Morining Break sponsored by MIERUNE Inc. & Cesium

This session will focus on the very early days of the open source geospatial community, including GRASS, UMN MapServer, the first FOSS4G conference in Bangkok, and the establishment of OSGeo.

The session will be featuring the three keynote speakers together with OSGeo President Jeroen Ticheler and several current OSGeo leaders and contributors.

The B2B session is for companies, organizations, and individuals who are using FOSS4G technology in their business, or who want to newly challenge themselves to adopt it.

The history of Open Source Software, including FOSS4G, has grown significantly not only through individual and academic contributions but also through adoption and support from many businesses. The spread of FOSS4G in business has played, and will continue to play, a major role in contributing to a sustainable community and society. We encourage you to take this opportunity to share your case studies of solving challenges with FOSS4G or new ideas, and to use it as a chance to find new business partners.

This B2B session will have lunchtime as a buffet, featuring presentations from sponsor companies. After the lunch buffet, there will be a café time where you can enjoy famous Hiroshima sweets selected by the LOC (Local Organizing Committee). Please use this time to share new ideas over coffee or Japanese tea.

Ticket:3,000 JPY
B2B tickets can be purchased at the time of purchasing your main conference ticket or at any time afterward. Please note that B2B tickets are available only to main conference participants.

https://2026.foss4g.org/en/program-schedule/b2b-session/

Lunch sponsored by Eukarya/Re:Earth & Geo Technologies

This presentation introduces a practical workflow for converting proprietary stormwater model data into open formats using CSV and SWMM inputs, enabling reuse and visualization in QGIS and Giswater environments.

For years, browser-based raster visualization has depended on backend services to preprocess, reproject, and tile imagery.

We built something different: a way to stream unmodified COG data directly from object storage, reprojecting the imagery in the browser — without a server in the middle.

Comparative analysis of route and health care facilities for the emergency patients of Pune and Bengaluru

Adwait Priyadarshan1 and Manish Kumar Mishra2

1Department of Computer Science & Engineering, IIIT-Bangalore-560100 - Adwait.Priyadarshan@iiitb.ac.in
2Environmental Monitoring and Assessment Division, BARC, Mumbai-400085 - manishkm@barc.gov.in

Keywords: Emergency care, Route, Accessibility, Bengaluru, Pune

Introduction
Despite the vast and complex terrain of India, attempts are now made to extend the digital resources developed for a particular region to other parts of the country. Open-source software is now effectively used to assist the decision support system for better connectivity, service delivery, emergency care and disaster management (https://fossunited.org/public-policy). Accordingly, a concept based on comparison of health care infrastructure and patient accessibility has been developed for two mega cities e.g. Pune and Bengaluru of India. Pune (in Maharashtra) and Bengaluru (in Karnataka) in India are two rapidly urbanizing South-Asian metropolitan cities which have witnessed unprecedented and multifaceted infrastructural growth. Recently, traffic congestion (time lost in intracity travel), air pollution (AQI), educational institutions, and availability of multi-specialty hospitals (health) are leading reasons for opting or relocating to a city of choice. The Ease of Living Index (EOLI) data provided by the Government of India, Bengaluru is ranked #1 (66.70) followed by Pune (66.27). Both cities are ranked #1 and #5 in TomTom’s world congestion index and #1 and #2 respectively in India (https://www.tomtom.com/traffic-index/ranking/). The congestion in the traffic can be attributed to several dynamic as well as static factors. The dynamic factors are office/school rush hours, sudden influx of vehicles from outside, weekends, festivals, vehicle breakdown, road construction or repair etc. The static factors (which change with time but do not alter the traffic flow dynamically) are number of nodes, population density, number of lanes, access (one-way or two way), road width, number of traffic signals, bus-lanes, rail-crossings etc.
Providing rapid response during emergencies is considered an essential part of governance. The decision support system (DSS) ought to decide the patient admission at the nearest available facility that can cater for the emergency. Despite this, several associated uncertainties can affect the outcome of the decision based on limited information about the proximity of the emergency service facilities. A comparative picture delineating the network behavior, spatial saturation and the derivations of equity implications from the modeling for Pune and Bengaluru cities of India is discussed in the paper. Availability of time-distance based service areas for the admission of emergency and critical care patients at the nearest hospitals equipped to attend a specific type of emergency is provided in the text. The traffic bottlenecks (both dynamic and static factors), congestion index, KDE, population and road density have been considered for the coverage calculation.
Methods
After repeated verifications from accredited agencies, a total of sixty-two (62) and fifty-five (55) hospitals capable of catering an emergency has been selected from Bengaluru metro and Pune-

Pimpri cities of India. These hospitals were designated as centroids for modeling. The road layer which was extracted from Open Street Map (OSM), using QuickOSM plugin consisted of primary, secondary, tertiary, residential etc., types. Many highway types e.g., paths, treks, proposed etc. were programmatically or manually removed. The layer was processed for nodes and networks to enable routing. Free and open-source software QGIS (ver. 3.44, Solothurn) has been used for analysing the data. The population coverage around the hospitals were estimated using the Voronoi polygons and the Global Human Settlement Layer (GHSL) data. The traffic data has been integrated into road networks of both the cities, with demarcations for 1, 2 and 5 km service areas and the time isochrones of 5, 10 and 15 minutes. In either case, they have been converted into convex hulls. The percentage of the covered service area has been calculated along with the overlay population raster and population percentage coverage. TomTom Stat one month traffic data available for August 2024 has been used as source data. For effective spatial data organisation, the data acquired in GeoJSON format has been converted to GeoPackage (GPKG). For generating the analytical dataset, an expansion of GPKG has been done. Within the traffic performance fields, this has enabled to derive a structured CSV, via unpacking the nested JSON strings. The complex time-set data has been transformed through this “flattening” process into individual columns that can be used for providing the Congestion Index- a ratio of baseline speed limit to the actual harmonic speed. Lastly, the geocoded CSV has been merged to the non-traffic road network. This was done through the nearest-neighbour spatial join breaking the traffic centroids to the nearest road geometries for creation of a combined, traffic-sensitive master network.

Discussion and conclusion
As far as the health care accessibility and emergency response is concerned the two cities provide a contrasting urban layout. Unlike the dense-sponge structure of Bengaluru, a tree-like pattern is found for Pune city. The worked-out road length of Bengaluru (16,319 km) is almost 1.5-times more than that of Pune (10,659 km). Also, an extensive network of high road density of Bengaluru (16.3 km/km2) provides the ambulance multiple route options thereby ensuring alternative route flexibility in case of block on roads. But slow movement is a likely phenomenon due to numerous intersections and street issues with regular turns and stops. Tree-like structural feature of Pune has a lower density of 12.1 km/km2, the city relies on the main arterial road that can facilitate fast movement during regular days. But the underlying sensitivity of blockage of any kind, cutting off from the access network can deeply affect the emergency access. Apart from the road layout, the cities are challenged due to the associated health care inequalities.

Observations based on the Spatial Coefficient of Variation (CV) use, a relatively higher score of 2.48 for Pune was obtained. A clustering of hospitals in the central region of Pune has been observed; leaving the peripheral areas scarce in emergency care hospitals. Although Bengaluru provides a robust infrastructure, the high inequality score of 2.28 corroborates the uneven distribution of hospitals with unplanned growth as added limitation.

References

1. https://fossunited.org/public-policy
2. https://www.tomtom.com/traffic-index/ranking/

The presentation describes processes and opensource tools employed by the author and his team to build and consume digital models for urban environments. Attendees will be presented with an overview of our work related to 3D data visualizations and a selection of use cases implemented for the MapStore WebGIS framework

What does it look like to teach programming and GIS entirely online? This talk shares stories from running the Kartoza internship remotely, guiding university students from basic computer science concepts to building simple web GIS applications with Python and Django.

Problem Statement
Open 3D city model datasets are becoming increasingly available worldwide through standards such as CityGML. In Japan, the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) has released Project PLATEAU, one of the largest national open 3D city model initiatives, covering over 200 municipalities with building, terrain, vegetation, and infrastructure data at multiple Levels of Detail. However, while PLATEAU provides official SDKs for proprietary game engines (Unity, Unreal Engine), no documented workflow exists for integrating PLATEAU data into open-source interactive environments. More broadly, the geospatial community lacks a fully documented, reproducible pipeline for transforming CityGML data into lightweight, interactive 3D applications suitable for deployment on low-specification hardware typically found in schools or community centres where participatory planning workshops take place.

This paper presents and evaluates a complete open-source pipeline from CityGML acquisition to interactive real-time 3D environment, using exclusively free and open-source software: Blender for geospatial data processing and 3D optimisation, and the Godot Engine for real-time rendering and interaction. We validate this pipeline through deployment in a participatory urban design workshop with elementary school students in Yokohama, Japan.

Pipeline Architecture
The pipeline comprises four stages.
Stage 1: Data Acquisition. CityGML data for the target site (Takashima Central Park, Minato Mirai, Yokohama; approximately 113 m x 159 m) was downloaded from PLATEAU's open data portal, including buildings, terrain, vegetation, and infrastructure at LOD1 and LOD2. The LOD selection involves a critical trade-off: LOD2 provides roof shapes and facade detail but substantially increases polygon count. A selective strategy was adopted, retaining LOD2 for buildings within the workshop perimeter and simplified representations for surrounding context.

Stage 2: Conversion and Optimisation in Blender. CityGML data was converted to Blender-compatible formats (FBX, glTF). This stage involved coordinate reference system alignment, geometric cleaning (degenerate faces, inverted normals, non-manifold geometries), mesh simplification balancing fidelity with rendering performance, semantic layer structuring (terrain, buildings, vegetation, public space), and removal of elements beyond the perimeter. The processed model was exported in glTF to preserve coordinate alignment with the Godot engine.

Stage 3: Integration in Godot. The optimised model was imported into the Godot Engine (version 4.x). A critical step at this stage was the automatic generation of collision shapes for all CityGML-derived meshes (buildings, terrain, infrastructure), a requirement absent from GIS workflows but essential for game engine navigation: without collision data, the first-person character controller would pass through walls and terrain. A custom plugin system was then developed, comprising: a drag-and-drop asset placement interface sourcing 94 curated 3D objects from the Kenney asset library (CC0 licence) across six functional categories; a first-person character controller for pedestrian-scale navigation over the collision-enabled environment; and an embedded GDScript-based logging system that records every placed asset's identifier, category, position (X, Z), and bounding-box volume as CSV files suitable for post-hoc geospatial analysis.

Stage 4: Deployment. The environment was packaged as a standalone Godot project deployable on standard educational PCs without dedicated graphics hardware, requiring stable frame rates and responsive interaction under classroom conditions.

Technical Challenges
The CityGML-to-game-engine conversion raises several challenges that deserve documentation for reproducibility.
CRS handling. PLATEAU metric coordinates produce models offset by large values from the scene origin when imported directly. A re-centring step in Blender was necessary to preserve internal metric distances while ensuring that coordinates extracted from Godot remain mappable to real-world positions.
Conversion artefacts. CityGML-to-FBX/glTF conversion introduced flipped normals, Z-fighting on coplanar surfaces, and disconnected mesh fragments, requiring manual inspection and correction. Documenting these artefacts and solutions reduces the barrier to entry for future pipeline users.
LOD and performance. Full LOD2 for all buildings exceeded classroom-grade PC rendering capacity. The selective LOD strategy maintained interactive frame rates while preserving spatial realism sufficient for participants to recognise the site. Quantitative performance characterisation (polygon counts, frame rates, memory usage) will be reported in the full paper.
No official Godot SDK for PLATEAU. Unlike Unity and Unreal, Godot lacks a PLATEAU SDK. Blender proved essential as simultaneous format converter, geometry optimiser, and semantic structurer, positioning it as the key bridge component in any open-source CityGML-to-game-engine pipeline.
Collision generation. CityGML meshes are designed for visualisation and spatial analysis, not for physics-based interaction. Game engine navigation requires collision shapes that prevent characters from passing through geometry. Automatic collision generation from imported meshes proved effective for terrain and building shells but required verification for complex or thin geometries where collision approximations could produce unintended barriers or gaps.

Validation
The pipeline was validated through a participatory workshop conducted on 17 December 2025 at an elementary school in Yokohama, with approximately 50 sixth-grade students (aged 11-12) working in 12 groups on six classroom PCs. Over two 45-minute sessions, participants placed 1,192 assets without system failures, demonstrating the pipeline's robustness under real deployment constraints with non-expert users. The embedded logging system successfully generated coordinate-stamped data for all groups, confirming that Godot's scripting capabilities can serve as a lightweight geospatial data collection layer. Detailed evaluation of pedagogical and spatial outcomes is reported in companion publications.

Contribution
This work makes three contributions to the open-source geospatial community. First, it provides the first documented end-to-end workflow for integrating CityGML/PLATEAU data into the Godot Engine via Blender, filling the gap left by the absence of an official SDK. Second, it demonstrates that a fully open-source stack (PLATEAU + Blender + Godot + Python) can deliver interactive 3D urban environments on low-specification hardware, removing both licensing and infrastructure barriers. Third, it shows that GDScript can serve as an embedded geospatial data collection tool, producing coordinate-stamped logs amenable to standard GIS analysis.
The pipeline is fully reproducible: all software is free and open-source, the PLATEAU data is publicly accessible, and the asset library is CC0-licensed. The methodology is transferable to any context where CityGML data is available, including European national mapping agencies and the emerging OGC CityGML 3.0 ecosystem.
Future work will focus on automating the Blender processing steps through scripted add-ons, benchmarking the pipeline across different PLATEAU municipalities and LOD configurations, and developing a standardised Godot plugin for direct CityGML import.

Nikkei Visual Data produces digital news content. This presentation introduces how FOSS4G technologies contribute to article production and related developments. It demonstrates the application of geospatial information in journalism through visualization using MapLibre and vector tiles, as well as data analysis and infrastructure development with tools such as QGIS.

Overview of OGC APIs implementation in GeoServer. OGC APIs offer a modern, modular, RESTful geospatial services using JSON and extensible building blocks.

PoliMappers reflect on a decade of student-led open mapping at Politecnico di Milano. We share lessons on community sustainability, FOSS4G academic integration, and volunteer data quality. Discover how this YouthMappers chapter bridges the gap between university education and impactful, collaborative open-source geospatial contributions from a student perspective.

Learn of all that is new in MapLibre - changes to the web map renderer, the native, the Martin tile server, and numerous other changes including the short intro into MLT - new tile format.

This is a personal account of a two-month sprint to attempt to rebuild our corporate GIS system in the Brazilian Federal Police using modern tools, solo. I will touch on tools, efficiency, burnout, and the golden pot at the end of the rainbow.

The geonode-k8s chart is production-grade, community-driven solution for deploying GeoNode 4.x and 5. This talk presents a status update on the project’s progress, highlighting e.g. the upgrade to GeoNode 5 and the improvements on start up time of GeoNode.

Computing a representative point that is guaranteed to lie inside a geometry is essential for labeling, geocoding, and spatial indexing.
This talk introduces "interior-point", an open-source project that ports the JTS (Java
Topology Suite) InteriorPoint algorithm to both TypeScript and Rust/WASM.

Did you know QGIS has a community hub for sharing styles, models, 3D models, projects, processing scripts, and more? This talk introduces hub.qgis.org and the QGIS Hub Plugin, which lets you discover and use these resources directly from QGIS, without ever leaving the application.

Culverted waterways and minor drainage features are physically present but persistently absent in OpenStreetMap — limiting flood and dengue risk analyses alike. This lightning talk documents a YouthMappers-led effort to map hidden urban waterways infrastructure and shows how improved completeness changes spatial risk outputs.

The geospatial ecosystem has evolved, introducing formats optimized for rendering, cloud-native access, and large-scale analytics. This talk provides a structured overview of formats including MVT, MLT, 3D Tiles, PMTiles, FlatGeobuf, and GeoParquet, explaining their core concepts, trade-offs, and how to combine them in modern geospatial workflows.

Japan has witnessed a surge in human-bear interactions, raising concerns about public safety and wildlife conservation. This phenomenon can potentially be traced back to dual pressures of climate change, urbanization significantly altering bear habitats and food sources.

PINOGIO connects gPocket for field data collection with a web platform for project management, map production, and story-map publishing. This talk shows how open-source technologies, offline workflows, and external GIS interoperability were combined into a practical GIS workflow that non-developers could use.

This presentation introduces a workflow for converting numerical weather prediction data from the Korea Meteorological Administration into real-time 3D visualizations. The process includes algorithmic transformation based on NWP models, optimization for lightweight performance, and display rendering, utilizing the open-source CesiumJS platform for 3D visualization.

Exploring Japan’s geospatial technology, history, use cases, and challenges through Japan Geospatial Times, an open-source blog documenting geospatial applications in Japan.

Background
Public health nursing uses routinely collected data to characterize population needs, plan targeted outreach, and coordinate prevention activities with clinical providers. Two operational questions are central to these tasks: where health risks are geographically concentrated and where residents actually access preventive services. Spatial cluster detection and provider catchment analysis address these questions from different perspectives, but they are often reported separately, limiting their interpretability for local planning.

Objective
To describe and apply a geospatial workflow that combines provider catchment analysis and spatial scan statistics using preventive checkup data, and to report catchment patterns and chome level clustering of elevated HbA1c within a shared geographic framework relevant to public health nursing practice.

Methods
We analyzed 3,200 records from Japan’s Specific Health Checkups program for adults aged 40–74 years in a single local government area in Japan. The study protocol was approved by the institutional ethics review board of the author’s university (approval number 2024 053). Analyses were conducted at two spatial scales. Service use patterns were summarized across seven residential elementary school districts. Spatial clustering was assessed at the chome level, using Japanese neighborhood scale address units and polygon boundary data.
The clustering outcome was elevated glycated hemoglobin, defined as HbA1c ≥ 5.6%, treated as an indicator of increased diabetes risk. Spatial processing, polygon handling, and centroid derivation were performed in QGIS. Statistical tabulation and testing were performed in R. Spatial scan statistics were computed using SaTScan.
Geographic boundary data for elementary school districts and chome units were obtained from e Stat, the Portal Site of Official Statistics of Japan. Population denominators for residents aged 40–74 years were also obtained from e Stat. No individual locations were mapped or reported. Spatial outputs were based on aggregated chome level tables linked to polygon centroids.
Provider catchment patterns were quantified using an origin destination contingency table crossing residential elementary school district, with seven origins, and provider location category, with nine destinations: seven in area districts, out of area providers, and group screening. The association between residence and provider location was evaluated using Pearson’s chi square test, and adjusted standardized residuals were examined to characterize origin destination pairs used more or less often than expected under independence. District specific proportions of within district use, out of area use, and group screening were summarized.
For spatial clustering, SaTScan input files were prepared by linking chome level case counts, population denominators for residents aged 40–74 years, and chome centroid coordinates. A Poisson spatial scan statistic was applied to identify circular windows with elevated risk relative to the population denominator. Cluster reporting included population, observed and expected case counts, observed to expected ratios, relative risk, and statistical significance based on Monte Carlo testing.

Results
Provider catchment analysis indicated strong geographic structuring of preventive checkup utilization. Residential district and provider location category were strongly associated (Pearson’s χ² = 2864.8, df = 48, p < 0.01). Adjusted standardized residuals were positive for all seven within district origin destination cells, indicating higher than expected utilization of providers located within the same district across all residential districts. However, the magnitude of within district utilization differed substantially by district. Within district use ranged from 6.8% to 72.5%, out of area use ranged from 11.2% to 37.3%, and group screening ranged from 3.8% to 7.1%, demonstrating heterogeneous catchment patterns within the study area.
Spatial scan analysis identified one statistically significant spatial concentration of elevated HbA1c. The most likely high risk cluster comprised 40 chome units and had a population of 9,577 adults aged 40–74 years. This cluster included 400 observed cases and 315.80 expected cases, corresponding to an observed to expected ratio of 1.27 and a relative risk of 1.37 (p < 0.01). Two additional areas had relative risks greater than 1 but were not statistically significant. The second cluster comprised 10 chome units, with a population of 4,986 adults, 209 observed cases, 164.41 expected cases, an observed to expected ratio of 1.27, and a relative risk of 1.32 (p > 0.05). The third cluster comprised three chome units, with a population of 177 adults, 13 observed cases, 5.84 expected cases, an observed to expected ratio of 2.23, and a relative risk of 2.24 (p > 0.05).

Conclusions
Provider catchment analysis and spatial scan statistics provide complementary geographic evidence relevant to public health nursing practice. Catchment heterogeneity indicates that patterns of preventive service utilization differ by residential district, including substantial variation in out of area utilization, while spatial scan statistics identify chome level areas where elevated HbA1c is concentrated beyond random variation. Presenting these results within a single geospatial workflow supports interpretation of local risk concentration together with district level service use patterns relevant to planning outreach, communication strategies, and provider collaboration.
A limitation is that the Poisson scan used census based population denominators for residents aged 40–74 years rather than chome specific counts of checkup participants. Spatial variation in checkup participation may therefore influence cluster detection and interpretation. The use of open source software provides a practical pathway to strengthen evidence informed public health nursing activities through transparent, reproducible, and adaptable geospatial analyses.

Will plugins really break when migrating them to QGIS 4?

Presentation introduces the SLA4GIS concept – its idea, scope, technical support principles, membership certification process and request handling mechanism. It highlights how the initiative redefines FOSS4G support, driving digital transformation across business and government while dispelling the mythical argument about the lack of official support for OpenSource GIS.

1 Introduction and Objectives
The large-scale development and maintenance of high-quality 3D city models remain a key challenge in the digital transformation of urban planning and management. In Japan, Project PLATEAU, launched by the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) in 2020, promotes the creation, utilization, and open dissemination of 3D city models as foundational data for smart city initiatives. Despite steady progress, the time and cost required to generate detailed models—particularly at Level of Detail (LOD) 2 or higher—continue to pose significant constraints, as such models are still largely produced through labor-intensive manual processes.
Project PLATEAU aims to complete 3D city models for 500 municipalities by fiscal year 2027, with roughly two-thirds achieved by fiscal year 2025. Achieving full coverage will require substantial reductions in production cost. In addition, 3D city models quickly become outdated due to ongoing urban changes such as construction and demolition, making periodic updates essential. However, high update costs risk limiting their timely maintenance, underscoring the need for scalable and automated solutions that can reduce costs while meeting strict quality requirements for municipal use.

In response, this study presents the development and validation of a beta version of an AI-driven automated modeling tool, named AI City Model Maker, designed to support the generation and updating of 3D city models at LOD2 and above. The primary objective of the tool is not to fully replace human operators, but to significantly reduce manual workload by automating substantial portions of the modeling process, thereby enabling cost-effective and repeatable production workflows.

2 Requirements
The tool is designed to comply with the “Standard Data Product Specification for 3D City Models” established by PLATEAU, which is based on the CityGML standard and aligned with ISO 19100 series quality requirements. This specification defines stringent criteria for positional accuracy, completeness, logical consistency, and thematic accuracy, reflecting the expectation that 3D city models will be used in official municipal operations such as urban planning, disaster management, and infrastructure assessment. For example, when models are generated from Level 2500 source data, horizontal positional accuracy must be within a standard deviation of 1.75 meters, and vertical accuracy within 0.66 meters, while logical and thematic errors are not permitted. Meeting these requirements through fully automated processes remains challenging with current technology.

3 Architecture and Applied Technologies
The proposed tool addresses this challenge through a hybrid workflow that combines AI-based automation with targeted human editing. It takes as input quality-assured datasets commonly available to Japanese local governments, including aerial imagery, digital surface models, building footprints, and digital elevation models. Using these inputs, the tool automatically generates building models at LOD1 and LOD2, road models at LOD1 and LOD2, and city furniture and vegetation models at LOD3. Outputs are provided in standard formats such as CityGML, enabling seamless integration into existing geospatial workflows and open data ecosystems.

From a technical perspective, the tool integrates multiple open-source and state-of-the-art AI technologies. Building model generation is based on an OSS “Auto-Create-bldg-lod2-tool”, enhanced with deep learning models such as convolutional and transformer-based architectures. Road extraction and modeling leverage semantic segmentation techniques, while city furniture and vegetation modeling combine point cloud analysis, object detection, clustering, and parametric modeling. The system is implemented in both cloud-based and desktop-based configurations, allowing flexibility in deployment depending on data volume, network conditions, and organizational constraints.

4 Implementation and Validation
To ensure practical relevance, the beta version of the tool has been deployed for pilot testing at six surveying and construction consulting companies in Japan. These companies are actively incorporating the tool into their existing 3D city model production pipelines, and regular feedback is being collected to guide further development. This real-world testing distinguishes the present work from many prior studies, which often remain at the proof-of-concept or research prototype stage.
Clear performance metrics and automation targets were defined during implementation. These include overall cost reduction across an entire city model, cost reduction at the individual feature level, and automation rates required to achieve these reductions. The long-term target is an overall cost reduction of 30–50%, assuming that some degree of manual editing remains necessary to meet quality standards. Feature-level targets vary by object type, reflecting differences in geometric complexity and data availability.

Evaluation results at the beta stage indicate mixed performance. For building models at LOD2, the automation rate for high-quality outputs remains well below the target, highlighting the difficulty of reliably reconstructing complex roof geometries from standard aerial imagery and elevation data. In contrast, road models and city furniture and vegetation models generally met or approached their respective targets, although performance varied depending on local conditions such as urban density and road structure. These results suggest that while AI-based automation is already effective for certain feature types, further improvements are required for complex building geometries.

5 Conclusion
Despite not yet achieving all targets, the beta version demonstrates the feasibility of a tool-oriented approach to AI-driven 3D city model generation. By focusing on usability, integration into existing workflows, and iterative improvement based on practitioner feedback, the proposed tool represents a step toward the social implementation of automated 3D city modeling technologies. In particular, it supports the expansion of open 3D city model coverage by lowering the barrier to production and updates, which aligns closely with the goals of both Project PLATEAU and the broader open geospatial community.

Future work will focus on improving modeling accuracy for existing feature types and extending automation to more detailed representations, including building and road models at LOD3. Through continued collaboration with industry users and further advances in AI and open-source geospatial technologies, the proposed approach aims to contribute to sustainable, scalable, and openly accessible 3D city model ecosystems.

MapStore is an open-source webgis built on React and Redux to create and share maps, dashboards, and geostories. It is mobile-ready and supports OpenLayers, Leaflet, and Cesium. This presentation covers current features, future roadmaps, and real-world case studies.

pycsw project status presentation. Come and find out the latest news on the project as well as future plans, and how to get involved!

Discover how the QGIS Brazil community uses translation and local networking to democratize FOSS4G. This talk highlights how breaking the language barrier is essential for empowering local governments, students, and professionals in South America, turning translation efforts into massive software adoption and community growth.

Tracing the 16-year evolution of the QGIS Brazil community—one of the world's first official user groups. From initial 2005 translations to a network of 40,000+ members, we explore strategies for virtual integration, institutional maturity, and the impact of hosting the 2nd QGIS LATAM Meeting in 2024

With the focus on 10-year cycle data collections by the Spanish National Forest Inventory (NFI), remote sensing-based biophysical indicators such as leaf area index are integrated in a Discrete Global Grid System (DGGS) as a complementary tool for forest monitoring in Catalonia.

Very High Resolution satellite imagery (eg WorldView3) provides insights into Earth surface processes, but suffers from limited spatial/spectral accuracy. While tools exist to correct these errors, there is currently no robust pipeline. Vhrharmonize is a Python library, command-line interface, and QGIS plugin that automates preprocessing and mosaic generation.

Digital Twins are software systems that provide dynamic virtual representations of physical systems(1), enabling modelling and visualisation, with automated data exchange and analytics being key attributes. These systems are enabling the development of smart cities(2) and may also represent the natural environment(3–5). Common use cases for Digital Twins are to monitor and control manufacturing lines or smart cities, but in environmental applications they are less common. Digital Twins can be used to automate and connect computer models of the environment, enabling on-demand simulations or ingestion of model outputs in planning.

A key example of an environmental Digital Twin is the the EU's “Destination Earth” system, which is being developed as a Digital Twin for climate services, to facilitate access to weather and climate models which can be used for impact studies(6). Physics-based Digital Twins such as this will revolutionise access to and use of numerical model predictions. By connecting systems together through open-data and standards, a “Digital Twin web” will be created, powered by rapidly growing data and distributed cloud computing(7). Yet the development of each component remains challenging.

In this work, we describe the development of the Environmental Digital Data Intelligence Engine (EDDIE), an open-source framework for creating environmental Digital Twins. The concept of EDDIE is that it acts as a core engine which manages the ingestion and processing of spatial and other data, provides a modularised framework for running environmental models from these data, orchestrates them and ingests their results, and provides an (optional) web-based user interface and visualisation system. EDDIE is based on APIs, meaning that is it possible to connect two or more instances of EDDIE (or other Digital Twins) to share data and environmental models. For example, these Digital Twins can represent multiple different domains, such as hazard assessment, environmental monitoring, and community and urban planning. Here we describe the EDDIE system and provide some application examples.

EDDIE and its open-source module implementations help developers of novel Digital Twins by providing a structure to follow, and providing library functionality for key spatial data handling processes. A dashboard of existing spatial data becomes trivial to setup and fetching and combining open data for analysis becomes simpler by following existing workflows and patterns.

An application using EDDIE is comprised of multiple containers working together to form a web application. Key containers include PostGIS, GeoServer, TerriaJS and the EDDIE backend and processing containers. EDDIE’s Python library is used in the backend to prepare data and keep them up to date if required. When a model scenario is requested, the Python library is used within domain-specific modules to gather and process data to generate predictive outputs. TerriaJS is the typical frontend for an EDDIE application, allowing 3D visualisations as well as the ability to request model scenarios to be run. These requests use the OGC Web Processing Service standard, and return JSON results that are valid TerriaJS catalog items. This allows requests to use existing tooling with standardised inputs, with results that can be used in further processing scripts or can be automatically displayed on the web. The standard front-end for EDDIE applications is TerriaJS, with the backend containers able to expose detailed dynamic catalogs. These catalogs can also be used by other independent Digital Twins, enabling them to use all functionality available to create more powerful ecosystems of Digital Twins.

EDDIE is used in active research projects for multiple distinct Digital Twins developed by the Geospatial Research Institute Toi Hangarau. EDDIE was born from the Flood Resilience Digital Twin (FReDT), focused on automated prediction of flood risk and collation of data for impact analysis. Currently, FReDT allows users to select parameters relating to climate change to assess how sea-level rise and increased storm intensity may change flood inundation risk. Ongoing developments are focused on working with communities to develop nature-based solutions to reduce flood impact, while allowing them to trial many different scenarios using the web interface. The core modules were extracted from FReDT to be able to be reused to construct novel environmental Digital Twins, and this core has formed EDDIE.

From there, EDDIE was used as the basis for the Ōtākaro Digital Twin, a prototype environmental platform for monitoring the health of the Ōtākaro/Avon River in Christchurch, New Zealand. This Digital Twin was created in collaboration with Ngāi Tūāhuriri and Christchurch City Council. Modelling available within the platform currently focuses on the MEDUSA 2.0 stormwater pollutant runoff model using user-inputted rainfall event parameters, and potential future modelling may include linking this to rainfall gauge telemetry.

Most recently, EDDIE was the core framework used to create Te Awarua Kai Ora, a platform for Te Awarua / Porirua Harbour. This platform collates data from open data sources relating to the harbour, presents spatial data on environmental sampling, and allows the Porirua community to understand a flow model of the harbour created at the Geospatial Research Institute Toi Hangarau in collaboration with PHF Science. People can create story maps to describe the environmental data, as well as interact with overviews and detailed plots of flows within the harbour to understand how the catchment, streams, tide and rain contribute to sedimentation, flushing, or contaminant buildup.

Current and near-future developments of EDDIE include optimisations and templates for cloud deployments. Focusing on facilitating cloud deployments allows for dynamic scaling to occur, allowing for large amounts of processing power to be accessed for only the short amount of time needed. This will be invaluable for FReDT allowing us to run many proposed scenarios at once for communities. EDDIE was built on a containerised architecture, and these additional developments will remove barriers to deploying new EDDIE projects.

EDDIE provides a framework for building environmental Digital Twins with interoperable standards. This framework will help adoption of new Digital Twins and strengthen the community ecosystem of environmental Digital Twins. This will enhance access to data and insights for communities, for planning, for decision making and for research.

1. Madni A, Madni C, Lucero S. Leveraging Digital Twin Technology in Model-Based Systems Engineering. Systems. 2019 Jan 30;7(1):7.
2. Deren L, Wenbo Y, Zhenfeng S. Smart city based on digital twins. Comput Urban Sci. 2021 Dec;1(1):4.
3. Blair GS. Digital twins of the natural environment. Patterns (N Y). 2021 Oct 8;2(10):100359.
4. Bauer P, Stevens B, Hazeleger W. A digital twin of Earth for the green transition. Nat Clim Chang. 2021 Feb;11(2):80–3.
5. European Commission. Destination Earth (DestinE) [Internet]. 2022 [cited 2022 Apr 18]. Available from: https://digital-strategy.ec.europa.eu/en/policies/destination-earth
6. Hoffmann J, Bauer P, Sandu I, Wedi N, Geenen T, Thiemert D. Destination Earth – A digital twin in support of climate services. Clim Serv. 2023 Apr;30:100394.
7. Autiosalo J, Siegel J, Tammi K. Twinbase: Open-Source Server Software for the Digital Twin Web. IEEE Access. 2021;9:140779–98.

Awesome Spectral Indices (ASI) is an open, community-driven catalogue designed to make spectral indices easy to discover, document, and compute. Five years after its first release, ASI supports 260+ indices across multiple programming languages and platforms. This talk reviews its evolution, current state, and future directions.

GeoServer 3 completes a long-planned modernization effort updating core dependencies while preserving backwards compatibility. This talk revisits the transition, outlines how it was organized and delivered, shares lessons learned and future directions.

In this lightning talk, I’ll show how 3D data helps urban planning. I’ll outline key datasets and how open-source tools like GeoServer publish them as map tiles and GeoJSON. I’ll also introduce three uses: Cross-sections Analysis, Visibility Analysis, and 2D–3D Linking.

A QGIS plug-in is developed for supporting designing culverts in civil engineering projects so that manual work currently adopted by the industry can be automated.

SedonaDB is an open-source single-node analytical database engine for GIS. But, with established options like PostGIS and DuckDB, why do we need yet another engine? This talk will provide an overview of the current FOSS4G database landscape and explain why SedonaDB is a name you need to know.

Beyond LiDAR and automated sensors, human photography captures the "why" of a place. This talk draws inspiration from Panoramio and Flickr, questioning whether human-captured imagery remains a vital layer for modern mapping and exploring what this looks like today.

This session provides an overview of the GeoPlegma project, how it evolved the past year and what is in store for the next.

Training is often treated as an afterthought in the geospatial industry, yet it is essential for real adoption of open tools. This talk share practical lessons from delivering and creating training for GIS, remote sensing, and earth observation, what worked and what didn’t, and what I now do differently

Urban microclimate modelling has become an important tool for analysing urban heat island effects, thermal comfort, and the environmental performance of urban design interventions. High-resolution microscale models allow researchers to investigate the interaction between urban morphology, surface materials, vegetation, and atmospheric processes at the scale of individual streets or neighbourhoods. Among these models, ENVI-met is widely used for simulating urban microclimate conditions due to its ability to resolve airflow, radiation exchange, heat transfer, and vegetation–atmosphere interactions within complex urban environments.
Despite their advanced capabilities, microscale models require detailed meteorological forcing data as input. These forcing datasets typically include time series of air temperature, wind speed and direction, relative humidity, and radiation parameters, which are usually obtained from nearby surface meteorological observations. In practice, however, such observational datasets are often unavailable, incomplete, or insufficiently representative of the simulated urban area. This limitation is particularly pronounced in smaller cities, complex terrain environments, or rapidly developing urban regions where meteorological monitoring networks are sparse. As a result, the lack of reliable forcing data often represents a key barrier to the application of microscale urban climate models for real case-study scenarios.
At the same time, mesoscale numerical weather prediction models provide continuous spatial and temporal coverage of atmospheric variables and represent a valuable alternative source of meteorological data. The Weather Research and Forecasting (WRF) model (Skamarock et al. 2019), especially when coupled with an Urban Canopy Model (UCM) (Tewari et al. 2007), enables the simulation of urban surface–atmosphere interactions at spatial resolutions ranging from several kilometres down to hundreds of metres. Mesoscale models incorporate land-use characteristics, surface energy balance processes, and regional atmospheric dynamics, allowing them to simulate realistic meteorological conditions across large spatial domains. While such models cannot explicitly resolve street-level urban geometry, they capture the broader atmospheric conditions and synoptic influences that drive local microclimate variability.
This study presents a methodology for generating meteorological forcing data for microscale urban climate simulations using outputs from an open-source mesoscale numerical weather prediction model. The proposed approach integrates ERA5 reanalysis datasets and WRF coupled with an Urban Canopy Model (WRF-UCM) to provide boundary forcing for simulations performed in the ENVI-met microscale modelling environment. The objective of the study is to demonstrate that mesoscale model outputs can effectively substitute for local meteorological observations in situations where surface measurement data are unavailable or insufficient.
The methodology follows a multi-scale modelling framework linking mesoscale and microscale atmospheric simulations. In the first stage, WRF-UCM simulations are performed for the broader study region using a system of nested domains to achieve progressively higher spatial resolution over the target urban area (Košice city, Slovakia). The mesoscale model configuration includes appropriate land-use classifications, urban parameterizations, and atmospheric boundary conditions to simulate key meteorological variables such as air temperature, wind speed and direction, humidity, and radiation fluxes.
In the second stage, selected meteorological variables from the WRF-UCM output are processed and transformed into forcing datasets compatible with ENVI-met boundary conditions. These datasets provide time-series information for near-surface atmospheric parameters required to initialize and drive microscale simulations. The generated forcing data are subsequently used as input for ENVI-met simulations representing detailed urban morphologies at the neighbourhood scale. Within this environment, the microscale model resolves local airflow patterns, radiation exchange between urban surfaces, vegetation interactions, and turbulent heat transfer processes.
A key advantage of the proposed approach lies in its ability to provide physically consistent meteorological forcing derived from a dynamically simulated atmospheric environment rather than relying solely on point measurements from meteorological stations. Mesoscale model outputs capture the influence of large-scale atmospheric circulation patterns, regional weather systems, and surrounding terrain features, all of which significantly influence urban microclimate conditions. Consequently, the generated forcing datasets reflect broader environmental influences that may not be represented in boundary conditions by isolated observational measurements.
The methodology has been tested for several case studies focusing on urban environments where meteorological observations are limited or unavailable (Fedor and Hofierka 2022). Simulation results demonstrate that forcing data derived from WRF-UCM outputs can reproduce realistic temporal variability in key meteorological parameters required for microscale modelling. Comparisons with available observational datasets indicate that ENVI-met simulations driven by WRF-UCM forcing achieve satisfactory agreement with measured temperature and wind conditions, supporting the reliability of the proposed method.
These results suggest that mesoscale model outputs can serve as a viable alternative source of meteorological forcing for microscale urban climate simulations. The approach enables the application of high-resolution urban climate modelling tools in data-scarce environments while maintaining physically consistent boundary conditions. Additionally, the integration of mesoscale forcing improves the representation of regional atmospheric influences, including the effects of surrounding topography and synoptic weather patterns.
The presented framework contributes to the development of multi-scale urban climate modelling strategies that bridge the gap between regional atmospheric dynamics and local urban microclimates. By combining mesoscale numerical weather prediction models with detailed microscale simulations, the approach supports more comprehensive assessments of urban climate processes across spatial scales.
This research builds upon previous studies by the authors investigating numerical modelling approaches for urban climate analysis and the integration of mesoscale and microscale modelling techniques. The proposed methodology demonstrates the potential for expanding the applicability of microscale urban climate models to regions where conventional observational data are limited, while maintaining reliable simulation accuracy. In this context, the mesoscale model WRF-UCM, which is an open-source modelling system, provides a transparent and flexible framework for simulating urban atmospheric processes and enables reproducible coupling with microscale models.
References
Fedor, T.; Hofierka, J. 2022: Comparison of urban heat island diurnal cycles under various atmospheric conditions using WRF-UCM. Atmosphere, 13(12), 2057. DOI: https://doi.org/10.3390/atmos13122057
Fedor, T.; Hofierka, J. 2025: Estimating heat stress in the urban centre of Košice using ENVI-met and high-resolution surface data. Geographia Cassoviensis 19(2), pp. 81-94. DOI: https://doi.org/10.3390/atmos13122057
Huttner, S.; Bruse, M.; Dostal, P. 2008: Using ENVI-met to simulate the impact of global warming on the microclimate in central European cities. 5th Japanese-German Meeting on Urban Climatology, October 2008, pp. 307-312. Available online: https://envi-met.net/documents/papers/Huttner_etal_2008.pdf (Accessed 4. March 2026).
Skamarock, W.C.; Klemp, J.B.; Dudhia, J. et al. 2019: A Description of the Advanced Research WRF Model Version 4. Technical Report NCAR/TN-556+STR. Available online: https://www.ecampmany.com/docs/cheatsheets/WRF.pdf (Accessed 4. March 2026).
Tewari, M.; Chen, F.; Kusaka, H.; Miao, S. 2007: Coupled WRF/Unified Noah/Urban-Canopy Modeling System. Available online: https://ral.ucar.edu/sites/default/files/public/product-tool/WRF-LSM-Urban.pdf (Accessed 4. March 2026).

pgRouting project status presentation.

Come and find out the latest news on the project as well as future plans, and how to get involved!

Developing QGIS plugins and using VS Code as your IDE? Get comfortable with DevTools for QGIS.

This lightning talk traces the inspiring journey of OSGeo Brazil, from informal online forums to hosting the monumental FOSS4G in the Amazon. Join us to explore how a passionate community united to democratize open-source geotechnology, build a local chapter, and map a collaborative future.

Valhalla is my favorite routing engine. It's one of the most flexible options already. but we can make it even better! Help us build extensions that enable even more use cases and easier collaboration.

This session introduces an open-source QGIS plugin that makes it easy to use basemaps, geocoding, and routing in everyday workflows without leaving QGIS.

Proximity-based community planning has emerged as an important approach for improving urban livability by ensuring that essential services are accessible within short travel distances. Rather than relying solely on administrative boundaries, this planning approach focuses on the spatial organization of everyday urban activities and the accessibility of key services such as healthcare, education, retail, and public facilities.

However, operationalizing proximity-based planning requires analytical frameworks capable of integrating large-scale mobility data, accessibility modelling, and spatial optimization. In many cities, these analytical components remain fragmented across different tools and datasets, making it difficult to develop reproducible and scalable workflows for urban analysis. At the same time, recent advances in open geospatial technologies provide new opportunities to build transparent, interoperable, and reproducible analytical systems that support data-driven planning practices.

This research proposes an open geospatial analytical framework for proximity-based community planning that integrates mobility data, open-source geospatial software, and spatial optimization techniques into a unified and reproducible analytical workflow. The framework is designed to support two key planning tasks:

the identification of functional urban communities based on observed mobility patterns, and the evaluation of spatial strategies for improving equitable access to essential urban services.

By combining multiple open geospatial technologies, the proposed framework aims to provide a flexible analytical approach that can be applied across different urban contexts while maintaining transparency and reproducibility.

The analytical framework consists of three core components that correspond to different stages of proximity-based planning analysis.

The first component focuses on the delineation of functional urban communities using mobility-based community detection. Large-scale telecom mobility data are used to construct origin–destination interaction networks between spatial units, represented as a grid-based spatial system. These mobility networks capture aggregated daily travel patterns between locations and provide a behavioral representation of urban spatial structure. Community detection algorithms from network science are applied to these networks in order to identify clusters of spatial interaction that represent functional communities emerging from mobility patterns. Unlike traditional planning units that rely on administrative boundaries, these mobility-derived communities reflect actual patterns of urban activity and interaction. As a result, the framework allows planners to identify spatial units that function as integrated communities in terms of daily mobility and service access.

The second component performs multimodal accessibility analysis at a high spatial resolution using open geospatial routing tools. Accessibility is calculated on a 250 m grid by combining telecom-derived origin–destination mobility flows with network-based travel times. Road-based accessibility is estimated using open routing engines such as OSRM, which compute travel distances and travel times along road networks derived from open geospatial data sources. In addition, public transport accessibility is computed using the R5 routing engine through the r5py Python interface. The R5 engine enables multimodal routing that integrates pedestrian networks, road networks, and public transport schedules derived from GTFS data. This approach allows the framework to calculate multimodal travel times across different transport modes, including walking, road-based transport, and public transit. The use of R5 and r5py also supports efficient computation of accessibility metrics for large-scale urban datasets, enabling the evaluation of accessibility patterns at a fine spatial resolution. By integrating multiple travel modes, the framework provides a more realistic representation of accessibility conditions in dense urban environments where public transport plays a significant role in daily mobility.

The third component incorporates facility allocation models based on linear programming in order to optimize the spatial distribution of urban services. The optimization model uses accessibility indicators derived from the multimodal analysis to evaluate potential service locations and allocation strategies. The objective function simultaneously considers equity and economic efficiency in service provision, allowing the model to identify spatial configurations that improve service accessibility while minimizing spatial inequality and redundant infrastructure investment. The allocation model can therefore support planning decisions related to the placement of public facilities, community services, and other urban amenities that are critical for proximity-based community planning.

A key contribution of this study lies in the integration of multiple open geospatial tools into a unified analytical workflow. Rather than introducing a single standalone software package, the framework combines several existing open-source geospatial technologies, including Python-based network analysis libraries, open routing engines, and geospatial data processing tools. This integration demonstrates how different open geospatial components can be combined to support complex urban analytics tasks. Furthermore, the computational workflow is designed to support reproducible research practices by documenting analytical steps and enabling the release of scripts, models, and computational procedures under open-source principles. Such reproducibility is essential for ensuring transparency and enabling other researchers and practitioners to replicate and extend the analytical framework.

The framework is demonstrated through an empirical case study in Busan, South Korea, where telecom mobility data are used to analyze community structures and evaluate alternative service distribution scenarios. The case study illustrates how mobility-derived communities differ significantly from conventional administrative planning units and provide a more realistic representation of urban spatial interaction. The integration of multimodal accessibility analysis and spatial optimization further allows planners to examine how alternative facility configurations influence accessibility outcomes across the urban population. These results highlight the potential of combining open geospatial technologies with mobility data to support evidence-based urban planning.

This study contributes to the open geospatial research community in several ways.

First, it demonstrates how open geospatial technologies can support integrated urban analytics workflows for community-level planning.

Second, it connects methods from network science, multimodal accessibility modelling, and spatial optimization within a reproducible open-source analytical framework.

Third, it highlights how open geospatial ecosystems enable transparent and collaborative approaches to urban analysis and planning.

By presenting a reproducible analytical framework for proximity-based community planning built on open geospatial technologies, this research contributes to ongoing efforts within the FOSS4G community to advance open, scalable, and collaborative geospatial solutions for sustainable urban development.

high-performance, serverless spatial analytics framework using GeoArrow, DuckDB-Wasm, and Parquet. By shifting from row-based GeoJSON to a columnar architecture, the method achieves a 10x–50x performance boost, enabling the seamless browser-based visualization and processing of over ten million records without backend infrastructure.

Open data projects like OpenStreetMap and OpenAddresses have become some of the best data sources in the world. But translating from data to real-time guidance remains challenge. Learn how we're tackling this with Ferrostar, a cross-platform FOSS navigation SDK.

MapLibre GL’s efficient "inline-shading" is ideal for standard maps. However, our Navara system’s dynamic day-night cycle, featuring constant light motion, exceeded standard coupled architectures. This necessitated a more flexible pipeline, leading us to implement a decoupled terrain normal generation approach for real-time, high-performance rendering.

Conventional hazard maps, public GIS layers, and disaster education materials remain essential resources for community disaster risk reduction. However, they do not always support in-situ spatial understanding during field-based learning and training. In many disaster walking tours, participants can read hazard maps and view web GIS layers, yet still struggle to connect those representations with the terrain, streetscape, and built environment around them. This gap is especially evident for children, first-time visitors, and residents who are not accustomed to translating two-dimensional hazard information into situated judgment. To address this problem, we developed MUSUBOU-AR, an open-source geospatial augmented reality framework for disaster walking tours, community disaster risk communication, and place-based disaster learning. Rather than treating AR only as an immersive visualization layer, the framework is designed as a reusable geospatial system that connects public GIS resources, local scenario authoring, on-site AR interpretation, and post-activity review.
MUSUBOU-AR is a mobile application that switches between a conventional map mode and an AR mode. In AR mode, virtual hazards are displayed relative to the user’s current position so that disaster scenarios can be experienced where risks may emerge. The system supports multiple disaster types, including flood inundation, fire, landslide, building collapse, block-wall collapse, and liquefaction. Hazards can also be configured as time-varying events, enabling organizers to express dynamic scenarios such as expanding floodwater or spreading urban fire during field-based training. This design supports a transition from passive map reading to situated risk interpretation.
Existing studies in disaster education and geospatial learning have explored practical drills, ICT-based learning materials, game-based activities, and AR-supported virtual disaster experiences. In contrast, our contribution is not centered on AR experience alone. We position MUSUBOU-AR as an open geospatial framework that integrates publicly available GIS layers, reusable route and scenario authoring, field logging, and optional LiDAR-enhanced visualization within a single deployable workflow.
A core contribution of the framework is its interoperability with open geospatial data. In map mode, MUSUBOU-AR supports the standard XYZ tile scheme used in Japanese public map distribution, making it possible to overlay publicly available GIS layers, including Geospatial Information Authority of Japan tiles and hazard-related layers distributed through the national hazard map portal. Project-specific or locally prepared datasets can also be incorporated through standard GIS workflows by generating custom XYZ tiles with QGIS and plugins such as QTiles and deploying them on a web server for use in the application. The framework also integrates Apple Watch-based trajectory and activity logging, synchronizes positional and biometric records into a unified GPX-based dataset, and supports open-source web review through GPXreaderWeb. In addition, LiDAR-enhanced rendering on supported iPhone Pro and iPad Pro devices improves spatial recognition and allows floodwater to be rendered at a specified height above the recognized ground surface while preserving backward compatibility for non-LiDAR devices.
Another key contribution is the authoring workflow. Although public GIS and open data make disaster-related information widely available, preparing field-ready AR content remains a practical bottleneck. To reduce that barrier, we developed a web-based authoring environment for MUSUBOU-AR datasets by customizing the open-source Re:Earth platform. This tool allows users to create, edit, and export the core data required by the application, including points, routes, and GIS layers. The contribution is therefore not limited to the application itself; it also includes a reusable workflow for transforming public geospatial resources and local field knowledge into deployable AR scenarios.
We position MUSUBOU-AR not merely as an educational app but as an open geospatial workflow for field deployment. The framework has already been deployed in multiple community- and school-based activities in Japan, including disaster walking tours, local awareness events, and place-based disaster learning programs. In this paper, we draw on these deployments and present a joint disaster walking tour conducted in the Hiro area of Kure City, Hiroshima Prefecture, as a representative validation case. In that case, route design and hazard placement were based on publicly available flood and landslide hazard maps, local field inspection, and additional local disaster-related map information.
This case is used as field validation, not as the primary contribution of the paper. Before the activity, university students studied topography, hazard maps, and three-dimensional terrain representations of the area. During the field session, they walked with elementary school pupils and used MUSUBOU-AR to explain local hazards in situ. Pre- and post-activity questionnaires administered to the university students were analyzed using multilevel models. The results showed a significant overall shift toward higher response levels after the activity, with an odds ratio of 21.7 for moving to a higher response category in the cumulative logit mixed model. Particularly relevant from a geospatial perspective, significant gains were observed in noticing environmental danger signs, understanding hazard maps and disaster-related signs, identifying dangerous and relatively safe places on maps, and planning evacuation routes while using maps. Open-ended responses further indicated that participants frequently identified roadside ditches or waterways and narrow roads as hazardous places, mainly in relation to flood or inundation risk.
This study has limitations. The field validation involved a relatively small sample and did not include a control group, so causal claims about educational effectiveness should be made cautiously. Nevertheless, the main contribution of this work lies in presenting an openly reusable geospatial AR framework that combines interoperability with public GIS and open data, a web-based authoring workflow for routes, points, and tiled layers, mobile AR hazard visualization with optional LiDAR enhancement, and GPX-based field logging with open-source web review. The source code of MUSUBOU-AR is publicly available in a GitHub repository, supporting reuse and reproducibility. We therefore position MUSUBOU-AR as a practical contribution to the FOSS4G community, demonstrating how open geospatial data, open-source software, and field-deployable AR can be combined for disaster risk communication.

Open Software and Open Standards are complementary pieces of the geospatial ecosystem. In 2022, OSGeo and OGC signed a new Memorandum of Understanding (MoU) that aims to benefit the mission and goals of both organizations. This presentation will provide an update on collaboration, reference implementations, code sprints, and future plans.

QGIS Web Client (QWC) is a modular next generation responsive web client for QGIS Server, built with ReactJS and OpenLayers. This presentation gives a short introduction to the QWC project and presents new developments.

The MapLibre Tile Format is a new open, community-governed successor designed to overcome compression, interoperability, and extensibility limits of old.

This talk explains its design principles, and ongoing development, including compression, and tooling, while outlining the roadmap and opportunities for solving the hard problems we all face.

DGGS gaining popularity in the GeoAI space, we'll look at pros and cons of storing your vector and raster data in couple of indexing algorithms.

To provide geospatial mapping training and raise awareness on the use of geospatial tools to empower Pacific Women — including young women, women with disabilities, and women from outer islands in mapping to access, utilize, and apply mapping resources for community development and decision-making.

Building on previous research, we introduce a traffic simulation platform enhanced for real-world urban planning. By optimizing large-scale data processing and interactions, we demonstrate technical maturity as a practical decision-support tool, moving beyond visualization toward immediate, high-performance deployment in professional and production-ready environments.

This presentation introduces the R package seg developed nearly fifteen years ago. I would like to share the difficulties encountered while developing and maintaining a small academic package, especially as job changes and passions shift.

GeoReports is a java servlet web application that provides a powerful tool for creating a geographically rich PDF Report containing a series of predefined pages relating to a location of interest.

Testing a CityGML-tileset conversion pipeline using topologies does not work: tile writers can add or remove vertices and even change geometry types. We show how statistical + raster-based testing in Re:Earth flow catches bugs in its engine development.

This talk celebrates how women volunteers like us strengthen and diversify open source geospatial ecosystems, especially in underrepresented regions from novice starter to member and Ambassador of various open source communities.

Coffee Break sponsored by MIERUNE Inc. & Cesium

Historic agricultural terraces are often poorly documented, especially where woodland obscures morphology. This paper presents an open source GeoAI workflow for semi automatic terrace mapping from LiDAR DEMs. By combining terrain, solar irradiance, soil erodibility and accessibility predictors, the extended Random Forest model improved detection, supporting heritage documentation and assessment.

Abstract :

Building footprint extraction from high-resolution satellite imagery requires accurate building boundary raster masks and an effective shape reconstruction method to produce natural building footprints. Unlike vertex-centric graph approaches or mask contour tracing, we propose DINO-EdgeQuery, an edge-first paradigm that addresses this need. A frozen DINOv3 backbone, adapter neck, and an instance decoder inspired by Mask2Former provide building-aware boxes and masks, while a region of interest (ROI)-conditioned EdgeQuery decoder predicts edge activity, center, direction, length, and successor relations. By capturing dominant wall structures as edge primitives and reconstructing vertices purely through deterministic line intersections, we deliberately separate neural edge prediction from verifiable geometric assembly. Active edges are ordered by successor scores, and the P2 quality gate removes invalid or contained duplicate polygons. This geometric post-processing encourages sharp, line-intersection-based corners and enforces topology-safe polygons through deterministic validity checks. To improve generalization, we train segmentation and EdgeQuery branches separately, mask inactive losses to zero, and merge the trained weights for unified inference. Validation on the dataset publicly released on the website by the Wuhan University Geospatial Computer Vision Group demonstrates that our EdgeQuery framework, including the P2 quality gate, delivered high-quality, geometrically superior polygon outputs with zero self-intersections and triangle collapses, and simultaneously improved Intersection over Union (IoU) and suppressed duplicate building footprints.

This presentation introduces a spatial system integrating an Environmental Impact Assessment (EIA) scoping tool with online authoring and review functions.
Built with OpenLayers, the system connects spatial data, experts, and stakeholders to support collaborative environmental decision-making.

We're creating a new ecosystem for client-side raster data visualization in Python & the browser, enabling interactive WebGL rendering of COG and Zarr data.

This talk presents a high-level overview of how this works and how to leverage it in your projects.

A practical tour of GeoServer 3 covering upgrade steps, new requirements, refreshed UI, and updated module structure. Learn what’s changed, what remains familiar, and how to transition from GeoServer 2.x efficiently, with a focus on real-world adoption and getting up to speed quickly.

Although an alpha version has been available on the Leaflet website for over a year, it remained unclear for a long time what we could actually expect.

Now there is more information -let's take a close look at what is changing with Leaflet 2.0.

Discover all QField new features and Explore how QField empowers people to map and understand the world—supporting daily tasks, global challenges, and the UN SDGs through open-source, intuitive, and collaborative mobile geospatial tools.

As one of the sponsors, and as CTO of MIERUNE, I'll introduce our company by focusing on the technologies we work with.

Founded in 2016, MIERUNE has grown into a technology company specialized in geospatial technology and modern web architecture. In this session, I'll demonstrate how these two domains come together through our actual products and solutions.

A non-technical talk in regards to the structure of OSGeo, it's members, local chapters, and financing.
The work of OSGeo:UK, the prospect of OSGeo Europe, and our plans for FOSS4G 2027 in Bristol.
The impact of Gen AI for OSGeo and Open Source licences.

We present Z7 Explorer, a web application that computes IGEO7/Z7 grid indexes entirely client-side using a pure JavaScript port of DGGAL's ISEA projection engine. We demonstrate why IGEO7 — an equal-area, pole-seamless hexagonal DGGS — addresses fundamental limitations of H3 and conventional coordinate systems.

We introduce landlensdb, an open-source Python package for managing proximity sensing imagery, including action cameras, 360° cameras, and UAVs, using PostgreSQL/PostGIS. It automates metadata extraction, corrects geolocation errors via road network snapping, and enables scalable spatial-temporal queries and visualization for large-scale geotagged image datasets.

Jointly optimizing map styles and underlying data can significantly improve vector map performance. This talk shows how data- and style-driven techniques reduce tile size, speed up loading, and improve client rendering—without compromising visual quality - based on results from real-world datasets and styles.

In the winter of 2024, a corporate tech-for-good program was shut down, and with it nearly all its OS projects went dark. In this lightning talk, we’ll cover key lessons of what went right—and what didn’t—to help other OS efforts avoid the same fate.

MAPME is a community-driven initiative promoting access to open GIS resources for planning, monitoring, and evaluation in international development. This lightning talk introduces its mission, key open-source tools, and use cases, showing how geospatial data drives development impact and how participants can join and contribute.

GIS is amazing, but there are many people who would like to create a map, who don't know GIS. gitRmap is an easy way of creating a simple map. It is designed for people who know a little bit of Git, but don't know any R or GIS.

PLATEAU, point clouds, and vector datasets are rapidly expanding the scope of geospatial applications. This session introduces the latest developments in 3D Tiles 2.0 with Cesium and explores how GIS data, 3D city models, and reality capture data can be integrated into next-generation digital twin workflows.

The GDAL Sponsorship Program changed how the project operates, has resulted in project-wide functionality and improvements, and put the project on a previously unachievable sustainability path.

City2Graph is an open-source Python library bridging GIS, network science, and geospatial artificial intelligence (GeoAI) with Graph Neural Networks (GNNs). It offers a unified pipeline to construct, analyse, and visualise graphs from diverse data sources, with conversion between GeoPandas, NetworkX, and PyTorch Geometric.

Introducing a deep learning-based rooftop solar model using open remote sensing data and OpenStreetMap. We will explain the design of the model and demonstrate its use in the Climate Action Navigator.

DigiAgriApp is a free open-source suite to monitor agricultural fields. It is a comprehensive client-server platform designed to manage agricultural data with high granularity, from fields down to individual plants. It is built on standards to ensure flexibility and integration. Let’s see what’s new in 2026.

Apache Airflow manages tasks in a data pipeline from data ingestion and preprocessing to storage in ZARR format for multidimensional satellite imagery. ZARR supports efficient management of large-scale datasets and parallel processing, while Airflow automates and monitors workflow tasks.

Introduction
Grasslands historically occupied a substantially larger portion of the Japanese landscape than they do today. In western Japan, traditional tatara iron production—a pre-industrial smelting technology that used iron sand and charcoal in clay furnaces—has frequently been associated with the persistence of grassland landscapes. Historical narratives often portray these open landscapes as products of the massive forest clearance and charcoal extraction that iron production required. Early ecological studies reinforced this interpretation: Ito (1962) noted a correspondence between the spatial distribution of tatara production areas and grassland occurrence in the Chūgoku region, a pattern that has since been widely cited as evidence of a direct causal relationship. Despite the persistence of this interpretation, it has rarely been subjected to rigorous quantitative examination using spatially explicit data. This study examines the spatial relationship between tatara iron production sites and grassland distribution in Tottori Prefecture, western Japan, using an integrated open geospatial analysis workflow built entirely from open-source tools and publicly available datasets.

Study area and Data
The study area, Tottori Prefecture, occupies the northern slope of the Chūgoku Mountains and was historically one of the active tatara iron production regions in Japan. The prefecture's geology is dominated by granitic formations in the west and center, with distinct volcanic deposits around the Daisen volcano in the northwest. Three major river basins—the Sendai, Tenjin and Hino river systems—organize the mountainous landscape into hydrologically distinct units.
The analysis integrates three classes of data. Grassland area was derived from municipality-level statistics in the 1950 World Agricultural Census. Because the original records exist only as printed statistical tables, a structured digitization protocol was developed using AI-assisted table extraction, converting scanned census pages to machine-readable CSV format. Tatara iron production site locations were compiled from the Tottori Prefecture WebGIS cultural heritage database. Raw spatial data were extracted via browser developer tools and processed into structured point data using Python, then aggregated to municipality counts using spatial joins in QGIS 3.40. Environmental variables were constructed from open geospatial datasets, including the 1:200,000 seamless geological map of Japan, the national hydrological grid dataset, and DEM data from which terrain indicators including elevation, slope, and curvature were derived.

Analysis Methods
Spatial data processing and analysis were conducted using Python-based open-source libraries. GeoPandas was used for vector data integration and spatial joins. Terrain derivatives were computed from DEM data using rasterio and scipy, with zonal statistics aggregated at the municipality level via rasterstats. All vector data were managed in GeoPackage and FlatGeobuf formats. Statistical modelling was implemented using statsmodels and scikit-learn.
The analysis proceeded in three stages. First, ordinary least squares regression models were estimated to assess the relationship between environmental variables and municipality-level grassland area, evaluating geological entropy, dominant basin membership, mean elevation, slope, and curvature as predictors. Second, stepwise model comparison assessed the marginal contribution of tatara site density after environmental predictors were included. Third, DBSCAN spatial clustering was applied to tatara site point data to identify geographically concentrated production districts and characterise their environmental context. The complete analysis code is archived on Zenodo (https://doi.org/10.5281/zenodo.19042339).

Results
Tatara iron production sites exhibit strong spatial clustering, identifying two large production districts in western Tottori Prefecture concentrated within the Hino River basin. These districts correspond closely with granitic geological formations and mid-to-high elevation zones, consistent with the known requirements of iron-sand-based smelting for specific geological and hydrological conditions.
Grassland distribution is most strongly predicted by environmental variables rather than by tatara site density. Mean elevation is the most consistent predictor across model specifications. When tatara site counts are added to environmental models, the improvement in explanatory power is modest. The interaction term between tatara count and slope gradient is statistically significant and negative: municipalities with many tatara sites but steeper terrain show a weaker association with grassland area. This suggests that iron production promoted grassland formation primarily in accessible mid-elevation areas, while in steeper terrain the same activities relied more directly on adjacent forest without generating the open landscapes associated with grassland land use.

Discussion
These findings suggest that the long-standing narrative associating tatara iron production with grassland landscapes requires qualification. Both phenomena appear to have developed within similar environmental opportunity spaces defined by geology, basin structure, and topography. From the perspective of cultural evolution and niche construction theory, tatara iron production can be interpreted as a technological strategy adapted to a particular ecological niche, in which the apparent correlation between industrial sites and grassland landscapes arises partly from shared environmental constraints rather than from direct landscape transformation alone.
From a methodological standpoint, the study demonstrates that open geospatial workflows combining QGIS, GeoPandas, rasterio, rasterstats, and statsmodels can support rigorous historical landscape analysis. The integration of AI-assisted digitization, spatial data standardization, and reproducible statistical modelling within a single transparent workflow illustrates the practical potential of the open geospatial ecosystem for environmental history research. The approach is transferable to other historical industrial landscapes where documentary and archaeological data can be combined with environmental geospatial datasets to re-examine inherited interpretations of landscape change.

Rendering dynamic map typography is notoriously difficult. Standard engines rely on pre-baked PBF glyphs, breaking complex text shaping (like Arabic), and making custom fonts hard to use. This talk explores these pitfalls and reveals how Navara abandons PBFs entirely.

This talk introduces two QGIS plugins designed to boost OGC API adoption, which may be hindered by a lack of tools. The "pygeoapi configurator" simplifies publishing data, while the "QGIS OACS" plugin helps users discover and visualize datasets.

Reproducibility in geospatial science is hindered by opaque data, proprietary software, and complex machine learning workflows. This talk highlights challenges in deep learning reproducibility and presents practical strategies, tools, and documentation practices to create transparent, repeatable experiments using open‑source technologies.

Motivation and Problem Context

Advances in generative artificial intelligence are enabling new applications across many industry sectors. Large language models (LLMs) have the potential to bring geospatial decision support to non-technical professionals by allowing users to explore spatial questions through natural language. Compared with traditional geospatial workflows—characterized by manual layer selection and preprocessing, static maps, delayed updates, limited model transparency, fragmented analytical processes, and results that are difficult to reproduce—GeoAI-enabled approaches offer the promise of on-demand, ad hoc spatial analysis. By reducing the burden of integrating knowledge across multiple data silos, such approaches can broaden access to geospatial insight while maintaining essential principles of integrity, provenance, and trust (IPT).

However, LLMs lack intrinsic geospatial awareness and do not maintain structured knowledge of geographic features or their domain-specific relationships. As a result, they cannot natively reason about how disruptions propagate through interconnected spatial systems—such as transportation networks, health service accessibility, or school catchments—during events like floods, wildfires, or other natural hazards. While LLMs excel at synthesizing textual information, they lack an internal representation of the spatial networks and semantic relationships required for geospatial reasoning.

Overcoming this limitation requires a sustainable and open approach for making the wealth of geoinformatics data interoperable and accessible to retrieval-augmented generation (RAG) workflows that can provide geospatial grounding for LLMs. Rather than relying on custom point-to-point integrations between individual systems, interoperability should occur at the point where data are published. Such an approach would act as a force multiplier for data reuse by eliminating bespoke integrations and enabling on-demand data fusion across domains.

Discrete Global Grid Systems (DGGS), recently standardized by the Open Geospatial Consortium (OGC), provide an important step in this direction by enabling aggregate and statistical datasets to be aligned to a common geographic reference framework through standardized zone identifiers. When datasets across different domains are published using the same DGGS reference system, they can be integrated directly and consistently across organizational and disciplinary boundaries.

Spatial Knowledge Graphs (SKGs) provide a complementary mechanism for representing geographic entities and their semantic relationships, enabling machines to discover, retrieve, and reason over geospatial context across networks of features. However, while DGGS enables interoperability for statistical and aggregated data, no widely adopted standard currently exists for publishing SKGs in a way that supports interoperability by common geography across independent organizations. Without such a mechanism, spatial knowledge graphs remain difficult to integrate dynamically across domains and data providers.

Proposed Approach

This paper outlines a metamodel that enables disparate organizations to independently publish interoperable spatial knowledge graphs (iSKGs) anchored to common geographies. By adopting a federated architecture inspired by the Data Mesh paradigm, independently managed iSKGs can be combined into a broader Spatial Knowledge Mesh, allowing networks of geographically referenced knowledge to be discovered and integrated on demand. In this architecture, organizations publish domain-specific knowledge graphs that reference shared geographic abstractions while maintaining their own governance and provenance.

The proposed metamodel further supports the propagation of changes across dependent knowledge graphs, enabling updates to flow through the mesh and ensuring that downstream users—particularly those operating in lower-resource environments—can maintain up-to-date geospatial awareness. When combined with DGGS-based data integration and retrieval-augmented generation workflows, the resulting Spatial Knowledge Mesh provides the foundation for Geo-GraphRAG pipelines that allow LLM-based systems to reason over interconnected spatial systems with explicit semantic context.

LLMs lack geospatial awareness because they do not maintain a live model of the Earth or structured graphs of geographic features and their topological relationships. Instead, they learn statistical associations between words rather than networks linking roads, facilities, and populations, and therefore cannot determine which specific roads connect communities to hospitals without structured geospatial data at runtime. They also lack domain-specific geospatial semantics, spatial reasoning capabilities, and up-to-date local knowledge as infrastructure and hazards evolve.

Spatial Knowledge Graphs address these limitations by representing geographic features as interconnected networks of geo-objects and their semantic relationships. In an SKG, each node represents a geographic feature, while edges capture topological relationships or analytical findings derived from geoinformatics data and expressed through semantic labels. Together with a geo-ontology defining the graph schema, these relationships create a structured representation of spatial systems that can be traversed and queried. This structure enables Geo-GraphRAG pipelines in which LLMs translate natural-language questions into graph queries, such as GeoSPARQL or Cypher, enabling geospatial insights to be retrieved with full transparency and traceability.

To support interoperability across organizations, the metamodel enables geo-ontologies—comprising definitions of geo-object classes and semantic relationships—to be published as first-order graphs that provide shared schemas for interoperable graph creation. Integrations between iSKGs can also be published as reusable first-order entities. Provenance metadata records both the original source of attribute values and geometries and the publishing organization, supporting traceability and integrity using existing standards such as GeoDCAT. iSKGs can be published with explicit periods of validity and version identifiers to capture temporal evolution. Open-source software implementing this metamodel has been developed to support the creation of geo-ontologies and iSKGs, as well as graph operations such as pull, merge, and change detection across federated knowledge graphs.

This work has been supported by the U.S. Army Corps of Engineers (USACE) Civil Works Division and, through the OGC, by Natural Resources Canada and the United States Geological Survey as part of a collaborative research effort to integrate geospatial awareness into LLMs for disaster response and resilience. The proposed metamodel extends OGC Building Blocks—which model dependencies between specifications and promote reuse across geospatial standards—by adding support for change propagation, temporal alignment, and spatial knowledge graph interoperability.

Expected Impact

By providing a standard approach and open-source software for publishing geoinformatics data as interoperable geospatial knowledge networks accessible to large language models, this work aims to act as a force multiplier for positive impact. In particular, it seeks to enable lower-resourced settings to develop bespoke GeoAI solutions by providing access to integrated geospatial knowledge infrastructures optimized for use with LLM-based systems.

Transit accessibility is a crucial component of sustainable urban mobility, as it directly influences the effectiveness of public transportation systems. Many cities throughout the world use transit-oriented development (TOD) plannings to make it easier to travel to public transportation like bus and train stations. The goal of TOD policies is to get individuals to take public transit instead of private vehicle dependence. Using fixed-distance radial buffers to mark the areas surrounding stations where people can catch the bus or train is a common technique to plan transit. TOD plannings often provide guidance in improving first and last mile connectivity by using radial buffers. The distance between these buffers is normally between 400 and 1500 meters. These buffers are widely used in planning frameworks and policy recommendations since they are simple to use and understand. Radial buffers are useful, however they don't necessarily indicate how easy it is for people to get to locations.

Standard Euclidean radial buffers assume as if people can walk straight from any location in the buffer to the station. Cities are a lot more complicated than they look. It's often hard for people to move around because of obstacles like buildings, fenced developments, and other physical barriers. Because of this, the real distance and time it takes to walk to a station can be significantly different from what radial buffers suggest they are. This oversimplification could lead to incorrect estimations of how easy it is to travel to stations, misleading representations of area coverage, and possibly faulty planning decisions in policies that are meant to encourage transit-oriented development.

This study therefore presents a GIS-based walking isochrone that utilizes open-source geospatial data and network analysis techniques to delineate transit station catchment regions. Walking isochrones show where you can walk to in a specific length of time or distance on an actual street or pedestrian network. The work employs open-source geographic data and GIS tools to generate walking isochrones around designated rail transit stops. This study locates places that can be accessed within 5, 10, and 15 minutes of walking by service area analysis via network data from OpenStreetMap (OSM). In addition to time-based accessibility, catchments based on walking distances of 400 m, 800 m, and 1500 m are also studied. This was done using open-source geospatial data as well as open-source tools such as QuickOSM and QNEAT3. The values are in line with the planning standards that are typically used for making transit stations catchments in TOD guidelines. Following that, the isochrone polygons are compared to normal Euclidean radial buffers that are the same distance from the center.

Several spatial indicators are analysed to assess the differences between typical buffer and isochrones. First, the catchment areas are calculated to determine differences in size between isochrone-based service areas and radial buffers. Second, population are evaluated by estimating the population contained within each catchment area. Third, the entropy index or diversity of surrounding land uses are examined. Land-use mix indices are calculated to provide a quantitative representation of activity diversity within each catchment area. TOD documents often promote land use mix to increase public transportation usage.

The findings of this study indicate significant discrepancies between Euclidean radial buffers and network-based walking isochrones. Radial buffers tend to overestimate the accessible walking area within a given timeframe or distance. This difference is particularly obvious in urban environments characterized by inefficient street connectivity or physical barriers that hinder direct pedestrian movement. Subsequently, isochrone-derived catchment areas are typically smaller and reveal less regular geometries compared to radial buffers. Furthermore, network-based isochrones provide a more accurate representation of the population that can feasibly walk to a given station. The examination of land-use mix also demonstrates that the approach used to find the catchment region can change how much land-use variety is found for the station area. Isochrone-based catchments might not include some land-use zones that are in radial buffers but are cut off by barriers or street networks that don't connect well. The results have a big impact on how we plan for public transit and how we judge TOD policy. If planners simply use radial buffers to figure out station catchments, they might think that transit infrastructure is simpler to get to than it really is and that public transportation serves a greater population than it really does.

This study not only supplements to the ways that people plan transportation, but it also shows how helpful open-source geospatial tools and datasets can be. Anyone can use open-source GIS tools and data sources, which are what the whole process of this study relies on. For instance, OpenStreetMap is utilized to obtain data about road networks, and GIS-based network analysis is used to create service areas. The Free and Open Source Software for Geospatial (FOSS4G) group has goals that are similar to this study. It supports open, accessible, and repeatable geospatial research. Because it employs open-source data and analytical methodologies instead of high-priced software or proprietary datasets, cities all around the world may replicate this study in their own cities.

All in all, this study illustrates how open geospatial systems might enhance advanced spatial analysis in urban transportation planning. It demonstrates the practical application of open data and GIS-based network analysis for evaluating public transit accessibility. Furthermore, it also reveals how open-source methodologies can enhance the precision of planning assessments. The proposed method not only contributes to scholarly discussions regarding the measurement of transit accessibility but also provides planners, policymakers, and academics seeking to enhance first- and last-mile links in urban transit systems with improved strategies. This study ultimately provides a replicable approach for assessing transit station catchment areas, grounded in open-source geographical data. The results improve the methodological accuracy of transit accessibility studies and help planners make better decisions about sustainable urban transportation and TOD plans.

This study aims to evaluate the walkability of the area surrounding Shin-Yurigaoka Station for older adults, which is located in a hilly, suburban environment. This will be achieved by analysing the street network as a whole using a gradient-aware network approach.

In Japan, walkability has become an increasingly important issue in the context of rapid population ageing. Older adults continue to go out frequently in their daily lives and walking remains one of their main modes of travel. In this context, the ease of walking is shaped not only by distance or network connectivity, but also by the physical burden of slopes. This issue is particularly relevant in Japanese suburban hilly areas, where many residential districts have been developed on hilly terrain, often requiring residents to negotiate slopes and stairways to access stations, shops, and everyday services.

Against this background, this study focuses on the area around Shin-Yurigaoka Station in Asao Ward, Kawasaki City, examining how the walkability of older adults can be understood through a combined perspective of street network structure and slope conditions. Asao Ward is an appropriate study area for two reasons. Firstly, according to the 2020 Municipal Life Tables published in 2023, Asao Ward had the highest average life expectancy in Japan for both men and women, making it a notable area for longevity. Secondly, despite its hilly terrain, a Kawasaki City survey found that a significant proportion of older residents in the ward reported being able to walk for around 15 minutes or doing so in their daily lives. This suggests that walking remains an important mode of everyday mobility for older adults, even in areas with many slopes.

This raises a key analytical question: in hilly urban environments, do roads that are structurally central to the street network also function as physically walkable routes for older adults? To address this question, the study examines the degree of overlap between network centrality and low-slope conditions.

In Space Syntax research, configurational measures such as Integration have been linked to pedestrian movement (Hillier et al., 1993). In contrast, walkability is understood to be a multidimensional concept influenced not only by road connectivity, but also by factors such as access to destinations, land use and safety (Saelens & Handy, 2008). Studies of older adults have also shown that walking and physical activity are influenced by neighbourhood environmental factors, including walkability, access to destinations, and pedestrian infrastructure (Barnett et al., 2017). These findings suggest that street-network centrality alone may not fully explain actual walkability, particularly in hilly areas. Therefore, this study introduces a gradient-aware network analysis that integrates street-network centrality with topographic conditions.

The analysis was conducted using open geospatial data and open-source software. Road network data were obtained from OpenStreetMap and a 5-metre digital elevation model (DEM) was used to represent topography. To better focus on walkable public routes, roads classified as parking areas, private roads and indoor roads were excluded from the analysis. The study area was defined as a 15-minute walking catchment around Shin-Yurigaoka Station, based on an assumed walking speed of 1.0 m/s for older adults. In QGIS, the roads were divided into 10 m segments and the longitudinal gradient of each segment was calculated based on the elevation difference between its start and end points. In parallel, Integration values were derived through Angular Segment Analysis at a radius of 900 m (R900) using DepthmapX and the Space Syntax Toolkit. These values were then assigned to road segments at intersections. Integrating these variables enabled the study to construct a gradient-aware network analysis framework, evaluating each road segment in terms of its configurational importance within the overall network and walking difficulty due to slope.

The results reveal several important patterns. Within the 15-minute walking catchment area of Shin-Yurigaoka Station, high-integration roads — defined as the top 20% of the network in terms of integration value — accounted for 23.9% of the total road length. Low-gradient roads, defined as having a gradient of 8.0% or less based on Japanese sidewalk design standards, accounted for 84.7% of the total. Roads that satisfied both conditions simultaneously — namely, roads that were both highly integrated and low in gradient — accounted for 21.6% of the total. Examining the intersection of these two conditions more closely, it was found that 90.8% of high-integration roads were also low-gradient roads. This suggests that many of the structurally central roads around the station have relatively gentle slopes. In contrast, only 25.6% of low-gradient roads were classified as highly integrated. This suggests that, while gentle-slope roads are widely distributed across the study area, they do not necessarily form the core of the street network.

These findings suggest that the walkability of a hilly urban area for older adults should be evaluated not only in terms of the physical ease of the slope, but also in terms of the role that a road plays within the overall network. In other words, roads that are easy to walk on do not necessarily occupy a central or strategic position in everyday movement patterns. This is an important consideration when it comes to understanding mobility in suburban hilly areas, where the topography can reshape the relationship between urban structure and practical pedestrian accessibility. The study also demonstrates the value of using open street and elevation data alongside open-source spatial analysis tools to examine this issue from a reproducible and scalable perspective.

As a next step, road data in OpenStreetMap and QGIS will be refined to ensure the analytical network more accurately reflects the actual pedestrian environment, including sidewalks and other walkable links. The analysis will also be extended to a 30-minute walking catchment area to compare the relationship between centrality and gradient at a broader spatial scale. Furthermore, future work will validate the gap between the analytical results and the actual physical environment in order to examine the validity and applicability of gradient-aware network analysis for evaluating walkability for older adults in hilly urban areas.

Most of today's web maps are using the Web Mercator projection, which has a major distortion of area sizes far from the equator.

This talk shows recent improvements in web mapping libraries for using the Equal Earth map projection in interactive web maps and discusses remaining obstacles.

This talk presents a practical playbook for taking GeoServer to production, covering performance tuning, data preparation, caching strategies, and operational controls. Drawing from real enterprise deployments at GeoSolutions, including GeoServer Cloud, it provides actionable guidance to build stable, scalable, and high-performance geospatial services.

First, an history of CRS at OGC: how WKT and abstract model evolved in parallel, what GML can do, why it nevertheless became legacy encoding. Then a look ahead: how OGC handles JSON encoding in future standards, implications for a CRS JSON, and revision of ISO 19111 abstract model.

Comparing differences between basemaps or datasets in QGIS is difficult when only one map can be viewed at a time. QMapCompare solves this by bringing interactive map comparison into QGIS.
The plugin supports multiple visualization methods to quickly reveal changes across styles and datasets.

We'll discuss recent advancements in fAIr. Specifically, we'll explore our approach to enabling developers to share their GeoAI models for humanitarian mapping purposes. We'll also dive into the development of our MLOps architecture, which leverages STAC, MLM, Zenml, and Kubernetes to run models independently in the cloud.

Update on the FOSS Discrete Global Grid Abstraction Library (DGGAL), focusing on support for new Discrete Global Grid Reference Systems (DGGRSs), use of the library in the OGC AI-DGGS Pilot for Disaster Management, and new high-level DGGAL "High Vibes" tools https://dggal.org https://github.com/ecere/dggal

TerriaJS is an open-source TS/JS framework for building rich, 2D and 3D web-based geospatial data platforms such as Digital Earth Australia, the WA Digital Twin, and numerous government and research deployments worldwide. This talk presents the current state of the project.

First ever FOSS4G talk on QuickMapServices, the #1 QGIS plugin. Seriously.

GeoTechnologies has built geospatial data in Japan for nearly 30 years. As we evolve from a traditional map vendor into a dynamic-data company, open-source tools matter more than ever. In this sponsor session, we share who we are and how we hope to collaborate with the FOSS4G community.

Traffic congestion at signalized intersections is a common issue in urban cities, which is often caused by static and inefficient traffic light timings that cannot adapt to changing traffic conditions. These long wait times lead to a significant amount of time loss, higher car emissions, and higher fuel usage. During peak hours or unforeseen disruptions, traditional traffic light control systems that rely on preset or time-of-day-based signal cycles are unfit to handle the dynamic and variable nature of traffic volumes which resulting in slow vehicle movement and increased in travel time. In addition, current traffic light control methods do not include spatial data or intelligent prediction tools that are able to analyze for unpredictable characteristic of traffic in urban areas. This gap highlights that an adaptive method for signal optimization is required. Hence, this study aims to derive the optimal traffic light timing at the junction by integrating Geographic Information System (GIS) and Artificial Neural Network (ANN) models. Unlike the current fixed-time signal control practice, ANN was chosen as it is one of the best machine learning techniques that could be used in predictive modelling. It did not rely on human in making prediction instead, it will learn solely on data without relying on manual assumptions or fixed timing. The historical and current traffic volume data, together with existing signal timing parameters, were used to develop ANN models capable of predicting optimal green time allocations based on traffic demand patterns for each signal phase.

In this study, traffic signal optimisation is conducted for a selected signalised intersection at Section 13, Shah Alam, Selangor, Malaysia. The traffic movement at the intersection is divided into multiple signal phases, including straight and right-turn movements from different approaches. Historical traffic volume data and current traffic volume data are used as input variables in the ANN model, while the existing signal timing serves as the target output. The ANN model is trained to generate new optimal green time durations for each signal phase during morning and evening peak periods. To evaluate the effectiveness of the optimised signal timings, microscopic traffic simulation using SUMO is applied. The existing signal timing and the ANN-predicted optimal timing are simulated and compared using key performance indicators such as traffic volume discharge, queue length, and average waiting time. Through this approach, the study aims to assess how intelligent signal timing optimisation can enhance intersection performance and reduce traffic congestion.

Traffic volume at intersections fluctuates significantly between morning and evening peak hours, as well as across different approaches and movement types. However, the current signal timing plans apply uniform green times that may not correspond to actual traffic conditions, leading to congestion, long queues, and inefficient traffic flow. Therefore, this study focuses on traffic volume characteristics as the primary factor influencing signal timing optimisation, specifically historical traffic volume and current traffic volume for straight and right-turn movements. To address this problem, the first objective of this study is to develop an ANN model to optimise traffic light signal timing. The model used historical traffic volume data as input variables, current traffic volume data as predictor variables, and existing signal timing as the target output. Through training, validation, and testing processes, the ANN model generated optimised green time durations for each signal phase based on five weeks of traffic observations. The results demonstrate that the ANN model is capable of producing adaptive signal timings that better reflect real traffic demand compared to conventional fixed-time control.

The SUMO software was used in this study to simulate the optimized traffic signal timing produced by the ANN model. SUMO is an open-source, microscopic traffic simulation software that enables detailed modelling of road networks, traffic flows, and signal control systems. It provides a flexible platform for evaluating how different signal timing configurations affect traffic performance in terms of delay, queue length, and vehicle throughout. The traffic network for the selected study area was imported from OSM into SUMO. The network editing and configuration were carried out in NETEDIT, which is a graphical tool in SUMO used to edit and reorganize road geometry, lane connections, and traffic light control logic.
The simulation results demonstrate that the ANN-optimized signal timings generally improved traffic performance compared to the existing signal plans. The SUMO simulation enables the evaluation of vehicle throughput by showing the number of vehicles that can pass through the intersection under the new signal timing configuration derived using ANN. The evaluation process involved running two (2) separate simulation scenarios, existing fixed time signal timing and optimized signal timings produced by the ANN model. By simulating these two conditions under almost the same traffic demand, the performance differences could be directly attributed to the signal timings adjustment. SUMO will automatically record several key performance indicators (KPIs) during each running simulation, such as vehicle throughput using Simulation output detector. These indicators were then compared the vehicle volume pass the traffic light intersection.

In conclusion, it is possible for us to use ANN to predict and optimised the traffic light signal timings based on existing traffic volume conditions. By employing historical traffic volume data, the ANN model was able to generate optimal green times for multiple signal phases, thereby instead of setting traffic light timing based on personal experience, assumptions or manual decisions by engineers, this study uses data and the ANN model to decide the timings automatically. In addition, the usage of SUMO for simulation had helped in understanding how much the traffic volume can be controlled with optimized signal timings. This helps avoid human errors, personal references or bias that may affect the traditional signal timing design.

Background

Urban parks provide essential ecosystem services and recreational opportunities that contribute to physical health, mental well-being, and social interaction in dense metropolitan areas. Evidence-based park planning requires accurate and fine-grained measurements of visitor numbers, spatial distribution patterns, and movement trajectories. Conventional data collection approaches—such as manual observation, surveys, and GPS tracking experiments—are labor-intensive, costly, and difficult to implement continuously at scale. Although commercial mobility datasets derived from smartphones provide large-scale behavioral information, they are typically proprietary and financially restrictive. Meanwhile, geotagged social media data have been widely used to estimate recreational visitation; however, these data are often sparse and insufficient for capturing microscale spatio-temporal and trajectory-level behavior patterns.

With the proliferation of Web 2.0 and volunteered geographic information (VGI), volunteered street view imagery (VSVI) platforms such as Mapillary offer a novel open-data source with global coverage. Unlike traditional social media posts that consist of isolated image points, VSVI data are organized into sequential, geotagged photo trajectories, inherently embedding movement information. This structural advantage suggests potential for behavioral analysis at fine spatial and temporal resolutions. Nevertheless, VSVI contributions are generated by self-selected individuals and often dominated by a small number of highly active users. Consequently, it remains unclear whether VSVI contribution behavior reflects general park-use behavior or primarily captures the habits of specific contributors. Furthermore, given the spatial heterogeneity of VSVI contributions and the potential role of data volume in improving data quality, we hypothesize that the statistical validity of such data depends on contribution volume.

Methods

This study empirically evaluates whether VSVI can serve as a reliable proxy for park-use behavior and investigates whether its effectiveness increases with higher levels of contribution. The analysis was conducted in 49 metropolitan parks located in Tokyo’s 23 wards, one of Japan’s major hotspots of Mapillary activity. Three dimensions of park-use behavior were examined: (1) estimation of park visitor counts, (2) spatio-temporal distribution patterns, and (3) travel-path characteristics. Two reference datasets were employed. First, official annual park visitor counts (2019) published by the Tokyo Metropolitan Government were used to validate visitation estimation. Second, commercial Real People Flow Data (May 2023) provided anonymized GPS trajectories from approximately 869,840 users (about 6% of Tokyo’s population), generating around 65 million records per day. After preprocessing—restricting to walking trips within park boundaries and correcting for temporal gaps and cross-park trajectories—523,057 PeopleFlow trajectories were retained for analysis. VSVI metadata were collected from Mapillary (2014–March 2024), yielding 242,534 cleaned image points organized into 2,007 sequences contributed by 45 users within park areas.

Results

To assess visitation estimation capacity, Spearman’s rank correlation analyses were conducted between official visitor counts and four VSVI contribution indicators: number of image points, number of sequences, number of photo-user-days (PUD), and number of creators. When considering all parks collectively, none of the indicators showed statistically significant correlations with official visitor counts. However, when parks were grouped based on mean PUD into low- and high-contribution categories, moderate and statistically significant correlations emerged in the high-PUD group for PUD and number of creators (p < 0.05), whereas low-PUD parks showed no significant relationships. These results indicate that VSVI data may not universally estimate visitation levels but can approximate visitor counts in parks with sufficiently dense contribution activity.

Spatio-temporal distribution patterns were compared at a 20 m × 20 m grid resolution using three indicators derived from both datasets: point counts, dwell time per trip, and average speed. When pooling all parks, statistically significant but weak positive correlations were observed between VSVI and PeopleFlow data for all indicators (p < 0.001). However, park-specific analyses revealed substantial heterogeneity. The proportion of parks exhibiting significant positive correlations was markedly higher in the high-PUD group, particularly for spatial indicators such as point density and dwell time. Improvements for average speed were comparatively moderate. These findings suggest that increased contribution volume mitigates individual behavioral bias and enhances representativeness at the aggregated spatial level.

Travel-path characteristics were further compared at the park level, including temporal features (start time, end time, duration), standardized movement range features (travel distance, standard deviation ellipse (SDE) area, and SDE perimeter normalized by park area), and behavioral features (average speed and deviation rate). Strong and highly significant correlations were identified for standardized travel distance, SDE perimeter, and SDE area, indicating a high degree of consistency between VSVI-derived trajectories and reference mobility data in terms of movement range characteristics. Average speed showed moderate positive correlations. In contrast, temporal features exhibited no significant relationships, suggesting fundamental differences between photo-sharing behavior and general recreational timing patterns. Group-specific analyses indicated that correlations were generally stronger in parks with a higher number of VSVI trips, particularly for movement range indicators, although variability remained across parks and metrics.

Conclusions and implications

Overall, the results demonstrate that VSVI can serve as a meaningful proxy for selected aspects of park-use behavior, especially spatial distribution and movement range characteristics. However, its reliability is strongly conditioned by contribution volume. Parks with higher contribution density—often scenic or tourism-oriented sites—show substantially stronger alignment with commercial mobility data. Low-contribution parks exhibit weaker or inconsistent relationships, underscoring the importance of sufficient data accumulation in VGI-based behavioral inference. For the FOSS4G community, this study demonstrates how crowdsourced open street-level imagery can complement proprietary mobility datasets in large-scale urban behavioral analysis. By empirically examining the relationship between contribution volume and analytical validity, the study contributes to ongoing discussions on data-quality growth, representativeness, and the practical integration of open geospatial data into reproducible urban analytics workflows. While limitations related to GPS accuracy, timestamp uncertainty, and contextual heterogeneity remain, the results highlight the potential of open, crowd-sourced street-level imagery as a scalable complement to traditional data sources in spatial planning and geospatial research.

This presentation explores how Large Language Models (LLMs) and Function Calling can transform web map experiences. Built with the open-source MapLibre GL JS, our application enables users to query geospatial data through natural language — designing smarter experiences that replace complex layer toggling with intelligent, AI-driven interactions.

A development update from the latest version of GeoNetwork, as a 20-year-old project prepares for the challenges of the next decade!

Think of a map as an artistic still life. Our framing often excludes items just out of view. Being completely factual is not the same as being factually complete.
Let’s learn skills to enhance our digital stories from 3D environments. New perspectives lead to improved perception.

GRASS, Geographic Resources Analysis Support System, is a powerful engine for geospatial processing and analysis. This talk delivers the latest GRASS update, covering technical progress, new integration pathways, community developments, and key outcomes from the 2026 community meeting.

PDAL is Point Data Abstraction Library. It is a C/C++ open source library and applications for translating and processing point cloud data. This is an update on the status of the library and new and updated features and capabilities over the last 2 years.

Update on the FOSS libCartoSym, libCSCQL2 and libDE9IM implementing the candidate OGC Cartographic Symbology 2.0 Standard, Common Query Language (CQL2) and Simple Features. http://cartosym.org/ https://github.com/ecere/libCartoSym

A tour of the latest-and-greatest in the STAC software ecosystem, with a focus on demonstrations and use-cases.

Making a tile when it is asked for is not new: vector tiles get built from a database per
request, raster tiles get cut out of Cloud-Optimized GeoTIFFs. But the expensive end stayed
baked. Terrain, and buildings as 3D Tiles, still arrive as pyramids somebody built in
advance — because the arithmetic is heavy, so it gets done ahead of time.

This talk is about three services that do that heavy work inside the request instead, on the
machines that serve a CDN: nothing baked, and no server kept alive.

Re:Earth Terrain and Re:Earth Buildings already work that way — global terrain, and the
world's buildings. Both are usable today: open source, no API key.

And once the drawing happens inside the request too, a map style no longer has to be a list
of rules — a brush can be picked, and a map painted on paper, at the moment someone asks for
it. The third service goes live during this talk.

pygeoapi project status presentation. Come and find out the latest news on the project as well as future plans, and how to get involved!

This talk presents a fully open-source workflow for geospatial modeling of satellite-derived chlorophyll-a data using QGIS, GDAL, and R. We demonstrate how large raster datasets can be processed, interpolated with kriging, and visualized with uncertainty, entirely within the OSGeo ecosystem.

ES|QL is a powerful new declarative query language for Elasticsearch, opening the door to PostGIS-like ease of use for Geospatial querying and analytics.

Pacific Ocean Portal 2.0 is an open-source platform enabling seamless access to ocean data, advanced GIS-based data management, and real-time visualization. It supports collaboration, interoperability, and decision-making, strengthening ocean services and climate resilience across Pacific Island countries.

This presentation demonstrates how GeoServer enables efficient management, discovery, and visualization of large Earth observation datasets. Through real-world examples, it covers indexing (STAC/OpenSearch), Cloud Optimized GeoTIFFs, mosaicking and filtering, data extraction (WCS/WPS), time-based animations, and band algebra, highlighting GeoServer’s latest capabilities in satellite imagery workflows.

This presentation will introduce the attendees to GeoNode's new capabilities. We will provide a summary of the new features added to GeoNode in the last release together with a glimpse of what we have planned for next year and beyond, straight from the core developers.

We introduce "Virtual Shizuoka," a high-density 3D point cloud dataset published as open data by a prefectural government, Shizuoka prefecture. This open data has driven innovation and changed our community. We present its use cases, focusing mainly on transforming the community through collaborative efforts, and discuss its outlook.

An update on the OGC SensorThings API from the OGC standards working group: what's new in v2.0 (to be published in 2026), the open source servers and clients available today, real-world geospatial IoT use cases across sectors, and what's coming next.

This study calculated the travel times to medical institutions in a regional city in Japan using GTFS (General Transit Feed Specification) data to enable a more precise geographic accessibility assessment. It focuses on comparing and validating tools that utilize open data and FOSS4G (Free and Open Source Software for Geospatial). This study reviews previous studies and evaluates GTFS-based route search services, including FOSS4G tools. Based on this review, we developed a QGIS plugin to facilitate participation not only by researchers, but also by practitioners and policymakers.

According to Park (2021), the emergence of sophisticated transportation databases, such as GTFS, enables the estimation of travel times across different transportation modes (e.g., public transit and private automobiles), as well as dynamic travel times under time-variant traffic conditions. The increased availability of dynamic mobility data has therefore facilitated the implementation of time-sensitive accessibility measures.

In Japan, Tanimoto (2020) pointed out that discussions on the selection of analytical methods for accessibility analysis remain insufficient in two respects. First, there has been insufficient discussion on method selection that considers the difficulty and cost of acquiring, preparing, and manipulating data for analysis. Second, there has been insufficient discussion on the effectiveness of the tools in relation to the subject of analysis. Sekine (2018) also highlighted the difficulty of creating geospatial data for intermodal travel chains, although this study was published before the emergence of GTFS data.

Within Japanese geography, the only previous study utilizing GTFS data is Kasahara et al. (2021), who analyzed the spatiotemporal patterns of delay times using Sendai City Bus timetable data. Given the absence of GTFS-based accessibility studies in Japanese geography,. Furthermore, while attempts have been made to compare tools enabling GTFS-based route search services (e.g., Higgins et al., 2021), these studies primarily verified correlations between numerical outputs across tools, rather than validating results in terms of shortest routes or route correctness. In this study, we evaluated the tools in terms of runtime, agreement in travel time estimates, and route validity (e.g., rule violations in GTFS).

We compared and evaluated three tools, ArcGIS, OpenTripPlanner (OTP), and R5, capable of measuring the shortest paths using GTFS. OTP and R5 are FOSS4G tools, whereas ArcGIS is a proprietary software. To examine accessibility challenges in regional cities in Japan, we selected the area around Yamagata City as a case study. Although the study area has an intricate network of railway and bus routes, automobile dependence remains high; therefore, improving public transport services is an important local challenge. Using each tool, we measured the shortest path travel times to medical institutions and compared and validated the results.

In our evaluation, approximately one million origin–destination (OD) pairs were generated and analyzed using each tool in the same computing environment. ArcGIS required approximately 10 min to calculate one million OD pairs, OTP required over 60 min, and R5 required approximately 5 seconds. These figures are broadly similar to those reported by Higgins et al. (2021), confirming R5’s substantial speed advantage.

For R5 and OTP, the distribution of travel times per OD pair was nearly identical, with 93.2% falling within a 5-minute difference. By contrast, ArcGIS had a match rate of less than 50%. Furthermore, the cases in which only ArcGIS derived the shortest path were limited to paths that violated operational rules, such as boarding at a GTFS-registered stop designated for drop-off. R5 and OTP search for the shortest path at each departure time, whereas ArcGIS first determines the single shortest path and then checks whether a journey exists at that departure time along that path. Consequently, ArcGIS has difficulty in producing accurate results for OD pairs with multiple feasible routes or stops.

OTP and R5 travel time estimates matched those from a popular Japanese route search service in approximately 70% of cases, calculated shorter travel times in approximately 25% of cases, and produced longer travel times for the remaining 5%. Most discrepancies were due to different definitions of walking distance to stops, and the differences were within an acceptable margin (often within ±3 minutes).

Although OTP and R5 produced accurate results in most cases, travel times differed by more than five minutes in 6.8% of cases. Most of these discrepancies were resolved by shifting the analysis time by a few minutes. R5 calculated the shortest time in most cases; however, when boarding or alighting at stops more than 300 m from the origin or destination, OTP identified the shortest paths. This appears to be due to R5 prioritizing stops within 300 m and searching further if no route is found.

Overall, OTP and R5 are suitable tools for measuring shortest-path travel times using GTFS. OTP can calculate the highest number of shortest paths. However, because OTP is substantially slower than R5, testing many scenarios using R5 may be more practical for exploring improvements in accessibility. Accordingly, when R5 cannot calculate the shortest identifiable pathh.

These results enable the calculation of reachable populations based on arrival times at each medical institution, for example, using R5’s high-speed and precise travel time estimation. As a result, we provided a QGIS plugin utilizing R5. The plugin constructs OD tables through graphical user interface (GUI) operations, eliminating the need for commands and adding statistical values such as travel times to the QGIS as GIS data.

The ZOO-Project is an open-source software platform that implements the OGC API Processes standard, providing a comprehensive solution for creating, deploying, and managing web processing services tailored to geospatial and Earth Observation (EO) applications.

Morining Break sponsored by MIERUNE Inc. & Cesium

What can geospatial technology do for memory — and for peace? This keynote explores how maps, satellite imagery, and digital archives preserve and pass on the memory of war and disaster across generations, from Hiroshima to conflicts and disasters unfolding today. The session combines a keynote presentation, testimony from Hiroshima, and a moderated dialogue — inside Peace Memorial Park itself.

STAC enables structured geospatial search, but GeoAI introduces semantic search with vector embeddings. This talk shows how to combine both, using STAC for discovery and embeddings for similarity, to support modern geospatial analysis workflows.

Maplat is an open-source platform enabling bidirectional coordinate transformation between historical or illustrated maps and modern maps. Celebrating its 10th anniversary, this talk covers the project's technology, real-world applications, and philosophy — including ongoing developments and open challenges we would love to explore together with the community.

mago3DTiler and mago3DTerrainer are open-source tools designed to support standards-based digital twin workflows from diverse geospatial data.

This session highlights the latest capabilities of both projects, including OGC 3D Tiles 1.1 support, terrain generation, GIS-to-3D conversion, and performance optimization, showing how they simplify the creation of interoperable digital twin applications.

We conclude by sharing the projects' roadmap and our vision for continued open-source collaboration.

Urban Heat Island (UHI) effects are among the most visible climatic consequences of rapid urbanization, driven by the expansion of impervious surfaces, increasing building density, and changes in vegetation cover. Satellite remote sensing provides consistent spatial coverage and repeated observations over long time periods, making it a valuable tool for analyzing these processes. In particular, land surface temperature (LST) derived from multispectral satellite missions enables systematic monitoring of thermal conditions in urban environments. Satellite observations have therefore become an important source of information for examining how urban thermal patterns evolve over time.
However, interpreting multi-year satellite-derived land surface temperature time series is not always straightforward. LST observations often show strong interannual variability influenced by meteorological conditions, seasonal differences in data availability, and changes in land cover characteristics. Distinguishing a stable long-term urban warming signal from short-term fluctuations therefore remains a methodological challenge in many urban heat island studies. This difficulty becomes particularly relevant when relatively short satellite records are analyzed, as individual anomalous years can influence the interpretation of long-term trends.
This study investigates how stable UHI signals can be detected in multi-year satellite-derived land surface temperature time series using a reproducible open-source geospatial workflow. The analysis is demonstrated for the city of Zagreb, Croatia, using a ten-year satellite record (2015–2024) derived from Landsat observations (Landsat 8/9). Instead of analyzing the entire urban area as a single unit, the study focuses on several representative urban neighborhoods used as analytical units for evaluating temporal patterns in the temperature time series. Using several analytical units also allows temporal patterns to be compared across different parts of the city rather than relying on a single aggregated urban value.
Zagreb represents a typical Central European urban environment characterized by a mixture of dense built-up areas, residential neighborhoods, urban green spaces, and peri-urban zones. To capture this diversity, several representative neighborhoods were selected as analytical units. The selected areas reflect different urban development patterns within the city, including contrasts in vegetation cover and building density. This allows the analysis to compare thermal behavior across distinct types of urban environments within the same metropolitan area. Such variation in land cover and urban structure provides a useful setting for examining how temperature dynamics differ between neighborhoods with different surface characteristics.
Annual warm-season composites of land surface temperature are generated to reduce noise associated with individual satellite scenes and to provide a consistent representation of thermal conditions for each observation year. To place temperature dynamics in the context of surface characteristics, two commonly used spectral indices are analyzed alongside LST: the Normalized Difference Vegetation Index (NDVI) and the Normalized Difference Built-up Index (NDBI). Analyzing these indices together with temperature values helps interpret whether observed thermal changes are associated with variations in vegetation or built-up surfaces.
Satellite data processing follows a reproducible workflow that integrates image preprocessing, time series aggregation, and statistical analysis. Individual Landsat scenes are first filtered according to cloud cover and seasonal criteria. Annual warm-season composites are then generated to reduce scene-level variability and provide consistent yearly observations. LST, NDVI, and NDBI layers are subsequently extracted for each selected neighborhood, forming the basis for the temporal analysis. Using a consistent processing workflow also helps ensure that results from different years remain comparable.
Particular attention is given to the temporal behavior of the data through the analysis of interannual differences between consecutive observations. This allows year-to-year variations in land surface temperature to be examined together with corresponding changes in vegetation and built-up surfaces. Interannual differences help reveal short-term dynamics and provide additional insight into whether temperature changes occur in parallel with changes in surface characteristics. These comparisons provide an additional perspective on the variability present in the satellite record.
To assess the robustness of the detected trends, several analytical approaches are used. Linear trends and non-parametric slope estimates are calculated for the LST time series, and the results are compared across different parts of the observation period. This makes it possible to see whether similar trends appear regardless of the method used or the selected time interval. In this way, it can be evaluated whether the observed temperature signal reflects a persistent long-term pattern or mainly short-term variability in the satellite record. Examining the consistency of trend estimates therefore provides an additional check on the stability of the detected patterns.
The results help interpret satellite-derived temperature time series in urban heat island research more carefully. Examining temporal changes, interannual differences, and the consistency of trend estimates helps identify whether similar warming patterns appear across different parts of the time series. This comparison makes it possible to separate persistent temperature trends from short-term variability in the satellite observations.
Such an analysis also shows that time plays a key role when interpreting urban heat island dynamics. The use of reproducible open-source geospatial tools allows the same workflow to be applied to satellite temperature time series in other urban areas. This makes the approach potentially applicable to other cities where comparable satellite datasets are available.

GeoServer 3 is a significant update for this beloved OSGeo Application. This presentation provides practical tour of GeoServer 3 covering upgrade steps, new requirements, and refreshed UI!

A lightweight Geospatial Data Gateway built with Go that composes spatial features from multiple databases and APIs at request time, serving enriched GeoJSON through OGC API Features standard without upfront data preparation.

Visualizing national climate projection data in the browser requires going beyond standard raster layers. We present maplibre-gl-shader-layer, an open source TypeScript library enabling custom WebGL shaders on MapLibre GL JS — demonstrated through Météo France climate datasets rendered with high-precision multi-channel tile encoding and configurable colormaps.

OGC API Environmental Data Retrieval is a powerful new API standard for accessing time series data about geospatial areas. This session will explain how EDR standardizes time series data access and how it can be integrated with RDF vocabularies and JSON-LD to standardize cross-organizational terminology.

This framework simplifies STAC data access by using Gemma 3 to translate natural language into queries. It automates GISTDA’s disaster data retrieval and MapLibre visualization, transforming complex geospatial imagery into instant, actionable intelligence. This human-centered AI approach ensures rapid, expert-free decision-making during crises for societal safety.

Geographic Information System (GIS) data is inherently multidimensional, encompassing spatial extent, temporal dynamics, and diverse attributes. Understanding and presenting this complex data through cartographic visualization requires both comprehensive geospatial knowledge and sophisticated map design skills. However, a significant portion of geographic platform users—including web developers, data analysts, and professionals without formal cartography training—frequently encounter substantial challenges throughout the map visualization workflow. These difficulties span from initial data interpretation and selection of appropriate visualization methods to the configuration of critical cartographic elements such as color schemes, symbol sizes, opacity levels, and overall compositional layout.

In contemporary web mapping systems, data visualization is predominantly controlled through Map Style JSON, a structured format that governs data rendering in mapping libraries such as MapLibre GL JS, which operates according to the MapLibre Style Specification standard. While this specification is open and highly flexible, creating appropriate styles from actual datasets remains a task demanding both technical expertise and design experience. Consequently, many users invest considerable time manually experimenting with style adjustments, often through trial and error, which can be both frustrating and inefficient.

This presentation proposes an innovative approach to simplifying the map design process through the development of a Model Context Protocol (MCP) for automated Map Style JSON generation from users' spatial vector data. This data can be sourced through database connections or API service integrations. The fundamental concept underlying MCP is the creation of a "context layer" that enables language models to systematically understand the structure and semantic meaning of GIS data before applying this understanding to map style generation.

The proposed architecture integrates MCP with open-source Large Language Models (LLMs) capable of operating locally through Ollama. The system analyzes users' spatial data characteristics, including geometry types, attribute structure, and data distribution patterns. Subsequently, it automatically generates MapLibre-compliant Map Style JSON that can be immediately deployed in web mapping applications. This local processing capability addresses both performance and data privacy concerns that often arise with cloud-based solutions.

The distinctive advantage of this approach lies in how MCP extends beyond merely ensuring structurally correct JSON generation. The protocol fundamentally incorporates cartographic design principles based on the "Perceptual Properties of Linear and Spatial Systems" into the decision-making process. This integration manifests in several critical ways: the selection of color schemes aligned with data semantics, the assignment of appropriate symbols corresponding to geometry types, and the strategic application of color tones and opacity levels to enhance user perception and readability. The system also accommodates datasets with multiple classification classes and leverages modern color palettes to ensure that spatial data visualization achieves both clarity and accessibility.

The technical implementation combines several key components working in concert. First, the MCP server acts as an intermediary layer that processes incoming spatial data, extracting relevant metadata and structural information. This includes analyzing coordinate reference systems, identifying attribute data types, detecting statistical distributions, and recognizing spatial patterns that inform styling decisions.

The language model component, running locally through Ollama, receives this contextualized information and applies learned cartographic principles to generate appropriate styling rules. The model has been trained to understand the relationships between data characteristics and visual representation best practices. For instance, when encountering categorical data with distinct classes, the system automatically selects qualitatively different colors that maximize perceptual distinction. For continuous numerical data, it applies sequential or diverging color schemes appropriate to the data's semantic meaning.

The generated Map Style JSON adheres strictly to MapLibre specifications, ensuring immediate compatibility with MapLibre GL JS and other compliant rendering engines. The output includes properly structured layers, sources, paint properties, and layout configurations that reflect both the data's inherent characteristics and established cartographic conventions.

A critical innovation of this approach involves embedding cartographic design expertise directly into the generation process. Traditional automated styling systems often produce technically correct but cartographically naive outputs. This MCP-based system incorporates several levels of design intelligence:

Perceptual hierarchy: The system understands which data elements should be visually prominent and adjusts styling properties accordingly, considering factors such as feature importance, scale-dependent visibility, and visual contrast.

Color theory application: Beyond simple color assignment, the system applies principles of color harmony, considers color blindness accessibility, and ensures adequate contrast ratios for legibility across different display conditions.

Symbolic representation : The selection of point symbols, line patterns, and fill styles reflects both the semantic meaning of the data and established cartographic conventions, making maps intuitively interpretable even for non-expert users.

Scale responsiveness: Generated styles include appropriate zoom-level dependencies, ensuring that map elements appear at suitable scales and with appropriate levels of detail.

The objectives of this presentation extend beyond merely demonstrating automated Map Style JSON generation. Fundamentally, this approach democratizes quality map production, enabling individuals without GIS backgrounds to create professional-quality cartographic visualizations. This democratization has significant implications for data journalism, civic participation, educational applications, and small organizations that lack dedicated GIS expertise.

Furthermore, the utilization of open-source models and architecture capable of local execution through Ollama aligns perfectly with the principles of the Open Geospatial Ecosystem. This approach ensures data sovereignty, eliminates dependency on proprietary cloud services, and facilitates integration with other open-source tools prevalent in the geospatial community. The system can be extended and customized by users, fostering innovation and adaptation to specific domain requirements.

The protocol-based architecture also enables future enhancements and integrations. As language models continue to evolve, the MCP layer provides a stable interface that can leverage improved capabilities without requiring fundamental system redesign. Additionally, the approach can be extended to incorporate user feedback, learning from styling preferences and iteratively improving recommendations.

In this presentation, our goal is not only to automatically generate Map Style JSON from your data but, more importantly, to empower individuals without GIS expertise to create high-quality map visualizations with ease. Additionally, by utilizing an open model and architecture that can operate on users' devices via Ollama, we align with the principles of an open geospatial ecosystem and enable integration with other open-source tools within the community.

This talk shares results from a security review of widely used open‑source GIS libraries using a lightweight SAST and SBOM methodology. Attendees will learn how common patterns create risk and how simple, repeatable practices can strengthen the security and resilience of geospatial tools.

Streamlining Construction Infrastructure with Web-Based 3D Digital Twins.
This solution empowers anyone to effortlessly plan and design LandXML-compliant temporary roads directly in a web browser. Leveraging CesiumJS and advanced 3D Tiles, it transforms complex civil engineering workflows into an intuitive, visual, and highly efficient experience.

ArkEdge Insights is a web-based geospatial information platform that integrates satellite imagery, drone data, IoT sensor data, weather data, and public datasets for analysis, visualization, and decision-making. In this lightning talk, ArkEdge Space will introduce how the platform uses open source geospatial technologies such as MapLibre GL JS, STAC, and TiTiler to build reusable applications for agriculture, disaster response, environmental monitoring, and climate adaptation. We will briefly share practical lessons from developing user-oriented geospatial services that connect Earth observation data with real-world operational needs.

GeoSolutions is a core contributor to GeoNode, GeoServer, GeoTools, GeoWebCache and MapStore. This session shows, through concrete examples, how our client-first enterprise services — support, training, and custom development — consistently strengthen open source software.

Natural and technological disasters pose significant threats to communities, infrastructure, and the environment worldwide. Effective disaster risk management requires robust analytical frameworks capable of systematically assessing hazard, vulnerability, and exposure components across multiple disaster types, recognizing the complex and cascading nature of modern risks (Pescaroli & Alexander, 2018). This paper presents a comprehensive disaster risk analysis model that integrates Geographic Information Systems (GIS), machine learning algorithms, and Artificial Intelligence (AI) driven interpretation tools (GeoAI) to support multi-hazard risk assessment, urban resilience planning, and citizen-oriented risk communication.

The proposed model is designed to be generalizable across a wide range of disaster types, including floods, wildfires, earthquakes, landslides, droughts, and urban heat island events. Rather than developing isolated models for each hazard category, the framework adopts a unified, modular architecture in which the core analytical pipeline is consistently applied regardless of the disaster type under investigation. This design philosophy ensures scalability and reproducibility, addressing the critical need for understanding social vulnerability to environmental hazards (Cutter, Boruff & Shirley, 2006). This framework integrates geospatial analysis with supervised machine-learning classification models. Hazard and vulnerability layers are constructed using multi-source spatial datasets, including topographic, land-use and land-cover maps, hydrological networks, soil-type and lithology data, historical disaster records, population-density grids, and socioeconomic indicators. These spatial layers are processed and analyzed in QGIS, building on established approaches that employ geoprocessing to evaluate regional risks of major urban hazards and improve public safety (Zhao & Liu, 2017; Contini et al., 2000).

Once the geospatial feature matrices are assembled, machine learning classification algorithms are employed to derive hazard and vulnerability maps. The study evaluates and compares several state-of-the-art ensemble learning algorithms, with particular emphasis on Random Forest (RF) and Extreme Gradient Boosting (XGBoost). The integration of hybrid models and bivariate statistics has been shown to significantly enhance the accuracy of hazard mapping in GIS-based comparative assessments (Ali et al., 2020). Python-based machine learning libraries are used to implement, train, validate, and tune these models rigorously. Cross-validation strategies and hyperparameter optimization techniques are applied to ensure robust and generalizable model performance. Accuracy metrics such as the Area Under the Receiver Operating Characteristic Curve (AUC-ROC), F1-score, precision, recall, and Cohen's Kappa coefficient are used to evaluate model outputs.

The exposure component of the risk model quantifies the number and nature of elements at risk within identified hazard zones, including population counts, building footprints, and critical infrastructure. To accurately reflect structural vulnerabilities, the model framework allows for the integration of fragility curves, a method widely used for generating damage probability estimates for structural typologies like reinforced concrete buildings under seismic loads (Akkar & Yakut, 2005; Ramamoorthy, 2006). Spatial overlay operations between hazard maps and exposure layers yield exposure matrices that inform prioritization in disaster preparedness.

A particularly novel contribution of this work is the integration of AI agent architectures based on Large Language Models (LLMs) and the Model Context Protocol (MCP) into the risk assessment pipeline. The MCP serves as a standardized communication interface enabling LLM-based AI agents to dynamically query, retrieve, and reason over geospatial risk data stored in structured databases and spatial APIs. These AI agents are capable of performing automated, context-aware interpretation of multi-hazard risk maps and statistical outputs, generating region-specific risk narratives that translate complex quantitative results into clear, actionable insights. By leveraging the natural language generation capabilities of LLMs, the system can produce detailed risk assessment reports tailored to specific administrative units, districts, or neighborhoods, significantly reducing the analytical burden on planners and decision-makers. The MCP-based architecture also enables seamless integration with external data sources and third-party geospatial services, enhancing the model's capacity for real-time or near-real-time risk monitoring.

All data inputs to the model are sourced from open and freely accessible repositories, including OpenStreetMap for infrastructure and land use data, Copernicus Land Monitoring Service and USGS Earth Explorer for remote sensing products, NASA EARTHDATA for climatological and hydrological datasets, and national/regional open government portals for administrative and socioeconomic statistics. This open data philosophy not only ensures cost efficiency but also enables transparency, reproducibility, and independent verification of results by the broader scientific community and policy stakeholders.

The outputs of the model are made accessible through a web-based geospatial platform that allows real-time visualization of hazard susceptibility maps, vulnerability indices, exposure layers, and composite risk scores across multiple administrative levels. The platform is designed with usability in mind, targeting both technical users such as urban planners, civil protection agencies, and researchers, and non-technical users including local government officials and ordinary citizens. Interactive map layers, filtering tools, and downloadable reports are provided to support diverse user needs. A key feature of the platform is an integrated AI-powered chatbot, developed using LLM technology and connected to the underlying risk database via the MCP interface. Citizens and decision-makers can query the chatbot using natural language to obtain plain-language explanations of the risk status for their area, receive guidance on risk-reducing behaviors, and access information about emergency preparedness resources. This conversational interface significantly lowers the barrier to understanding complex risk information and fosters greater public awareness and engagement.

The model was developed and piloted in the Marmara Region of Türkiye, encompassing 11 provinces: Istanbul, Bursa, Kocaeli, Sakarya, Tekirdağ, Edirne, Kırklareli, Balıkesir, Çanakkale, Yalova, and Bilecik. The Marmara Region has outstanding strategic importance due to its high population density, industrial concentration, seismic activity, and exposure to multiple natural hazards. This geographically and socioeconomically diverse study area provided an ideal testbed for evaluating the model across varying physical and demographic conditions, allowing for a comprehensive evaluation of its cross-hazard applicability and cross-regional transferability. Pilot results demonstrated strong classification accuracy for hazard susceptibility mapping across all tested disaster types, with XGBoost consistently achieving higher AUC scores compared to baseline models. The LLM-based risk interpretation layer produced coherent, factually grounded, and contextually relevant narratives that were positively evaluated by domain expert reviewers.

In conclusion, the GeoAI-based disaster risk analysis model presented in this paper offers a scalable, open, and interoperable framework that bridges advanced spatial analytics, machine learning, and generative AI to support evidence-based disaster risk reduction. By making risk information accessible through web platforms and conversational AI tools, the model contributes directly to the development of sustainable, smart, and resilient cities aligned with the Sendai Framework for Disaster Risk Reduction 2015–2030 and the United Nations Sustainable Development Goals (UNDRR, 2015). Future research directions include the incorporation of real-time sensor data and Internet of Things (IoT) technologies to support global initiatives for dynamic multi-hazard early warning systems (UNDRR, 2024).

To address data integration and interoperability, we conducted a PoC using OSS “Ouranos GEX” in our high-speed spatiotemporal data management technology. We present PoC results and discuss the development of the Java-based OSS, with a focus on technical challenges and lessons learned.

Vector tiles are a lightweight map data format that enables fast, client-side map rendering with libraries such as MapLibre.
One of their key advantages is the ability to create multiple map visualizations from the same data simply by switching styles.
At PASCO CORPORATION, we manage these vector tiles in the cloud-native PMTiles format and provide an API that serves basemaps through our own service, "GeogrAPI."
In this session, we will showcase a live demo site and share the benefits of adopting PMTiles, along with practical insights and implementation techniques gained through building and operating the service.

We built a conversation-first data analysis tool for Japan's MLIT LINKS project that lets non-engineers explore location-rich datasets using natural language. The stack runs entirely in-browser DuckDB-WASM for SQL execution, Claude on AWS Bedrock, and MapLibre GL JS and Vega-Lite for declarative visualization.

This talk details an end-to-end analysis pipeline for early-season crop mapping. We demonstrate how to process multi-sensor Sentinel data , automate CWT scalogram generation , and deploy optimized PyTorch CNN models. This workflow enables rapid, scalable geospatial inferencing by minimizing data sequence requirements.

We present an open, global, building-level resolution, reproducible and dynamic exposure model with the aim to provide global exposure data on the building level for applications in natural hazard risk estimation, resilience planning, disaster recovery, rapid loss assessments and humanitarian tasks

The first hard requirements of the European Cyber Resilience Act (CRA) comes into effect in September of this year. Attend this talk to learn more about what this means for you, your employer, and our open source community.

Learn how project support works at the Open Source Geospatial Foundation, the initial vision of the “incubation committee”, and what is being setup as as a replacement.

Pacific Spatial Solutions (PSS), founded in Tokyo in 2012, builds its business around open geospatial standards and open source software rather than in spite of them. As an OGC member, our team also contributes directly to the community — from OGC/CityGML data pipelines for Japan's Project PLATEAU to a widely used Japanese QGIS textbook co-authored by our CEO and a director.

That same philosophy shapes our product lineup. Fully open-source CesiumJS and TerriaJS power Japan's national 3D city model viewer and Tokyo's Digital Twin. Around that open core, we partner deliberately with tools that strengthen the ecosystem rather than compete with it: FME converts hundreds of municipal datasets into OGC standards; Felt — QGIS's flagship sponsor — maintains Tippecanoe and funds PMTiles/Protomaps; Fused builds its serverless spatial analysis on DuckDB and H3; CARTO, a core PostGIS contributor, open-sources its Analytics Toolbox.

In five minutes, we'll share what more than a decade of treating open standards and open source as a growth strategy, not a cost, has meant for our business — and what we've learned along the way.

Every month Overture Maps Foundation releases more than four billion records of global multi-source map data as GeoParquet files, available in public AWS and Azure buckets. We also release a modular schema as Pydantic packages, a data changelog in Parquet, metadata in STAC, PMTiles, and other open tools to make all of it usable. The spine of Overture's work is the Global Entity Reference System (GERS), an open ID system for maps, and we're building it with 50 member companies. Most map ID systems belong to a single company. An open one is unusual, and building it with companies that compete with each other is harder still. In five minutes I'll cover what a GERS ID is, where it comes from, what the GERS registry captures, and how bridge files help you join your own data to GERS.

Connect with sponsors of FOSS4G Hiroshima 2026 and explore career opportunities in the open source geospatial community.

The Hiring Session brings together FOSS4G Hiroshima 2026 sponsors and conference attendees to learn more about the organizations supporting the event. While the session offers a great opportunity for students, early-career professionals, and job seekers to explore career opportunities, it is open to all attendees interested in learning about the participating companies and their work. As the session takes place during the lunch break, attendees are welcome to enjoy their lunch while listening to the presentations.

Each sponsor will introduce their organization. Presentation times are allocated according to each sponsor's FOSS4G Hiroshima 2026 sponsorship level. Following the presentations, attendees and sponsors will have time for informal networking and conversation.

https://2026.foss4g.org/en/program-schedule/hiring-session/

Lunch sponsored by Eukarya/Re:Earth & Geo Technologies

Released 1.5 years ago, GeoParquet 1.1's features had evolving ecosystem support in late 2025. We built a custom spatial indexing pipeline using Dagster and GeoPandas while waiting for native tools to fully mature.

This presentation introduces a QGIS plugin developed under Japan’s Project PLATEAU to help municipal staff calculate and visualize urban structure evaluation indicators using 3D city models and public datasets.

The increasing volume and heterogeneity of digital research data have intensified the need for integrated research data management (RDM) infrastructures that support the full lifecycle of data stewardship across scientific domains. In agricultural and environmental sciences, datasets commonly combine geospatial information, experimental measurements, tabular records, and model outputs, demanding systems that align with the FAIR principles—ensuring data are findable, accessible, interoperable, and reusable. While many research repositories provide strong metadata and archival capabilities, they often lack interactive exploration tools, whereas geospatial platforms, though rich in visualization features, are typically not designed for comprehensive RDM workflows. This disconnect hinders data reuse, as users cannot assess dataset content, structure, or spatial context without downloading data or using specialized software. To bridge this gap, this work presents the adaptation of GeoNode as a unified RDM infrastructure that integrates robust repository functionalities—such as persistent identifiers, extended metadata models, and structured publication workflows—with advanced web-based visualization and interactive exploration capabilities. By enabling users to inspect both geospatial and non-geospatial datasets directly in the browser, the platform facilitates early understanding of data context and quality, significantly improving discoverability and reusability. This approach demonstrates how GeoNode can be extended beyond its traditional role to serve as a scalable, FAIR-compliant foundation for managing diverse research data in agricultural and environmental sciences.
The presented platform has been developed at the Leibniz Centre for Agricultural Landscape Research to operate the BonaRes Repository, a research data repository supporting the German agricultural research community. The repository was originally established within the BonaRes research initiative (meaning soil as a sustainable resource) and has since become part of the national research data management initiative FairAgro, which aims to build a FAIR-compliant data infrastructure for agricultural research in Germany. The system currently hosts more than one thousand datasets and is accessible to researchers across the German agricultural research community.
One objective of this work was to investigate how a geospatial data platform can be adapted and extended to support the requirements of a domain-specific research data management infrastructure. Particular attention is given to the architectural integration of repository workflows, metadata interoperability, data validation mechanisms, and scalable deployment infrastructure.The objective of this work is to investigate how an open-source geospatial data platform can be adapted and extended to support the requirements of a domain-specific research data management infrastructure. This work should motivate others to follow our path.
A central requirement for research data infrastructures is the support of interoperable and domain-relevant metadata models. The BonaRes Repository integrates multiple metadata standards, including the DataCite Metadata Schema, together with the ARC metadata model developed within the NFDI4Plants initiative to represent annotated research context. Further requirements are addressed through the BonaRes Metadata Schema, an extended INSPIRE-oriented metadata schema for soil and agricultural research.
Supporting efficient data ingestion while ensuring high data quality remains a critical challenge for research data repositories. To address this, we developed an open-source upload tool (https://github.com/zalf-rdm/upload-tool) that interfaces with the GeoNode backend via its REST API, offering researchers a structured, guided workflow for dataset submission. The tool enables users to prepare datasets locally, validate data structures and metadata prior to upload, and publish datasets through a transparent and reproducible process. A key enhancement to the workflow is the integration of a data steward review process, which allows data stewards to inspect uploaded datasets and request revisions to metadata or data structure to improve quality, consistency, and FAIRness before final publication. This collaborative feedback loop strengthens data governance and ensures adherence to domain-specific standards. Implementing this workflow required significant extensions to the GeoNode 4.x REST API to support programmatic dataset management, extended metadata handling, and multi-step publication states, including draft, review, and public release stages. To further enhance interoperability and automation, we developed geonodectl (https://github.com/GeoNodeUserGroup-DE/geonodectl), a dedicated Python client and commandline client for the GeoNode API. Geonodectl provides a clean, structured interface for managing datasets, metadata, and publication workflows, enabling seamless integration with external tools and automated ingestion systems.
Additional extensions in GeoNode enhance support for research data workflows. Native non-geospatial dataset support—allows tabular data (e.g., crop trials, sensor records) to be managed alongside spatial datasets. These datasets are visualized as interactive tables and grouped into a Tabular Collection, a tabbed interface enabling seamless exploration of related data. The metadata model was extended for domain-specific fields, and the publication workflow now includes multi-stage curation with review and embargo. ORCID integration ensures persistent researcher attribution. To improve usability for non-GIS experts, the web UI was streamlined by hiding advanced features and enhancing visual clarity, making the platform more accessible to researchers without geospatial expertise. These improvements foster broader adoption across agricultural and environmental research.
Reliable and scalable operation of the research data repository is essential for sustained use in research environments. To meet this requirement, we adopted a Kubernetes-based deployment architecture using the geonode-k8s (https://github.com/GeoNodeUserGroup-DE/geonode-k8s) Helm chart, a production-ready, open-source deployment solution developed and maintained by the GeoNodeUserGroup-DE. This Helm chart provides a fully automated, declarative, and reproducible way to deploy GeoNode on Kubernetes clusters. By leveraging Kubernetes, the infrastructure can dynamically scale based on user demand, ensuring consistent performance during peak usage, shown in FoSS4g 2024 (Performance Benchmarking for Resource Allocation Optimization in GeoNode Ecosystems on Kubernetes Clouds). This approach not only enhances system reliability and maintainability but also enables deployment across diverse environments—from local development to cloud-based research infrastructures.
The presented work demonstrates that GeoNode can serve as a foundation for domain-specific research data management infrastructures when complemented by targeted architectural extensions. By combining repository workflows with interactive geospatial exploration capabilities, the platform enables researchers to both publish and explore datasets within a unified environment.This approach supports improved dataset discoverability and practical data reuse, while the architectural patterns and open-source components developed in this project provide a reference for other institutions seeking to implement interoperable research data infrastructures based on free and open-source geospatial technologies.

Introduction

Urban platform capitalism relies on spatial databases such as OpenStreetMap (OSM) to connect people with nearby services and commodities (Michel and Schröder-Bergen 2022; Alvarez Leon 2024, p. 139). The set of features in OSM thus influences the set of locations that users of these platforms can know about and interact with (Graham and Dittus 2022, p. 17-18). Given the power of OSM to both represent and shape places, there is a need for deep place-based studies of OSM evolution, completeness, quality, and utility, with a focus on social and economic dynamics behind the map’s production (Schröder-Bergen et al. 2025). Of particular interest are those who add high-value information from local knowledge, as well as remote contributors and corporate mappers who standardize data tags and topology for smoother integration into platforms.

Research purpose and questions

This research offers an example of a deeply place-based and contributor-focused study of OSM by presenting “ the story of OSM in a small town”. I reconstruct the history of OSM’s development in Othello, Washington, USA (population 9001) using qualitative and quantitative data from the OSM history extracts, changeset comments, user profiles, and a walking survey of the town’s main thoroughfares.

Othello’s rural economy is supported principally by farming and frozen food processing. Latino immigrants and their descendants play a major role in the town’s daily life: over 79% of residents identify as Hispanic or Latino, compared with 15% for Washington state. Approximately 24% of Othello’s residents are considered by the US Census Bureau as living in poverty, compared with about 10% for Washington state (https://www.census.gov/quickfacts/fact/table/othellocitywashington/PST045224).

In this study, I focus on the influence of local mappers, one-time contributors to the project, corporate mappers, and armchair hobbyists, including the kinds of features contributed by each. In particular, I ask:

When did the map begin to be infused with local knowledge that would be difficult to trace or import from remote locations?

What can we determine about the contributors who added this local knowledge, especially their region of residence and level of involvement in the project?

What kinds of activity by corporate editors can be detected in this place?

When we walk the streets of Othello, how much information matches what we see in OSM? What kinds of things are present or missing?

This study demonstrates how FOSS4G attendees and other researchers could combine OSM history files, public profiles, and on-the-ground surveys to learn more about OSM data quality and completeness, especially in rural or low-income regions.

Methods

I gathered the edit history of OSM in Othello by downloading the OSM full history extract from Geofabrik for the US West region (download.geofabrik.de), along with the full changeset history from planet.osm.org. I used the osmconvert utility to help narrow down the edits to only those within the Othello city limits. I used Python text parsing functions for cleaning, reading, and organizing the data. I studied all unredacted edit information for nodes and ways from the year 2007 until 31 January 2026.

For each of the 383 changesets in Othello, I examined the extent and type of the edits, along with any comments or hashtags left by the contributor and discussion from the user community. I browsed the contributor’s public OSM profile page for any disclosure of home location, editing interests, or corporate affiliations. I also used Pascal Neis’ utilities How Did You Contribute to OSM and Your OSM Heat Map to understand the contributor’s activity levels and preferred editing locations

I evaluated the current state of the map in Othello by downloading all OSM data for the town on 31 January 2026 using the QuickOSM utility in QGIS. I compared the content of those datasets with information I gathered from a ground survey of Othello that same day. I walked 6.8 kilometers of street frontage of two commercial thoroughfares, recording all visible institutions such as businesses, places of worship, government offices, and nonprofit organizations.

Preliminary Results

OSM in Othello has been built by a diverse set of 122 contributors. Thirty-six of those edited the town on more than one day, and only two of them edited more than 10 days. Many of the most active contributors have at least regional ties, with either a home location or substantial mapping history within Washington state. Some of the mappers with the most detailed local knowledge edited on only one day. The amount of mapping activity in Othello has varied greatly from year to year, but was highest in 2025.

I found corporate mapping activity by Mapbox, Meta, Microsoft, and Telenav. These companies focused on improving the positioning and topological integrity of the road network, with the latter two primarily editing driveways and parking lots. Several other companies appeared to be engaged in brand promotion by adding metadata tags to businesses.

Out of 154 institutions identified in the walking survey, 29 were in OSM (18.8%). Half of the 26 national chain businesses in the Othello survey were represented in OSM, a substantially higher representation than other institutions. Institutions with Spanish-language words or names in the title, or those selling products aimed at the Hispanic/Latino population, had 13.3% representation in OSM (4 out of 30).

Othello has a largely complete and accurate street network in OSM, and extensive coverage of building footprints that make the map look detailed; however, many of the institutions that contribute to day-to-day livelihoods in the town are absent from OSM and its representation of place.

References

Alvarez Leon, L. (2024). The Map in the Machine. University of California Press.

Graham, M., & Dittus, M. (2022). Geographies of Digital Exclusion. Pluto Press.

Michel, B., & Schröder-Bergen, S. (2022). The Politics of Geodata in Urban Platform Capitalism. In A. Strüver & S. Bauriedl (Eds.), Platformisation of Urban Life: Towards a Technocapitalist Transformation of European Cities. transcript Verlag.

Schröder-Bergen, S., Michel, B., Glasze, G., & Dammann, F. (2025). Open Geospatial Data within Digital Capitalism: OpenStreetMap and the Overture Maps Foundation. Cartographica, 60(3), 160–172. https://doi.org/10.3138/cart-2024-0030

geospatial-audio-js is an open-source JavaScript library that adds real-time 3D spatial audio to web maps. Sound sources placed at geographic coordinates respond to camera movement, opening new possibilities for accessibility, immersive audio guides, and soundscape data visualization.

This presentation introduces GeoServer’s authentication and authorization subsystems, covering supported protocols, identity providers, and integration strategies. It explores combining mechanisms into a unified framework, custom plugins, and proxy-based solutions. It concludes with GeoFence, highlighting advanced rule-based access control, fine-grained data security, and flexible integration options.

This presentation introduces a National Map Agent that integrates open geospatial standards, knowledge graphs, and GraphRAG to enable intelligent, standards-aware mapping workflows. Built on open-source tools, the system transforms authoritative mapping specifications into machine-readable knowledge, supporting automated feature modeling, validation, and collaboration between mapping agencies.

The IFRC launched the Global Crisis Data Bank (Montandon) — the world's largest repository of natural hazard and impact data, enabling evidence-based decisions for financial and operational crisis planning. This talk covers how we're building a harmonized data repository using open standards like STAC for fast, reproducible analysis.

I needed to analyze pedestrian flow patterns and draw them on map in urban spaces, so I built a simple pipeline to extract trajectories from smartphone videos which shot from a low line of sight.

Poster and Demonstration Presentations

Japan’s cadastral land-parcel open data ships as massive proprietary XML files. This talk is about how I made a Rust port of the official Python converter in Rust, rewriting and optimizing, cutting nationwide conversion down from hours to minutes.

To transform discrete geospatial data into an insightful heatmap, this proposal focuses on the selection of geospatial indexing methods with traversal functionality, and the propagation of data like waves through space. This approach is implemented using open source libraries and applied to two real-world phenomena in Japan.

Mapillary hosts over 2 billion street-level images as open data. Their quality and capture conditions vary. This presentation introduces filtering methods to remove low-quality or irrelevant images, enabling more effective use of Mapillary imagery.

In earth volume calculation using the spot level method, computational speed was improved by adopting image-based rasterization for the inside/outside determination of the calculation area.

Volunteers from the FOSS4G 2026 Hiroshima LOC are building a cross-platform event guide app using React Native and MapLibre. The app offers map-based navigation from Hiroshima Station to the venue, interactive indoor floor maps, and session information linked to each floor.

An interactive web application that visualizes Tokyo's last train network fading away over time in a firework-like radial display. It combines public transit open data (ODPT) from 17 rail operators to show when and where you can no longer catch the last train home.

The orientation of Christian churches has long attracted attention in liturgical studies, architectural history, and archaeoastronomy. In the Catholic tradition, churches are often associated with an eastward-facing sanctuary and a westward-facing entrance, reflecting theological symbolism linked to sunrise, resurrection, and the anticipation of Christ’s return. In actual urban settings, however, church orientation is rarely determined by theology alone. Existing streets, neighbouring buildings, plot geometry, topography, and later rebuilding campaigns can all shape the final disposition of a church building. This tension between sacred orientation and urban form makes church directionality a productive field of inquiry at the intersection of religion, architecture, and urban studies.

This proposal investigates church orientation patterns in Milan, Italy, using open geospatial data derived primarily from OpenStreetMap (OSM). Milan is an appropriate case because it is a historic Catholic city with a long continuity of ecclesiastical development and a complex urban morphology shaped by Roman, medieval, early modern, and modern transformations. Rather than treating churches only as isolated monuments, this study approaches them as urban objects embedded in streets, blocks, and neighbourhood structures. It asks three questions: first, whether church entrances in Milan display a statistically visible directional pattern; second, whether that pattern suggests the persistence of the traditional east–west liturgical axis; and third, how directional variation can be interpreted in relation to urban morphology and spatial constraints.

This proposal also responds to a methodological opportunity. OSM has become an important infrastructure for transportation analysis, humanitarian mapping, land use studies, and urban modelling, yet its value for architectural-historical and religious-spatial research remains underexplored. By using OSM building footprints as the spatial basis for city-scale analysis, this proposal contributes to an emerging dialogue between open geospatial science, digital humanities, urban history, and the study of religion.

Previous scholarship has proposed several explanations for church orientation. One long-standing interpretation emphasizes alignment toward geographic East as a symbolic and liturgical norm. A second associates church alignment with the sunrise azimuth on the feast day of the patron saint. A third suggests that deviations from true East may reflect the historical use of magnetic compasses during church construction. In this debate, Arneitz et al. (2014) provide an important reassessment based on a statistical analysis of medieval churches in Lower Austria and northern Germany. Their study compares deviations from geographic East, magnetic East, and sunrise azimuths associated with patrons’ feast days. The smallest mean deviations were found relative to geographic East: −5.5° in Lower Austria and −2.1° in northern Germany. By contrast, deviations from magnetic East were much larger, at −19.0° and −14.0°, leading the authors to reject the compass hypothesis statistically. They also found that the patron-saint sunrise model showed greater scatter and only limited explanatory value overall.

Equally important, Arneitz et al. argue that deviations from East should not automatically be treated as evidence of alternative symbolic rules. They emphasize that church orientation may be affected by neighbouring buildings, pre-existing streets, foundation conditions, vegetation, and topography, especially in urban settings. They also note that a raised horizon can substantially shift the apparent sunrise, complicating straightforward solar interpretations. Church orientation should therefore be understood as the result of both sacred principles and material-spatial constraints. This framework is particularly useful for interpreting Milan, where churches are situated within a dense and historically layered urban fabric.

The dataset used in this proposal consists of 114 church buildings in Milan. For each case, a web-based mapping tool was developed using MapLibre GL JS and Turf.js to measure entrance orientation and automatically analyse church directionality. The analysis tool has been released as open-source software under the CC0-1.0 license, and the resulting measurements were organized in a spreadsheet. Direction was defined as the azimuth from the building centroid toward the main entrance. Although this value does not directly measure the liturgical axis or altar orientation, it provides a consistent and scalable proxy for analysing the public-facing directionality of church buildings in relation to the urban fabric. The azimuth values were analysed using circular statistics and visualized through a wind rose diagram. For exploratory analysis, the bearings were grouped into 16 directional sectors to evaluate concentration and dispersion.

The results show that church entrances in Milan are neither uniformly distributed nor tightly concentrated around a single bearing. Instead, the pattern is dispersed but structured. The wind rose reveals a noticeable concentration in the western to west-southwestern sectors. The two most frequent classes are W and WSW, each with 19 cases, representing 16.7% of the sample. Together, they account for one-third of all observed entrance directions. Secondary peaks appear in the E sector, with 12 cases (10.5%), and in the SW sector, with 11 cases (9.6%). Circular statistics yield a mean direction of 245.4° and a mean resultant length of 0.283, indicating a relatively weak but visible directional tendency rather than a strongly concentrated system.

These findings suggest that the traditional Catholic arrangement of west-facing entrances and east-facing sanctuaries may still be partially visible in Milan at the aggregate level. At the same time, the relatively low concentration and the presence of substantial eastern and southwestern groups indicate that the city cannot be explained by a single orientation rule. Instead, Milan appears to contain a mixed spatial morphology of church orientation. Some churches likely preserve the canonical east–west liturgical axis, while others appear to have adapted to street alignment, constrained parcels, public squares, topographic conditions, or later phases of rebuilding and urban redevelopment.

This interpretation aligns closely with the implications of Arneitz et al. (2014). Their work suggests that the most productive way to analyse church orientation is not to choose between symbolic and practical explanations, but to examine how these forces interact. In Milan, church entrances should therefore be read not only as liturgical markers but also as indicators of how sacred buildings negotiate the urban environment. Entrance direction becomes a meaningful variable through which one can explore the relationship between ecclesiastical architecture and surrounding city form.

Methodologically, this proposal demonstrates the value of combining OSM building geometries, a custom open-source web mapping tool, manually validated directional attributes, and circular analytical techniques in an open and reproducible workflow. The approach is lightweight and transferable to other cities where church inventories and building footprints are available. It also opens possibilities for comparative research across Catholic and non-Catholic cities, as well as diachronic analyses incorporating construction dates, denominational affiliation, or street-network orientation.

The contribution of this proposal is therefore twofold. Substantively, it offers new evidence that church orientation in Milan reflects both sacred tradition and urban form. Methodologically, it shows that open geospatial data, OSM-based analysis, and openly released analytical tools can extend church orientation research beyond isolated monuments toward city-scale spatial humanities. In the context of FOSS4G and ISPRS-related academic discussion, the study demonstrates how open mapping ecosystems can support new forms of interdisciplinary scholarship across religion, architecture, urban history, and geospatial science.

This talk explores how cloud-native principles- combined with STAC, COGs, and TiTiler - enable dynamic, scalable, expression-driven raster workflows without precomputing products.

"Cognitive Motion" leverages Motion UI and perceptive psychology to simplify complex Smart Agriculture data. By using multiscale continuity, temporal animations, and real-time visual feedback, it reduces cognitive load and prevents spatial disorientation. Purposeful design aligns data visualization with human instinct, fostering trust and enabling more intuitive, precise decision-making.

Decision-making depends on trust, but how often can you trace a reported figure back to its source? We're solving that with an open-source platform to repeatedly analyse any dataset against any geometry cloud-natively, with an auditable record of every step and full provenance for each value.

How can open-source GIS support large-scale and collaborative validation of urban water networks? This talk presents a scalable workflow built with QGIS, Giswater, PostGIS and QFieldCloud, enabling structured field data collection, multi-organization collaboration and high-quality datasets for hydraulic modelling in a national water utility.

This study applies Sentinel-2 satellite imagery and open-source Python tools
to detect and map quarry pond remnants on Kitagi Island, a designated heritage
site in Japan's Seto Inland Sea, revealing spatial distributions consistent
with the island's historical quarrying records.

Ensuring reliable map tiles is essential for modern web mapping applications. This talk explores how quality is maintained around MapLibre through automated testing, visual regression testing, and continuous integration workflows. We then demonstrate how TileGuard, an independent vector tile inspection tool developed during this work, complements these automated processes by helping developers investigate rendering regressions, inspect vector tile data, and diagnose quality issues more efficiently. The session concludes with practical engineering patterns that other open-source geospatial projects can adopt to build more reliable tile quality workflows.

Global demand for GIS is rising, but 3D city model adoption takes different forms across countries. This talk explores Japan's Project PLATEAU in its mature phase and shares field experiences of introducing similar initiatives abroad, highlighting Ukraine's reconstruction-driven adoption alongside cases in Thailand and Peru.

An open-source geospatial system in Slovenia provides a unified accessibility database and public web viewer for inclusive mobility. It helps municipalities and other decision-makers identify barriers, prioritize interventions, and plan their gradual removal for vulnerable groups.

From GeoNetwork semantic search to agentic GIS automation and the French National Digital Twin, Camptocamp shares two years of hands-on open source GeoLLM experimentation — what works, what doesn't, and what the OSGeo community should build next.

This paper presents the current state of development of a spatial data infrastructure, based on GeoNode, focusing on data applied to natural disasters, especially geological hazards and susceptibility to mass movements and flooding.

The rapid growth of Earth observation (EO) data, from satellite missions such as the Copernicus Sentinel missions to Digital Twin Earth frameworks like Destination Earth, which integrate high-resolution atmospheric and oceanic simulation models with the ever-growing satellite remote sensing datasets, poses significant challenges for data storage, interoperability, and analysis at the global scale. Traditional gridding approaches, typically based on regular latitude/longitude grids or projected coordinate systems such as UTM, suffer from well-known limitations, in particular non-uniform cell areas, singularities at the poles, and poor scalability for multi-resolution analysis. There is a growing need for a common, efficient, and interoperable data representation that can accommodate heterogeneous sources, support large-scale analytics, and integrate naturally with cloud-native storage paradigms.

The Grid4Earth project investigates how HEALPix (Hierarchical Equal Area isoLatitude Pixelization), combined with the Zarr storage format, can serve as a unifying framework for Earth observation and Digital Twin data. Although HEALPix takes a fundamentally different route to spherical tessellation than the polyhedron-based constructions that dominate the classical DGGS literature, it satisfies the desirable DGGS geospatial core criteria: congruent refinement levels, strictly equal-area cells, and unique hierarchical indexing that enables efficient spatial queries and data indexing. Originating from the demanding computational needs of cosmic microwave background analysis in astrophysics, HEALPix is now well integrated into the workflows of climate modellers and astrophysicists and is thus positioned as a prime candidate for terrestrial geospatial and remote sensing applications. Our approach extends HEALPix for geospatial applications by generalising the spherical tessellation to an ellipsoidal HEALPix, ensuring true equal-area properties on the WGS84 ellipsoid. Unlike traditional geographic grids (e.g. Lat/Lon grids, UTM), HEALPix cells have identical areas regardless of latitude, thus eliminating polar distortions and facilitating statistically unbiased global analyses. Combined with the cloud-native, chunked array data format Zarr, HEALPix data can be efficiently stored, accessed in parallel, and integrated into modern Python data science workflows.

Within grid4earth, we developed a complementary suite of four composable open-source Python packages that address the full workflow from data ingestion to analysis: healpix-geo, healpix-resample, healpix-plot, and healpix-analyse.

healpix-geo: Provides the foundational geospatial layer of the grid4earth ecosystem. While standard HEALPix implementations assume a perfect sphere, healpix-geo extends the tessellation to the WGS84 ellipsoid, ensuring true equal-area properties for real-world Earth observation data. The library supports refinement levels from 0 to 29, spanning global scale down to ~1.2 cm resolution, and provides coordinate conversions, spatial subset selection through bounding box, polygon, and cone coverage queries, and Multi-Order Coverage (MOC) via the zuniq scheme. All operations are fully vectorised and support parallel processing, enabling memory-efficient workflows at scale.

healpix-resample: Tackles the core data conversion problem: regridding arbitrary Lat/Lon or UTM gridded data onto the ellipsoidal HEALPix grid. It implements a forward modelling approach in which source data points are projected onto their corresponding HEALPix target pixels using precomputed index mappings. This design enables efficient batch processing of large datasets, including multi-temporal satellite image stacks and model outputs. Flexible aggregation strategies (nearest neighbour, averaging, majority/mode) make it applicable to both continuous fields (e.g., surface temperature, reflectance) and categorical data (e.g., land cover).

healpix-plot: Provides visualisation capabilities tailored to HEALPix data. Building on established Python visualisation libraries, it offers functions for projecting HEALPix maps onto standard cartographic projections (Mollweide, orthographic, and regional), overlaying ancillary geographic information, and producing publication-quality figures. Particular attention has been paid to handling multi-resolution data and supporting interactive exploration in Jupyter notebook environments.

healpix-analyse: Completes the ecosystem with diagnostic and analytical tools, including, but not limited to computation of power spectra via spherical harmonic decomposition, spatial convolution with custom kernels on the sphere, and standard statistical diagnostics (mean, variance, histograms, spatial correlation). These operations benefit from the isolatitude property of HEALPix and decades of optimisation in astrophysical data processing, which makes spherical harmonic transforms in HEALPix exact and fast. These capabilities are valuable for model evaluation, intercomparison between EO products and Digital Twin outputs, and characterisation of spatial variability at global scale.
The HEALPix+Zarr approach builds on existing adoption. HEALPix is already being used in the Destination Earth's Climate Digital Twin, and Zarr underpins the EOPF format for Sentinel data re-engineering, thus providing a validated foundation for the Grid4Earth toolchain.

All four packages are released as open-source software under permissive licences and are publicly available on GitHub, PyPI, and conda-forge. The project follows open development practices, with documented APIs, unit tests, and example notebooks. Contributions from the community are actively encouraged, and the packages are designed to integrate seamlessly with the broader Python geospatial ecosystem (Xarray, NumPy, Matplotlib, and PyProj). In parallel, the Grid4Earth team is working together with the OGC DGGS Standards Working Group to advance recognition of ellipsoidal HEALPix as a conformant DGGS, and with the CF Conventions community to establish standardised metadata conventions for HEALPix-indexed data in netCDF and Zarr, ensuring that the toolchain aligns with evolving community standards for interoperability.

In conclusion, the Grid4Earth Python suite provides a complete, open-source toolchain for converting, storing, visualising, and analysing Earth observation and Digital Twin data in the ellipsoidal HEALPix+Zarr format. By combining geodetically accurate indexing, efficient regridding, rich visualisation, and spherical analytics, the ecosystem lowers the barrier to adopting HEALPix as a common representation for global-scale geospatial workflows. We believe that this approach has strong potential to improve interoperability across the Earth system science community.

This talk presents MapConductor, a neutral open-source layer designed to improve interoperability among mobile map SDKs. Rather than replacing existing providers, it bridges ecosystems, supports shared conceptual models, and promotes flexible integration across diverse geospatial platforms.

Open-source software was used to develop water flow and tracing models of Te Awarua o Porirua (Porirua Harbour), Aotearoa-New Zealand. To facilitate model outputs’ interpretation by the local community, an opensource dashboard was developed together with 3D visualisations of the data. All developed products remained with the community.

Introducing the Climate Action Navigator, a FOSS4G website containing high-resolution spatial indicators of actionable insights to support climate action.

We present a shift from tool-centric to intent-centric wildfire monitoring. By integrating LLMs and VLMs, we eliminate the technical barrier of manual tool selection. This enables any user to utilize satellite data through simple dialogue, achieving accessibility for everyone and bridging technology and humanity.

verture Maps Places offers 53 million POIs with a clean taxonomy — but what can you actually do with them? In 50 lines of Python and DuckDB, I turn raw Overture places into a visual neighborhood typology. Before/after, pipeline, and a ready-to-run notebook.

Cloud Native Geospatial formats like FlatGeobuf and GeoParquet allow applications to access geospatial data directly from an object store with no API server maintenance. This session will explain how the Geoconnex project has leveraged such formats to improve the user experience of analyzing hydrological information across the United States.

GeoParquet is a file format based on Apache Parquet for efficient storage of large geospatial datasets. Well-sorted Parquet achieves better compression and better query performance. However, it is not enough for streaming features with acceptable performance. To address this, I added hierarchical dimension for sorting and achieved better results.

Maps often rely on color to communicate spatial information, yet about 3% of the global population has color vision deficiency, making interpretation difficult. This study proposes accessible map design approaches that reduce reliance on color using inclusive palettes, symbols, patterns, and visual hierarchy, resulting in practical guidelines for GIS developers.

Web map engines perform dozens of optimizations to render geographic features at interactive frame rates. This talk reveals the internal mechanisms – from GPU optimization that collapse 10,000 draw calls into a handful, to zero-copy parsing pipelines and Web Worker task scheduling that keep the main thread free.

Iwfgara is an open-source web-based 3D GIS platform for visualizing numerical weather prediction data and enabling interactive exploration of atmospheric phenomena.

This talk introduces Kartore, a project centered around a style editor for MapLibre. It covers the motivation and design approach, as well as current editing capabilities, style management approaches, and planned extensions such as sprite and glyph generation, exploring possibilities for improving cartographic workflows.

GeoAI embeddings span models, representations, and systems, but are often conflated. This talk introduces a clear framework and compares approaches like Clay and TESSERA to help practitioners understand and apply geospatial embeddings in real-world workflows.

This presentation introduces the public release of vector tile data for the Actual Vegetation Map 2024 and the launch of the Satellite-based Vegetation map 2030. It outlines technical design, open data distribution, and interoperability with open-source GIS to support biodiversity and environmental applications.

Impact-based flood forecasting asks not "how intense?" but "who gets hit, and how badly?"

This talk documents a open-source pipeline — Python, PostgreSQL, and QGIS — that translates rainfall forecasts and hazard layers into impact assessments for critical facilities, with honest discussion of what works, what doesn't, and why.

Biological invasions represent one of the most significant threats to global biodiversity and agricultural systems, causing substantial ecological and economic damage worldwide. Among emerging invasive pests in East Asia, Aromia bungii, commonly known as the red-necked longhorn beetle, has become a serious threat in Japan. The species attacks several Prunus species, including ornamental cherry trees (Cerasus spp.), peach (Prunus persica), and plum (Prunus salicina). Because cherry trees play an important ecological and cultural role in Japan, the spread of this invasive beetle has raised growing concerns for landscape management and biodiversity conservation.
Since its first detection in Aichi Prefecture in 2012, A. bungii has expanded rapidly across urban and peri-urban areas. Understanding its spatial pattern is therefore essential for effective monitoring and early intervention. Spatial analysis can be applied to these processes. However, such analyses fundamentally depend on how spatial data are defined, including the geometry of the spatial grid, which can influence the results.
Thus, to examine how grid shape influences spatial analysis results, this study evaluates spatial autocorrelation measures using different tessellations of invasive species occurrence data. Specifically, we compared rectangular and hexagonal grids for analysing spatial patterns in A. bungii occurrence records and density of rivers in Saitama Prefecture, Japan.

Volunteer-based occurrence data for A. bungii were compiled from field surveys across Saitama Prefecture from 2017 to 2024, yielding 2,349 confirmed presence records. Records were classified as confirmed presences if either adult beetle observations (Adult-yes = 1) or evidence of tree damage (Tree_damage = 1) was recorded. All records were georeferenced using latitude–longitude coordinates (WGS84) and then reprojected to UTM Zone 54N (EPSG:32654) to ensure metric accuracy in spatial calculations. We assembled the occurrence data into predefined grid cells to explore spatial patterns across the study area and to enable consistent spatial aggregation and neighbourhood-based analyses.

Two grid tessellation schemes were constructed over the study area. The rectangular grid consisted of 4,013 cells at a 1 km × 1 km resolution. Each rectangular cell overlaid with the occurence point of location Aromia bungii presence in Saitama. This configuration corresponds to rook contiguity, where only four directly adjacent neighbours are considered. To evaluate the effect of diagonal bias, the same rectangular grid was also analysed using queen contiguity, which includes eight neighbouring cells by incorporating both direct and diagonal neighbours. The hexagonal grid was generated using Shapely and Geopandas with hierarchical spatial indexing system, producing hexagonal cells with an almost equivalent spatial resolution to the rectangular grid. To keep the overall grid coverage comparable to the rectangular representation, 4,007 hexagonal cells were generated to cover the study area. The rectangular grid had a side length of 1.0 km, whereas the hexagonal grid had a slightly larger side length of 1.074 km. Each hexagonal cell has a circumradius of 620m and a cell area 0.999km square approximately equivalent to 1km square rectangular grid ensuring comparable spatial resolution between two tessellations. From these two tessellation we set four spatial weight configurations which are; Rook, standard Queen, Distance-Weighted Queen and Hexagonal KNN-6 and then calculated the spatial autocorrelation.

We found Global spatial autocorrelation analysis confirmed significant and positive clustering of A. bungii occurrences across all configurations Moran's I = 0.4525--0.5380, Geary's C = 0.5527--0.5970, Getis-Ord G all p-value = 0.001, demonstrating that the spatial agglomeration characteristic is robust to the choice of tessellation geometry and neighbourhood definition. Local spatial autocorrelation analyses consistently identified two primary hotspot zones at the northern (Kazo--Gyoda) and the southeastern (Soka) parts of Saitama Prefecture across all configurations, providing spatially explicit evidence of persistent infestation core that may serve as priority zones for targeted surveillance and countermeasure deployment.　Across all configurations hotspot areas were consistent, however coldspot delineation was found to be unstable with certain configurations inconsistent coldspot that undermine the accurate identification of management priority zones.

Hexagonal tessellation produced higher counts of similar cells associated with hotspot clusters compared to rectangular configurations due to six- neighbourhood structure equidistant by design, ensuring the spatial relationships are evaluated uniformly in all directions thereby reducing directional bias. In contrast, rectangular configurations evaluate neighbours at unequal distances, introducing directional bias that result in fragmented cluster boundaries and lower hotspot cell counts. The inconsistency of coldspot indicating the spatial clustering outcomes are sensitive to tessellation choice. Practitioners should therefore consider the tessellation configuration carefully when interpreting spatial cluster results for management planning, as different configurations may lead to different prioritisation of monitoring and eradication efforts, specifically at cold spot areas.

The limitation of different geometry structure should be acknowledged. Although, the comparable spatial resolution were used, the exact areal equivalence between rectangular and hexagonal tessellation could not be perfectly achieved due to their difference geometric structure.

OpenStreetMap (OSM) is a prominent platform for Volunteered Geographic Information (VGI), wherein a diverse array of geographic data, including road networks, buildings, land use, and points of interest (POI), are updated daily by the community. However, owing to the collaborative nature of data generation by numerous contributors, ensuring the quality of edits and ongoing monitoring has emerged as a significant challenge in recent years (Choe et al., 2023). The OpenStreetMap Changeset Analyzer (OSMCha) serves as a quality assurance tool for monitoring OSM changesets and automatically flags suspicious edits based on a catalogue of detection rules covering geometric and tag plausibility, edit scale, contributor behavioural patterns, and edits to specific feature types. Nonetheless, few studies have systematically collected OSMCha data via its API for large-scale, country-wide spatiotemporal analysis.

In this study, we developed a series of Python scripts to systematically collect changeset data from the OSMCha API, amassing 740,038 changesets spanning four years from 2022 to 2025 for the Japanese context. Given API constraints, such as limits on the number of records retrievable per request and potential timeouts, we implemented monthly batch processing and GeoJSON storage, a checkpoint function to allow resumption from interruptions, and a validation logic that alerts when the acquisition rate falls below 90% of the expected total, thereby ensuring the stable collection of large-scale data. The collected data were consolidated into a FlatGeobuf database in which 60 reason_id-to-name mappings—classified into four categories (geometry/tag plausibility, edit scale, contributor attributes/behaviour, and feature-specific rules)—were embedded as attributes. By spatially joining the centroid of each changeset with Japanese administrative-area polygons (with 98.8–99.2% coverage), we appended prefecture names and municipal codes. The developed scripts and the resulting dataset—comprising 339,114 detection occurrences across the four years—will be released as open-source on GitHub and Zenodo, supporting reuse for similar analyses in other countries and regions.

Our analysis corroborated a significant upward trend in OpenStreetMap (OSM) editing activities in Japan. The annual number of changesets increased by 63.3%, from 144,625 in 2022 to 236,203 in 2025. Concurrently, the number of unique contributors expanded by 72.5%, from 3,660 to 6,315. Conversely, the rate of edits identified as suspicious (is_suspect rate) decreased by 11.9 percentage points from 40.7% to 28.8%, suggesting an overall enhancement in the community's data quality. An examination of editing software revealed that while iD remained predominant, accounting for 63.1% of edits, mobile-oriented editors such as StreetComplete experienced a 4.5-fold increase over four years, with changesets rising from 5,709 to 25,576 in the same period. EveryDoor exhibited a 10.6-fold growth in the same period (from 539 to 5,692), indicating increasing adoption of field-survey-style editing. Regarding data sources, survey-based edits consistently demonstrated lower suspect rates (15.4% over the four years, declining to 10.6% in 2025) than non-survey-based edits (38.2% in aggregate, 33.4% in 2025), demonstrating clear quantitative advantages in data quality. AI-assisted edits (RapiD/mapwithai) also showed a marked improvement in suspect rates, declining from 64.0% in 2022 to 16.0% in 2025, although their annual volume remained modest at approximately 1,700–1,800 changesets—in line with the cautious adoption pattern surrounding AI-assisted mapping (Andorful et al., 2026). PLATEAU-derived edits, leveraging Japan's national 3D city-model dataset, also grew 4.3-fold over the same period (Seto et al., 2023).

A longitudinal analysis spanning four years identified a significant alteration in the detection rules of OSMCha around 2024. Detection reasons associated with geometry and tag plausibility, such as "Invalid tag modification" (reason_id=42) and "Motorway/trunk geometry modified" (reason_id=91), were recorded close to or over 10,000 times annually in 2022 and 2023, but these instances nearly vanished post-2024. Concurrently, there was a notable increase in detections based on user attributes, including "User has multiple blocks" (reason_id=83) and "suspect_word" (reason_id=1), indicating a shift in the algorithm's emphasis from geometry and tag plausibility assessments to monitoring contributor behavioural patterns. The frequency of review requests escalated from 3,151 in 2022 to 8,600 in 2025, reflecting an increase in peer-review activities within the community. In terms of user retention, 84.7% of the 14,978 contributors were active for only one year, with only 2.9% (433 contributors) maintaining activity throughout the entire four-year period.

Spatial analysis employed kernel density estimation (KDE) to visualise the geographical concentration of editing activities, utilising the centroid coordinates of all changesets. Significant concentrations were identified in three major metropolitan areas—Tokyo, Osaka, and Nagoya—replicating, at the within-country prefecture scale, the cross-national spatial bias pattern reported by Quattrone et al. (2015), whereby power users concentrate in urban centres while occasional contributors distribute their edits more uniformly. The January 2024 Noto Peninsula earthquake further triggered a marked increase in crisis mapping activities within Ishikawa Prefecture, resulting in a nationwide doubling of changesets to 30,228 for that month, with participation from 1,536 unique users in Japan. This phenomenon underscores the spatial and temporal responsiveness of VGI during disaster events. Furthermore, substantial regional disparities were observed in the suspect rate by prefecture: over half of the edits were flagged in Iwate (59.3%) and Kumamoto (55.4%) prefectures, whereas the rates were considerably lower in Kyoto (20.3%) and Oita (20.6%) prefectures.

This study makes a significant contribution by developing a method for large-scale data extraction from the OSMCha API and presenting a comprehensive spatiotemporal analytical framework encompassing 740,038 changesets over four years for the entirety of Japan. The scripts and nationwide dataset developed in this study will be released as open-source on Zenodo, thereby providing a reproducible workflow that can be utilised for similar studies in regions beyond Japan. Looking forward, we intend to pursue qualitative analysis of harmful-flagged cases, international comparative analyses using OSMCha (with Benelux as an initial target), and panel-data analysis tracking the activity trajectories of individual contributors, with the objective of accumulating knowledge that contributes to the sustainable enhancement of data quality in VGI communities.

Open-source communities can't afford to ignore the growing use of AI agents to write code. This talk shares lessons from building MapLibre Agent Skills: an open, eval-tested knowledge base that corrects AI hallucination in MapLibre implementations, with lessons for other projects looking to do the same.

Geospatial data powers critical decisions . TrustChain is an open-source implementation of the OGC IPT framework that makes trust computable: every dataset carries a cryptographic fingerprint proving its origin, integrity, and provenance. The data answers that question itself.

We share how the Open Exposure Taxonomy integrates several open data taxonomies and improves the quality of data for disaster risk.

This study develops an R package "spCF" that provides functions for coarse-to-fine spatial modeling, enabling scalable spatial prediction, regression, and multi-scale analysis for large samples.

This study proposes an automated slope estimation method using cross-sectional profiles derived from large-scale point cloud data to reduce operator dependency and improve reproducibility, and demonstrates its practical applicability for the safety inspection and maintenance of retaining walls.

The GeoNetwork opensource project is a catalog application that makes it easy to discover resources at local, regional, national or global level.

By using openEO and titiler, geospatial analysts can reduce data preparation time for Sentinel-2 datasets, enabling rapid, actionable insights for accelerating climate action. This cloud-native workflow delivers insightful GIS dashboards, empowering decision-makers with timely analytics on forest health.

This presentation introduces the "Data Transition" framework, using FOSS4G pipelines to transform satellite imagery into actionable disaster solutions. By bridging technical capability and human necessity, it provides real-time situational awareness, strategic urban planning, and predictive intelligence, empowering vulnerable communities to save lives and protect livelihoods through informed decision-making.

This study proposes a Conversational GIS system that integrates LLM with MCP, allowing non-technical users to access and analyze geospatial data through natural language, supporting flood risk analysis and lowering technical barriers for GIS interaction.

In Nagaoka, Japan, 23 citizens aged 11–70+ with no GIS experience mapped the 1945 air raid using Re:Earth, an open-source WebGIS platform. This talk explores how no-code CMS architecture enabled real-time "input-to-visualization" workflows, transforming community members into active contributors to geospatial heritage preservation.

Background

Flood inundation models are an essential component of flood hazard assessment, emergency management, infrastructure planning, and climate adaptation. Most contemporary two-dimensional flood models operate on structured raster grids or unstructured meshes and typically require repeated conversion between raster, vector, and computational representations throughout the modelling workflow. While these approaches are mature and widely adopted, they can introduce complexity in data management and interoperability. Discrete Global Grid Systems (DGGS) provide an alternative spatial framework based on hierarchical tessellations that support globally consistent indexing, multi-resolution analysis, and standardised spatial referencing. Although DGGS have been widely applied in Earth observation, geospatial analytics, and environmental data management, comparatively little work has explored their use as the primary computational mesh for hydrodynamic simulation.
This paper presents FloodA5, an open-source flood modelling framework built on the A5 equal-area pentagonal DGGS. The A5 DGGS provides an equal-area hierarchical tessellation of the Earth composed of pentagonal cells. At any given resolution, all cells possess identical area, while refinement follows a strict parent–child hierarchy in which each cell subdivides into five children. Interior cells possess a uniform five-neighbour topology, and compact hierarchical identifiers provide efficient indexing and storage. These properties make the grid attractive for hydrodynamic modelling because water storage calculations can be performed directly from cell area and depth, while the hierarchical structure provides a potential pathway towards future adaptive multi-resolution simulations.

Methods

FloodA5 was developed to investigate the feasibility of performing flood inundation modelling within a fully DGGS-native workflow, maintaining a single spatial representation from mesh generation through terrain processing, simulation, storage, and visualisation. The framework is implemented primarily in Julia, with DGGS operations provided through the pya5 ecosystem via a lightweight Python interoperability layer. FloodA5 integrates mesh generation, digital elevation model (DEM) processing, hydrodynamic simulation, sub-grid terrain representation, visualisation, and data storage within a unified software architecture.
FloodA5 currently supports two hydrodynamic formulations. The standard solver applies the inertial shallow-water approximation of Bates et al. (2010) on the A5 mesh. Because A5 cells form a non-orthogonal polygonal grid, a first-order correction based on the angle between the edge normal and the cell-centre connection vector is applied when calculating water-surface gradients. The framework also includes an optional Sub-Grid Sampling (SGS) formulation intended to represent terrain variability below the computational mesh resolution. Rather than storing a single representative elevation for each cell, the SGS approach derives hypsometric relationships from high-resolution DEM samples and pre-computes volume–elevation, wetted-area, hydraulic-radius, conveyance, and edge-sill relationships. During simulation, water storage is tracked as volume and converted to water-surface elevation through inversion of the hypsometric curves, while flow routing uses hydraulic properties derived from the pre-computed SGS tables.
The framework was evaluated using two synthetic benchmark problems and a real-world flood case study. The first benchmark consisted of a point-source injection on a flat domain. Under isotropic conditions, the resulting inundation pattern should be circular and therefore provides a simple test of directional bias. The second benchmark consisted of a planar slope intersected by a perpendicular embankment. This benchmark was designed to evaluate routing behaviour under a known flow direction and assess the ability of the SGS formulation to represent sub-cell topographic barriers. A larger-scale evaluation was performed using the January 2005 Carlisle flood event, using the same domain configuration and inflow hydrographs employed in previous LISFLOOD-FP studies. Three FloodA5 configurations were tested: a resolution 18 standard solver (approx. 22 m cell spacing), a resolution 20 standard solver (approx. 5.6 m cell spacing), and a resolution 18 SGS solver. Results were compared against a 5 m LISFLOOD-FP reference simulation using inundation extent intersection-over-union (IoU) and depth RMSE metrics.

Results

The synthetic benchmarks demonstrated that physically plausible flood propagation can be simulated on the A5 DGGS. In the point-source benchmark, the resulting inundation pattern exhibited a high Polsby–Popper circularity score of 0.96, indicating near-circular expansion. However, visual inspection revealed a preferred northwest–southeast propagation axis, suggesting the presence of directional routing bias. The planar slope benchmark provided stronger evidence of this behaviour, with the flood wave deviating approximately 30–60 degrees from the expected downslope direction. These results indicate limitations in the current treatment of non-orthogonal gradients and suggest that more sophisticated gradient reconstruction approaches may be required for accurate routing on pentagonal meshes.
The SGS benchmark revealed a second important limitation. Although the SGS formulation successfully represented sub-cell elevation variability, it failed to reproduce the hydraulic effect of an embankment located entirely within individual cells. Water was able to pass through the barrier because the current SGS representation preserves elevation distributions but not the spatial arrangement of topographic features. This finding highlights a potential limitation of storage-based sub-grid approaches when internal barriers are important controls on flow routing.
Comparison with the Carlisle reference simulation demonstrated that useful flood simulations can nevertheless be produced using the current implementation. The Resolution 18 standard solver produced the closest correspondence with the LISFLOOD-FP reference, achieving an IoU of 0.67 and a depth RMSE of 1.17 m. While these results indicate only reasonable rather than strong agreement, they demonstrate that a DGGS-native flood model can reproduce broad inundation patterns within a real-world catchment. Interestingly, neither increased resolution nor the SGS formulation improved performance, suggesting that numerical formulation errors currently dominate resolution-related effects.
Computational performance was also evaluated. A Resolution 18 mesh containing approximately 30,000 cells was generated in approximately two minutes and completed a 120-hour flood simulation in 39 minutes on a workstation-class AMD Ryzen Threadripper system. These results indicate that DGGS-native flood modelling can be performed efficiently without specialised high-performance computing infrastructure.

Conclusions

The results demonstrate the feasibility of hydrodynamic flood modelling on an equal-area pentagonal DGGS and establish FloodA5 as one of the first open-source frameworks to provide a complete DGGS-native flood-modelling workflow. At the same time, the synthetic benchmarks identify important methodological challenges, particularly in relation to non-orthogonal gradient treatment and sub-grid representation of internal topographic barriers. These findings should not be interpreted as limitations of DGGS-based modelling in general. Rather, they highlight the importance of numerical formulations specifically designed for non-orthogonal polygonal meshes. Beyond flood modelling, FloodA5 illustrates the broader potential of DGGS-native environmental simulation frameworks and provides an open-source platform for future research into multi-scale environmental modelling, integrated geospatial analysis, and environmental digital twins.

GeoGirafe is a framework-free WebGIS built on standard WebComponents. This talk presents its feature set, plugin-based architecture, community governance model, and the funding structure that keeps the project independent and sustainable.

GeoServer is a web service for publishing your geospatial data using industry standards for vector, raster and mapping, as well as batch or on-the-fly data processing.

Accurate fiber maps often rely on proprietary operator data. This talk presents an evidence-based methodology to assess fiber presence and connectability using Overturemaps, network measurements from a Brazilian city with SIMET, and routing constraints—explicitly accounting for uncertainty.

An experimental CesiumJS project that reconnects scenes from Barefoot Gen to real locations in Hiroshima, using historical maps and open geospatial data to explore spatial storytelling, cultural memory, and how narrative can be read through geographic space.

Urban traffic congestion costs Southeast Asian economies 2–5% of GDP annually, yet the analytical methods needed to disentangle its spatial and temporal drivers—exploratory spatial statistics, network topology analysis, multilevel variance decomposition—have traditionally required proprietary GIS platforms whose licensing fees are prohibitive for the developing-country cities that need them most. This extended abstract presents traffic-congestion-pipeline, an open-source Python package (v0.4.0, MIT license, pip install traffic-congestion-pipeline) that delivers a complete, reproducible workflow for large-scale spatiotemporal traffic research using exclusively FOSS4G tools. The package, its source code, and full API documentation are publicly available at https://github.com/firmanhadi21/traffic-analyses and https://firmanhadi21.github.io/traffic-analyses/.

The pipeline integrates five mature open-source components into a unified eleven-command CLI and Python API. (1) PySAL's esda and libpysal compute Global Moran's I and Local Indicators of Spatial Association (LISA) with K-nearest-neighbor spatial weights for hotspot detection. (2) PySAL's giddy module fits classic Markov and Spatial Markov transition models to quantify how congestion hotspots persist over time and whether their transitions depend on neighboring segments' states (spatial contagion). (3) OSMnx downloads street network graphs and computes betweenness centrality and graph-based capacity-drop detection, linking network topology to congestion. (4) statsmodels.mixedlm fits nested mixed-effects models—null, temporal, and full—that rigorously partition within-segment (temporal) and between-segment (spatial) variance using absolute speed (km/h) rather than the normalized jam factor, avoiding circularity from free-flow speed normalization. (5) Uber H3 hexagonal aggregation re-runs spatial autocorrelation tests at multiple resolutions (6–9) to check robustness against the modifiable areal unit problem (MAUP). Each component maps to a dedicated CLI command (traffic-pipeline multilevel, markov, speed-validation, h3-robustness, etc.), allowing researchers to execute any stage independently or chain the full workflow from automated data collection to publication-ready figures.

We apply this pipeline to over 264 million traffic observations collected at 15-minute intervals from the HERE Traffic API across three Indonesian cities over 11 months (March 2025–February 2026): Jakarta (14,549 road segments, population 10.5 million), Bandung (3,069 segments, 2.5 million), and Semarang (1,076 segments, 1.8 million). These cities span a 20× population range and feature high motorcycle mode shares, representing traffic dynamics markedly different from car-dominated Western cities.

Multilevel variance decomposition shows that 88–89% of total speed variance lies between segments (ICC), reflecting road design differences. Within segments, time-of-day explains 57–67% of speed fluctuations, while betweenness centrality adds less than 1% explanatory power beyond road type. ANOVA across four speed metrics confirms this temporal dominance is not a normalization artifact: η² = 8–10% for jam factor and speed reduction, 5–6% for absolute speed, and effectively zero for free-flow speed. Centrality correlations with absolute current speed are near zero (R² < 0.003); moderate centrality–jam factor correlations (R² = 0.06–0.14) are mediated entirely through free-flow speed. Evening peak congestion exceeds daily averages by approximately 40% across all three cities.

Global Moran's I is non-significant for all cities (p > 0.35). LISA identifies local clusters in ~10% of segments along known corridors, though none survive FDR correction. LISA Markov analysis reveals an inverse relationship between city size and hotspot persistence: Semarang retains hotspots at 18.8% probability versus Jakarta's 6.5%, yet no segment persists across all eight daily periods. Spatial Markov testing detects significant contagion in Bandung (χ² = 8.43, p = 0.004) and Semarang (χ² = 6.48, p = 0.011), but not in Jakarta, where denser network topology dissipates spillovers. H3 robustness analysis confirms null spatial autocorrelation at neighbourhood scales for Bandung and Semarang, while revealing a weak signal for Jakarta at resolution 8 (I = 0.030, p = 0.033).

This work is relevant to the FOSS4G community for four reasons. First, it demonstrates that PySAL, OSMnx, statsmodels, and H3—all open-source—can handle a dataset of 264 million observations end-to-end in approximately 15 minutes on consumer hardware (Mac Mini M2 Pro), establishing that FOSS4G tools are ready for operational, city-scale traffic analysis, not just prototyping. Second, the pipeline is released as a versioned, installable PyPI package with eleven CLI commands and comprehensive documentation, lowering the barrier for adoption by transportation agencies in resource-constrained settings where proprietary licenses are unaffordable. Third, the modular, command-per-analysis architecture provides a reusable template for applying FOSS4G tools to other urban analytics domains—air quality, land use change, public health surveillance—extending the impact beyond traffic. Fourth, the substantive finding—that congestion is primarily a demand synchronization problem rather than a spatial infrastructure constraint—carries direct policy implications for cities investing in road expansion versus demand management, and was only discoverable through the multi-method integration that Python's composable ecosystem uniquely enables.

The convergence of four independent spatial tests on null results for absolute speed, validated through MAUP-robust H3 aggregation, provides a level of methodological rigor that strengthens confidence in this conclusion. We believe this combination of open-source tooling, reproducible packaging, large-scale empirical validation, and policy-relevant findings makes a strong case for presentation at the FOSS4G Academic Track, and we look forward to engaging with the community on extending the pipeline to cities beyond Indonesia.

Keywords: open-source, reproducibility, PySAL, OSMnx, traffic congestion, FOSS4G

Technology can scale solutions, but the impact on the ground happens through human action. Drawing on HOT’s experiences from around the world, this talk shows how FOSS tools when paired with collaboration and local engagement empower communities to become local changemakers.

GeoTools is an open source Java library that provides tools for geospatial data.
Attend for a quick check in with the GeoTools as a vibrant OSGeo Project.

Introducing ImageN as amazing raster image procession engine for the Java GeoSpatial Community.

Coffee Break sponsored by MIERUNE Inc. & Cesium

How can companies engage with QGIS continuously while balancing community relationships and user support? Based on MIERUNE’s experience in Japan, this presentation shares practical insights into how companies can support and extend QGIS through user support, knowledge sharing, plugin development, and product development that expands its practical use.

Every serious MapLibre project hits this wall: a monolithic style.json nobody fully owns. Designers want a GUI, developers edit JSON by hand, version history is one huge diff. Mapstrata is a visual editor that makes collaboration possible: spec validation, support for related stylesheets with different themes, and a project format built for git.

We use LLM to extract and structure geospatial data buried in 100K–1M+ PDF and Office files held by Japan's MLIT, enabling visualization, spatial analysis, and evidence-based policymaking — demonstrated through real-world use cases, no coding required.

This talk presents the implementation of Object-Selective Post Effects (Outline/Bloom) in a GIS-oriented rendering system. It explains how selective effects were achieved through object identification, mask generation, and render pass composition, while addressing GIS-specific constraints such as rendering order and mesh structures.

This presentation introduces an efficient algorithm using "Extended Spatial ID" and "V-Bit" encoding to solve computational explosions in 3D overlap detection. By mapping 3D geometries to 1D keys, our method enables fast, scalable spatial set operations. This approach will play a key role in drone 3D airspace management.

High-resolution flight trajectory data are essential for analyzing airport operations, environmental impacts, and aviation safety, yet publicly available aviation statistics typically provide only aggregated indicators such as annual movement counts and hourly traffic summaries. Such datasets do not preserve the individual trajectories required for spatial analysis of airport-adjacent phenomena at fine spatial and temporal scales. As a result, researchers and practitioners often lack accessible and reproducible methods for constructing trajectory datasets suitable for geospatial analysis.

This study presents a reproducible open workflow that converts raw ADS-B (Automatic Dependent Surveillance-Broadcast) reception data into analysis-ready flight trajectory datasets while explicitly examining uncertainties introduced throughout the processing pipeline. Implemented with low-cost open reception infrastructure and open geospatial tools, the workflow is designed to be transparent, transferable, and applicable across airports and research contexts.

The workflow is demonstrated using data collected in the vicinity of Narita International Airport, one of the largest airports in Japan and a complex operational environment well suited to evaluating trajectory reconstruction methods. ADS-B signals broadcast by aircraft were received using a reproducible hardware configuration consisting of a 1090 MHz antenna, an RTL-SDR (Radio-Television Tuner Software Defined Radio) receiver, and a Raspberry Pi 4B single-board computer. Signals were decoded with the open-source software dump1090-fa and stored as records containing timestamps, aircraft identifiers, positions, altitudes, and speed information.

The empirical dataset covers a full year, from 1 April 2024 to 31 March 2025, enabling analysis across seasonal and operational variation. Initial processing stages harmonize timestamps, remove records without valid positional information, and spatially filter the dataset to the area surrounding Narita Airport. Coordinates are transformed into a projected system suitable for geometric operations, and altitude values are converted to metric units to ensure consistency across analytical steps.

Because ADS-B altitude values represent pressure altitude rather than true altitude above mean sea level, meteorological observations from the Japan Meteorological Agency are incorporated to estimate corrected altitude values using the ICAO standard atmosphere relationship. This correction improves the interpretability of vertical trajectory profiles and is particularly relevant for analyses that depend on accurate vertical geometry, including approach-path diagnostics and environmental modeling.

Individual observations are then segmented into candidate flight trajectories based on temporal continuity between successive aircraft messages. Arrival and departure movements are inferred from geometric relationships between reconstructed trajectories and runway-end locations together with vertical trajectory characteristics. Low-altitude trajectory points are associated with specific runways using a minimum-distance rule applied to runway centerline extensions. Each movement is further categorized into operational classes defined by runway usage, movement direction, and arrival or departure status.

To enrich analytical interpretation, aircraft attributes are integrated by linking Mode S identifiers with the OpenSky Network aircraft database, enabling the inclusion of information such as aircraft type, manufacturer, registration, and nationality. Additional quality-control procedures remove trajectories with large deviations from runway geometry, exclude low-altitude outliers, and filter tracks with insufficient observation density.

Applying the workflow to the one-year dataset produced 245,205 reconstructed arrival and departure trajectories associated with Narita Airport. This figure is broadly consistent with published annual aircraft movement statistics, suggesting that the reception and processing pipeline captures real operations at a practically meaningful scale. The most frequently observed aircraft types were A320, B767-300, B737-800, B777-200LR, and B787-8, reflecting the typical fleet composition operating at the airport.

Temporal analysis of the reconstructed dataset indicates approximately 650 aircraft movements per day on average, with activity concentrated primarily between 06:00 and 22:00. Late-night operations are comparatively limited, a pattern relevant for interpreting time-dependent operational and environmental impacts. The classification results also reveal a clear asymmetry in runway usage, with departures predominantly associated with Runway A and arrivals with Runway B, demonstrating how reconstructed trajectories can reveal detailed operational characteristics of airport traffic.

Data-quality evaluation shows that the proportion of records excluded during filtering and outlier removal remained below one percent of the total dataset, suggesting limited distortion of population-level representativeness. Examination of vertical profiles indicates that many arrival trajectories follow the expected approximate three-degree descent path used in instrument approaches. Nevertheless, some trajectory segments fall below this reference path, with residual discrepancies on the order of approximately 100 m. These deviations likely reflect the combined effects of ADS-B measurement uncertainty, reception gaps, and the spatial limitations of meteorological pressure observations used in altitude correction.

The principal contribution of this study is the specification of a transparent and transferable geospatial data-engineering workflow that converts raw ADS-B reception data into curated flight trajectory datasets suitable for spatial analysis. By explicitly documenting processing steps and identifying sources of uncertainty throughout the pipeline, the workflow provides a reproducible foundation for aviation-related geospatial research across different airports and operational contexts. Potential applications include airport operational analysis, time-of-day airspace utilization studies, emissions-inventory preparation, environmental impact assessment, and the creation of public information products describing aircraft activity. More broadly, the study demonstrates how open reception infrastructure and open analytical tools can be integrated into reusable spatial data pipelines, thereby contributing to transparent and reproducible geospatial practice within the FOSS4G community.

High School Student Poster Presentation

Understanding how landscapes evolve across decades to millennia is fundamental for environmental planning, hazard mitigation, infrastructure management, and heritage protection. Many environmental processes affecting land stability operate at timescales far exceeding typical planning horizons. Hillslope diffusion, fluvial incision, sediment transport, and gradual terrain adjustment continuously reshape catchments and river networks, influencing erosion risk, water quality, and the preservation of archaeological and culturally significant features. However, contemporary geospatial workflows are largely observational. Geographic Information Systems (GIS) and remote sensing platforms excel at representing present conditions and monitoring short-term change, yet provide limited capacity to evaluate how landscapes may respond to long-term environmental or human disturbances. Consequently, planners and land managers must often make decisions without accessible tools for understanding long-term terrain response.

Numerical landscape evolution models (LEMs) offer a framework for investigating long-term terrain change. A LEM is a computational model that represents a landscape as a digital surface and simulates how its elevation evolves over time according to governing physical relationships between topography and environmental forcing. Rather than analysing a single snapshot of terrain, the model iteratively updates the land surface across many time steps, allowing users to explore how landscapes may respond to natural variability or human intervention. Unlike conventional spatial analysis, which evaluates patterns at a fixed moment, LEMs simulate dynamic landscape behaviour and enable both reconstruction of past terrain conditions and exploration of potential future environmental states. For geomorphologists, they function as a virtual laboratory for examining Earth-surface change over timescales that cannot be directly observed.

Despite their scientific maturity, LEMs remain largely absent from operational geospatial practice, and it is still unclear how they can be effectively applied in real-world decision-making contexts. To investigate this issue, we conducted a comprehensive systematic review of landscape evolution modelling studies focused on anthropogenic landforms. The review identifies a restricted scope of application. Existing implementations are concentrated predominantly within mining-related environments, particularly post-extraction rehabilitation landscapes, with comparatively little use in other human-modified terrains. While many studies simulate overall landscape evolution, they seldom examine the evolution of specific localized features within those landscapes that are directly relevant to management or planning. Furthermore, stakeholder or community participation is largely absent; modelling activities are typically undertaken as research-driven academic exercises rather than collaborative decision-support tools.

Following the review, we suggest that the principal barrier preventing broader adoption of landscape evolution models is not scientific validity but geospatial usability. Existing modelling frameworks provide physically robust process simulation, yet they remain inaccessible to most practitioners outside geomorphology and scientific computing. Configuration typically requires scripting knowledge, parameter editing through code, and specialised understanding of numerical modelling workflows. Consequently planners, heritage practitioners, and community stakeholders cannot meaningfully interact with models even when the questions addressed are directly relevant to land management. The challenge, therefore, lies not in modelling capability but in translating simulation into a form usable within everyday geospatial practice.

Building on this observation, we propose an approach centred on intuitive landscape evolution modelling. Rather than positioning numerical simulation as a specialised scientific activity, the study extends landscape evolution modelling into an accessible interactive environment that communities can freely use and explore. The modelling backend is implemented using Landlab, an open source Python library that provides modular components for representing Earth surface processes, including hydrological routing, fluvial incision, sediment transport, and hillslope diffusion. This modelling engine is connected to a graphical user interface (GUI) developed using Qt for Python, allowing model configuration and execution to occur through direct visual interaction rather than scripting. Through the interface, users can define the area to be simulated, specify simulation duration and timestep structure, and choose which geomorphic processes to include, all without modifying code. In this way, process-based geomorphic modelling is translated into a form that can be understood and used without scientific computing expertise.

Simulation outputs are presented through interactive two dimensional and three dimensional terrain visualisations, elevation change maps, and comparisons between initial and simulated topography. Rather than producing static results, the system allows landscape change to be observed progressively across extended timescales. Users can explore how environmental conditions influence gradual terrain development and examine how different assumptions alter long term outcomes. By visualising processes that normally occur over centuries to millennia, the model supports interpretation of environmental dynamics that cannot be directly observed within human lifetimes and encourages discussion about potential future landscape states.

For participatory application, the framework is applied to Indigenous landscapes in Aotearoa New Zealand, where relationships with ancestral land emphasise continuity across generations. The project seeks to undertake community engagement workshops with Māori communities in which participants interact directly with simulations and collectively examine possible environmental futures. Participants explore alternative scenarios and discuss the implications of gradual environmental change for culturally significant landscape features. Through this collaborative process, modelling becomes a shared interpretive activity that supports dialogue and understanding rather than a purely technical analysis conducted by researchers alone.

This engagement reveals an additional dimension often absent from numerical geomorphology. The study highlights the importance of geoethics within landscape evolution modelling by incorporating Indigenous knowledge and stewardship perspectives into model interpretation. The simulations function as a medium for communication between scientific understanding and community knowledge, supporting culturally grounded responses to environmental change. By combining accessible modelling, transparent workflows, and participatory engagement, the work demonstrates how predictive environmental modelling can move beyond academic research and contribute to collaborative environmental stewardship and long term landscape care.

Finally, the study emphasises methodological transparency through an explicitly open and reproducible workflow. The system is implemented entirely using freely available open-source geospatial software, and the modelling code, configuration files, and demonstration datasets will be publicly released under an open-source licence. This allows independent verification of results, replication of simulations, and adaptation of the framework to different environmental and planning contexts. By removing dependence on proprietary platforms, the approach lowers practical barriers for practitioners and communities while supporting reproducible research standards. In doing so, the work positions landscape evolution modelling as a transparent and transferable analytical tool rather than a closed, specialist research product.

This talk shows how using ComfyUI and the Qwen Image model to build a "magic eraser" for high-res street imagery. We’ll dive into how to use AI-driven masks to scrub away the clutter while keeping the city’s geometry intact, plus some tips on scaling this up.

The Pasig River in Metro Manila is a major global source of ocean plastic pollution. Leveraging digital twin and AI-Machine Learning models, this project enables real-time monitoring and predictive analysis of plastic waste flows.

Voxelizer, which produces 3D grid data—voxels—from 2D/3D spatial information raw data.

MapboxVectorTiles have changed the geospatial ecosystem. Vector data distribution has become more efficient, map rendering has moved from server-side to client-side, map styling has become more common. But does that mean we no longer need raster tiles? No. Raster tiles still matter.

The Mineral Research and Production Support Platform (P3M) was conceived and is being implemented by a team of specialists from the Geological Survey of Brazil (SGB). with the support and participation of public and private entities directly or indirectly related to the Brazilian mineral industry, such as: Secretariat of Geology, Mining and Mineral Transformation (SGM) of the Ministry of Mines and Energy (MME); National Mining Agency (ANM); Agency for the Development and Innovation of the Brazilian Mineral Sector (ADIMB); Brazilian Association of Mineral Research and Mining Companies (ABPM); National Association of Aggregate Producers for Construction (ANEPAC); Brazilian Institute of Geography and Statistics (IBGE), and Brazilian Mining Institute (IBRAM).

The core development of P3M is being carried out by the UFLA (Federal University of Lavras) Agency for Innovation in Geotechnologies and Intelligent Systems (Zetta). Launched in late 2022, the Platform's main objective is to increase the attractiveness of investments and promote the sustainable development of the mineral industry, through the integration of data on mineral potential, infrastructure, costs, legislation, protected areas, and socioeconomic indicators of Brazil.

It has a crucial role in planning mineral research and production, generating qualified knowledge that is essential to support strategic decisions in both the public and private sectors. For industries and investors in the mining sector, the data available on the Platform contribute to supporting development plans and the exploitation of mineral deposits. Government entities also benefit primarily from information for monitoring the competitiveness of mineral research and production, thereby promoting greater attractiveness of investments for national development.

In short, P3M provides information free of charge through two main environments: (i) a map and data viewer that integrates geoscientific, economic, environmental and legal information, mining rights, among others, from various national and governmental entities; and (ii) a dashboard system with statistical data presentation, with filters by territorial unit (e.g., region, state, or municipality) and commodity.

One of the main differentiators of the P3M platform, among the various software products of SGB, is its complete conception and development using FOSS4G. The main objective of this work is to present the solution architecture used in the development, focusing primarily on the use of FOSS4G projects. In early February of this year, version 2.4.4 was released, with several improvements in terms of user experience and performance.

The platform deployment is entirely container-based, provisioned in a Kubernetes environment using Helm Chart, and data storage relies on PostgreSQL 14 databases with PostGIS 3.4. In general terms, the P3M Platform can be subdivided into three main components: (i) map and dashboard visualization system, (ii) data pipeline orchestration, and (iii) spatial data infrastructure.

The map and dashboard applications are deployed in two container images: the first, a backend project with Django + Rest framework, brings a system of REST APIs to define the content controls for the maps and dashboards. The application also has an administrative interface so that content managers can modify the application's behavior and the list of available layers and groups. The front-end component is a TypeScript application that relies on libraries such as OpenLayers to render the layer tree, delivered by the backend, and the dashboard data.

The data pipeline orchestration environment was deployed from a custom Apache Airflow container image, supplemented with libraries such as GDAL, GeoPandas, and Apache Arrow/Parquet.

For P3M, some data pipelines were developed to collect data of mining processes and financial compensation – available from the National Mining Agency (ANM). This data, originally delivered in spreadsheets and File Geodatabases, is sanitized, enriched, and transformed using GDAL, and finally saved in proprietary PostgreSQL/PostGIS tables of active mines, commodity associations, and mining process statuses. Currently, these pipelines run on Mondays, Wednesdays, and Fridays and are crucial for maintaining the dashboards and authorial layers.

The provision of maps and metadata on P3M strictly follows the use of OGC Open Web Services standards, such as WMS, WMTS, and WFS, and the platform consumes maps and metadata from a GeoNode installation owned by the SGB, in order to fulfill the role of Spatial Data Infrastructure.

The SGB’s GeoNode was deployed to a production environment in late 2025, initially to support this platform. Prior to this GeoNode, P3M had a dedicated GeoServer instance in its own workspace to serve the maps. After going into production, all layers and configurations were moved to this GeoNode. Currently, about 320 vector and raster layers from various Brazilian government data sources are cataloged and registered. Initially, this data was organized and cleaned to create a database that was placed as a datastore in GeoNode's GeoServer.

The option of hosting duplicate data instead of consuming OGC services from other institutions was chosen due to the need to have this data available on the platform, even if the source provider is unavailable, in addition to the geological service having a robust IT infrastructure. The P3M authorial layers are also hosted on GeoNode, but directly from the exposure of the spatial tables to the GeoNode Geoserver. Some of these tables, such as mineral resource data and systematic mapping, are also fed from pipelines in Airflow.

Finally, the P3M Platform, since its conception, has always been guided by the use of free and open-source libraries and frameworks to ensure transparency and interoperability. Alongside major FOSS4G projects, P3M stands out among the select group of data and service providers applied to Geology and Mining sectors by adopting frameworks that are free from restrictive licensing and bundled sales practices. The application is publicly available at https://p3mgeo.sgb.gov.br/.

FOSS4G software often has an international user base.Therefore, many FOSS4G software programs include internationalization features when they are created. These internationalization features are used to localize the software for various languages. This presentation will discuss the history, current status, and challenges of Japanese localization.

Hydrological ML requires costly upstream catchment aggregation. We present an efficient flow-accumulation-based method bypassing per-pixel delineation, achieving orders-of-magnitude speedups. Implemented in GRASS and Python, this open-source approach enables scalable, high-resolution modeling, demonstrated by a countrywide 90 m Random Forest nitrogen prediction.

Soil organic carbon (SOC) and total nitrogen were estimated using Sentinel-2 vegetation indices and machine learning in northeastern Thailand. After outlier removal, Random Forest achieved R² = 0.63 for SOC and R² = 0.39 for total N, with BSI and BAEI as dominant predictors.

The Kaii-Yokai Densho Database, maintained by the International Research Center for Japanese Studies (Nichibunken), a
Japanese national research institute for Japanese studies, is one of the largest structured archives of Japanese yokai
folklore. Each record contains a yokai or phenomenon name, a prefecture field, source metadata, and a short textual
summary. These fields make the archive searchable by name or region, but they do not directly support geographic
interpretation. Folklore records often refer to place in ways that are difficult to reduce to a single coordinate. A
summary may contain a formal place name, but it may also describe a riverbank, road, house, grave, pass, shrine,
village edge, or boundary without giving a mappable toponym. This creates a methodological problem for geographic
analysis: treating every record as a point can hide the uncertainty and evidence structure that make the record
spatially interpretable.

This study proposes a Folklore Geospatial Support Model for representing vague folklore place descriptions in the
Kaii-Yokai Densho Database. The model shifts the analytical unit from an asserted event coordinate to the set of
evidence channels that support spatial interpretation. These channels include prefecture or municipality support,
formal place mentions, place-description terms, coordinate-derived terrain evidence, text-derived terrain cues, human-
condition terms, boundary-interface types, source metadata, resolution level, and confidence reason. In this
representation, a coordinate remains useful for display and spatial computation, but it is not treated as the whole
geographic meaning of the record. The model keeps administrative metadata, gazetteer-resolvable place names, vague
place descriptions, and GIS-derived polygon context separate so that each record can be inspected according to the
evidence it actually supplies.

The workflow transforms 33,378 records from the Kaii-Yokai Densho Database into resolution-aware geospatial features.
First, all records receive prefecture-level support from the structured prefecture field. This creates complete
national coverage and provides a baseline for later refinement. Second, Japanese named entity recognition and rule-
based extraction are applied to the summary field. Formal place mentions are stored separately from terrain and place-
description terms because many folklore summaries describe settings without naming a municipality or landmark. Third,
a conservative local GADM level 2 gazetteer refinement searches for municipality-like names within each record’s
prefecture. A record is upgraded only when exactly one local municipality candidate is found. Fourth, GIS enrichment
adds coastline, lake, and river context using GADM, Natural Earth, and National Land Numerical Information River Data
(W05). The workflow also computes coordinate-only and text-aware terrain labels separately, so that GIS thresholds and
summary-derived terrain cues are not merged into a single evidential category.

The resulting coverage shows why folklore geography cannot rely only on formal toponym resolution. All 33,378 records
receive prefecture-polygon support. Formal place mentions occur in 9,361 records, or 28.1% of the archive. Place-
description terms occur in 25,023 records, or 75.0%. Terrain terms occur in 20,242 records, human-condition terms in
16,066 records, and strict boundary-interface terms in 9,047 records. Local GADM admin2 refinement upgrades 1,231
records, or 3.7% of the archive, to municipality-level representative points. Although this subset is small, it is
analytically useful because it provides a diagnostic set for measuring how much prefecture-centroid representation
changes environmental interpretation.

The admin2 subset reduces support area by a median 96.4% relative to prefecture support. It also shows that coarser
coordinates can change coordinate-derived terrain interpretation. Among the 1,231 locally refined records, 528
records, or 42.9%, change coordinate-only terrain class when the record is represented by an admin2 representative
point rather than by the prefecture centroid. The most common changes are coastal to inland-water, coastal to plain,
plain to inland-water, valley to inland-water, and plain to coastal. This result is not an archive-wide accuracy
estimate, because the refined subset is produced by successful local name matching. Its value is diagnostic: it
demonstrates that prefecture-centroid maps can change the environmental reading of records that contain enough
information for municipality-level support.

The study also evaluates place-function vocabulary under a prefecture-stratified permutation model. Category labels
are shuffled within prefectures, preserving regional composition while breaking category-specific association with
extracted place-function terms. This tests whether category labels are associated with particular kinds of setting
descriptions beyond prefectural background composition. The strongest associations include Kappa, or water spirits,
with water-setting vocabulary; Yurei, or ghosts, with death-ritual vocabulary; and Snake/Dragon, or serpentine beings,
with boundary-interface evidence. Observed-to-expected ratios provide interpretable effect summaries: Yurei/death-
Dragon/water-setting vocabulary at 2.23 times, and Snake/Dragon/boundary-interface vocabulary at 1.47 times. These
results describe archive language and evidence structure rather than verified physical event locations.

0.371 km. A prefecture-stratified category-shuffle null with 10,000 permutations expects a median of 0.295 km. The
observed median is larger than the shuffle expectation, so the Kappa result is treated as summary evidence rather than
a demonstrated physical water-proximity effect. This separation is central to the model: text-derived place-function
associations and GIS-derived proximity measurements are related but distinct forms of evidence.

Robustness diagnostics further test whether the strongest associations depend on category-name leakage or dictionary
design. Category-specific surface forms such as kappa, tengu, kitsune, yurei, and snake/dragon terms are masked from
summaries before extraction. Dictionary sensitivity is tested by randomly removing 10%, 20%, and 30% of terms from
each dictionary group across seeded trials. Boundary-interface definitions are also varied across broad, strict, and
conservative versions. The focal associations remain positive across these diagnostics, although their magnitudes
change. This supports the use of place-function and boundary-interface channels as inspectable evidence fields within
the archive representation.

The main contribution is a resolution-aware GIS representation for vague folklore places. The model preserves
uncertainty instead of forcing all records into single asserted coordinates. It also shows that coarse spatial support
is not only imprecise; in the locally refined subset, it can alter coordinate-derived environmental interpretation. By
separating formal toponyms, place-description vocabulary, coordinate-derived terrain, text-derived terrain, and
boundary-interface evidence, the workflow makes the geographic structure of a large folklore archive inspectable while
keeping the evidential basis of each record visible.

Open-source geospatial data and Overture Maps can generate detailed dwelling, transport, and service models supporting decision making. Case studies from Australia, New Zealand, Fiji, and Spain demonstrate applications for accessibility analysis, disaster risk reduction, and resilience, highlighting data strengths, limitations, and infrastructure needs.

point-tiler is a Rust CLI that converts massive LAS/LAZ point clouds into modern 3D Tiles (v1.1) via a glTF/GLB-centric workflow. This session covers city-scale reliability: LAZ-first I/O, coordinate/axis handling, external sorting beyond RAM, and practical compression choices.

Maplibreum: framework that makes the creation of web maps quite easy in a Python environment, making it well suited for interactive tools. Similar to Folium, but using the ever-growing MapLibre instead of Leaflet. I will present my motivations and the development of this new contribution to the FOSS4G community!

Invasive canopy-smothering vines such as Decalobanthus peltatus threaten tropical forests but are difficult to map under persistent cloud cover. This talk presents a SAR-based workflow using Sentinel-1 and open-source geospatial tools to detect and map infestation across Pacific islands, supporting reproducible, transferable monitoring for management.

In this presentation, we will try to understand the relevance of geospatial metadata in the modern technological landscape, and we will reflect on their usefulness.

Noodles.gl is an open-source visual programming environment that simplifies the creation of high-performance, cinematic geospatial animations by integrating Deck.gl, DuckDB and MapLibre into a reactive, node-based workflow. This session explores how GIS practitioners can use its "noodles and wires" interface to create beautiful GPU-accelerated visuals in the browser and allow you to "steer" AI-based maps

This session is aimed at developers, project leads, and organisational decision-makers who are navigating the question of how to sustain FOSS4G. We aim to leave attendees with a clearer framework for thinking about sustainability, and a more honest picture of what each path demands in practice.

The M4S Project leverages a low-barrier FOSS stack to map seagrass in Metinaro, Timor-Leste. By integrating participatory drone imagery and mobile data collection into a community-led workflow, we transition from manual methods to digital baseline mapping, empowering community stewardship to protect marine biodiversity and sustain local livelihoods.

Official statistics lack impact without spatial context. This talk presents UNICEF’s open-source stack - GeoRepo, GeoSight, and SDMXConnector - enabling seamless integration of SDMX data with harmonized boundaries. We address key challenges and show how to transform complex indicators into map-ready insights for better decision-making and GeoAI readiness.

A few collected thoughts about the impact of the current coding revolution for the future of FOSS4G projects, including a brief take on the economics of FOSS4G.

Apache SIS is a Java library for metadata, referencing, feature and grid coverage services with a focus on implementing OGC/ISO abstract models. This talk shows how SIS can handle some non-trivial cases such as non-linear localization grids and rasters crossing the anti-meridian.

What actually constitutes a valid polygon is a question that various stakeholders have very different views on – and even open-source libraries like GEOS and S2 are far from a consensus. This talk explores practical problems at the crossroads of coffee industry, geodata, and EU regulation.

As Open Data and FOSS4G grow, "open" does not always mean "accessible" due to language and local technical barriers. This presentation highlights these "invisible walls" and explores how Generative AI can bridge these gaps, making local geospatial data truly usable for the global community.

How far can a fully open geospatial stack take a natural-language request like “Show me cafes in Hiroshima City”? This talk presents TRIDENT, text2geoql-dataset, and a self-hosted planet-scale OpenStreetMap stack for building a grounded, interactive AI map agent.

Cogniscape is an open source iOS and Android app combining OpenStreetMap pedestrian data, spatial audio beacons, and on device AI obstacle detection for eyes free navigation. This talk covers the OSM data pipeline, offline routing, binaural audio, and real time computer vision for visually impaired users.

Bridge is a plugin for your favorite desktop GIS that makes it easy to publish your data to map and catalog services. It also allow you to search and consume your OGC services now.

A Search panel has now been added to the plugin, making the process of finding and using your and other resources much, much easier.

We present a pipeline transforming 190,000+ Japanese geological borehole records into interactive CesiumJS 3D Tiles. From open-sourcing a custom XML parser (boring-parser), through Implicit Tiling with glTF metadata extensions, to a browser-based WebGIS viewer — we share the journey of making subsurface data accessible to everyone.

In recent years, demand for geography education using GIS has been growing among junior and senior high school teachers. Since 2022, the compulsory geography curriculum in Japanese senior high schools has emphasized the use of GIS in classrooms. QGIS has attracted attention because it is freely available and enables advanced spatial visualization and analysis. However, the implementation of QGIS in education for junior and senior high school students remains limited compared with GIS education in universities. In this study, we conducted educational events for high school students to examine three aspects: students’ ability to operate QGIS, their understanding of spatial data concepts and processing, and their interpretation of geographical features across multiple layers.
The exercises were conducted as part of one-day or half-day educational programs introducing various GIS applications. The events were held eight times between 2018 and 2025. Although the teaching topics varied slightly across events, each QGIS-based exercise was consistently implemented using the same instructional content. At the beginning of the events, we delivered a 20–30 minute lecture introducing fundamental GIS concepts. The exercises, which lasted 60–75 minutes, required each student to operate QGIS individually. During the exercises, we focused on teaching basic GIS operations and spatial thinking by processing data related to the Kyodo River alluvial fan in Yamanashi Prefecture, Central Japan.
The QGIS-based exercise component consisted of five steps. In the first step, the students imported a 5-m resolution DEM (Digital Elevation Model) into QGIS. At the beginning of the session, the educator briefly explained basic PC terminology and the topographic characteristics of an alluvial fan. In the second step, the students attempted to identify the Kyodo River alluvial fan using the default DEM display, in which elevation values were represented from black to white. In the third step, they edited the DEM color settings based on cell values to enhance the visualization of the alluvial fan. In the fourth step, they created a slope map, a hillshade map, a 5-m contour map, and a 3D visualization to interpret the apex, middle, and toe areas of the alluvial fan. In the fifth step, they added land-use data to QGIS to examine the relationship between topography and land-use distribution by overlaying multiple layers. Finally, students attempted to complete two assignments to assess students’ learning outcomes: (A) whether they understood GIS operational procedures, and (B) whether they could interpret the geographical relationship between the alluvial fan and its land use. Assignment A was to answer “How should the raster dataset be styled in QGIS to display the rivers more clearly? Please include the following terms in your answer: elevation values, color allocation, hillshade, and layer overlay.” Assignment B was to answer “Explain the geographic characteristics by analyzing the relationship between orchards and man-made structures such as buildings and roads.”
At the end of the exercises, we asked the students to submit their assignment answers and complete a questionnaire. The survey was conducted only after obtaining the students’ consent to participate. The questionnaire consisted of two items: (a) simplicity of the exercise and (b) moments of difficulty or stumbling during the exercise. Item a was evaluated using a five-point Likert scale, where 1 indicated “complex” and 5 indicated “simple.” Item b consisted of seven options: “difficulty in following the instructor’s GIS operations,” “understanding computer terminology,” “communicating with educators when asking questions,” “importing and saving GIS data,” “typing on the keyboard,” “other,” and “none.” Students were allowed to select one or more options.
We collected responses from 167 students for Assignment A and from 150 students for Assignment B. To assess students' understanding of GIS processing and the features of DEM data, Assignment A was scored based on whether their responses included the four terms: “elevation values,” “color allocation,” “hillshade,” and “layer overlay.” The results show that approximately 80% to 95% of the students answered correctly, regardless of grade level. On the other hand, the overall correctness rate for Assignment B was 47.3%, which varied by students' grades. Over 75% of 11th- and 12th-grade students submitted correct answers, compared to 30%–40% of 7th- to 10th-grade students. To examine whether the distribution of correct and incorrect responses differed among grade levels, a chi-square test of independence was conducted for Assignment B. The result indicated a significant association between students’ grade level and achievement levels (χ² = 21.78, df = 5, p < 0.001). The following types of incorrect responses were observed: descriptions that focused solely on agricultural areas or building areas, and misinterpretations of the GIS maps or the characteristics of an alluvial fan. Another criterion for Assignment B was whether they could identify elevation or slope values from the GIS data; however, only three senior high school students included elevation or slope values in their responses. Most participants expressed general representations such as “this area is a steep slope,” “gentle slope,” “high elevation,” or “low elevation.”
In the questionnaire survey, 181 students responded to Item a and 142 to Item b. The perceived simplicity of the exercise was distributed as follows: 35 students (19.3%) selected “complex,” 75 (41.4%) “relatively complex,” 34 (18.8%) “neither,” 26 (14.4%) “relatively simple,” and 11 (6.1%) “simple.” For Item b, no single difficulty was reported by more than 50% of respondents. The most frequently selected option was “following the instructor’s GIS operations,” selected by 46 students (32.4%). Approximately 20% of respondents selected “understanding computer terminology” and “none.”
The results indicate that geography education using QGIS is feasible for junior and senior high school students. While most students understood basic GIS operations and spatial data concepts, many, particularly younger students, struggled to interpret their maps from a spatial thinking perspective. These findings highlight the importance of strengthening existing map-reading education in schools. In addition, adopting self-directed learning approaches that allow students to practice GIS operations step by step may be more effective for some students than simultaneous lecture-based instruction.

Open geospatial data have become increasingly available in recent years, enabling researchers and practitioners to analyze complex spatial phenomena at unprecedented spatial and temporal resolutions. Government statistics, environmental monitoring networks, and socio-economic indicators are now widely distributed through open data platforms and can be integrated with geographic information systems (GIS). However, many spatial datasets contain multiple interrelated variables that evolve simultaneously across both space and time. Examples include regional economic indicators, demographic statistics, environmental measurements, and infrastructure activity. Understanding the interactions among these variables is essential for studying spatial systems, yet most existing spatial modeling approaches treat each variable separately.

Traditional spatial regression models such as the spatial autoregressive (SAR) model and the spatial error model (SEM) are widely used in spatial econometrics and spatial statistics. These models capture spatial dependence through spatial weight matrices representing relationships among geographic units such as administrative regions or grid cells. While these methods have proven useful for analyzing spatial spillovers and diffusion processes, they are typically formulated for a single response variable. When multiple spatial variables are analyzed separately, potential interactions among them cannot be represented explicitly. As a result, the joint dynamics of spatial systems may remain partially unexplained.

This presentation introduces a multivariate spatio-temporal regression framework designed to analyze multiple spatial variables simultaneously. The proposed model extends classical spatial regression models to a multivariate setting by representing dependent variables as vectors that evolve across both geographic space and time. In this framework, spatial dependence is represented through spatial weight matrices commonly used in GIS-based spatial analysis, while temporal dependence is incorporated through autoregressive structures. The resulting model captures three important types of interactions: spatial spillovers across neighboring regions, temporal persistence within each variable, and cross-variable interactions among multiple spatial indicators.

The proposed modeling framework is expressed in a compact matrix form and estimated using maximum likelihood methods. The likelihood-based estimation approach enables efficient parameter estimation while maintaining statistical interpretability. Under suitable conditions, the parameters of the model are identifiable, meaning that spatial effects and cross-variable interactions can be uniquely recovered from observed spatial data. Because the model structure generalizes classical spatial econometric specifications, it naturally nests widely used models such as multivariate spatial autoregressive and spatial error models.

To evaluate the statistical performance of the proposed framework, Monte Carlo simulation experiments were conducted under a variety of spatial and temporal dependence scenarios. The simulations demonstrate that the estimation procedure can accurately recover true model parameters even when complex cross-variable spatial interactions are present. These results indicate that the multivariate formulation provides a robust statistical tool for modeling multidimensional spatial dependence.

The practical applicability of the approach is demonstrated using regional data from Japan. The empirical analysis focuses on the joint spatial dynamics of prefectural fertility rates and regional gross domestic product (GDP). These variables are closely related through demographic and economic processes and are likely to influence one another across neighboring regions. Using GIS-based regional datasets and spatial weight matrices representing prefectural adjacency relationships, the multivariate spatio-temporal model is estimated to capture both spatial spillovers and cross-variable interactions. Model comparison based on the Akaike Information Criterion indicates that the multivariate specification provides a better fit to the data than conventional univariate spatial regression models. The results suggest that explicitly modeling multidimensional spatial dependence improves both statistical performance and interpretability of regional processes.

Beyond this specific example, the proposed framework has broad implications for spatial data science and geospatial analytics. Many open geospatial datasets—including environmental observations, urban indicators, transportation data, and socio-economic statistics—contain multiple variables that interact across both space and time. The multivariate spatio-temporal regression framework provides a statistical methodology for analyzing such datasets while accounting for spatial dependence structures commonly used in GIS analysis. Because the model relies on spatial weight matrices and likelihood-based estimation, it can be integrated with existing geospatial workflows and open-source statistical environments. It is also potentially useful for policy evaluation, regional forecasting, disaster recovery assessment, and evidence-based planning using linked spatial indicators.

For the FOSS4G community, this work highlights the importance of combining statistical modeling with open geospatial infrastructures. While GIS platforms provide powerful tools for visualizing and managing spatial data, rigorous statistical models are essential for understanding spatial interactions and making reliable inferences from complex datasets. The proposed modeling approach contributes to this integration by offering a statistically grounded framework for analyzing multidimensional spatial processes using GIS-based regional data.

Future work will focus on implementing the proposed modeling framework in open-source statistical and geospatial software environments, enabling researchers and practitioners to apply multivariate spatial modeling techniques to a wide range of open geospatial datasets. By bridging statistical methodology and open geospatial data analysis, the approach aims to support more comprehensive modeling of complex spatial systems in both research and applied domains.

CesiumJS's built-in APIs hit CPU bottlenecks when rendering massive dynamic objects in real time. This talk presents how we used Vertex Texture Fetch (VTF) to store position data in GPU textures, enabling real-time updates without geometry reconstruction, applied to a satellite trajectory monitoring system.

An in-depth look at GeoServer’s support for Mapbox Vector Tiles, covering recent enhancements, flexible styling mechanisms, and performance optimization techniques such as feature consolidation. The session demonstrates generating high-quality, multi-projection base maps using OpenMapTiles and Planetiler, offering practical guidance for scalable, production-ready vector tile workflows.

This talk explores the co design process and technology stack behind a visual cultural database for Indigenous organisations in Australia, using open source geospatial tools and rapid field based software prototyping to make maps and media more accessible for non GIS users

This presentation describes how we connected Japan's national open geospatial data platform to AI agents using the Model Context Protocol — covering architecture, real-world use cases including infrastructure maintenance, and our vision for AI-ready open spatial infrastructure.

Open-source GIS tools such as QGIS and PostGIS excel at spatial analysis, but in earthquake-prone regions like Japan, layered maps alone are often insufficient. This talk introduces a practical extension that adds relational modeling to existing workflows, improving transparency and decision support in complex disaster scenarios.

Re:Earth is an open-source geospatial data platform for government and
public sector.
High school students in Hiroshima have used Re:Earth for a range of
initiatives, from peace education to inquiry-based GIS learning.
In this session, held in conjunction with FOSS4G Hiroshima, the students
will introduce their own initiatives and what they learned.

Japanese map culture has evolved through a pursuit of intuitive map designs as sophisticated infographics. We are redefining this cultural legacy within the digital space using FOSS4G technology to ensure its inheritance and evolution for the next generation.

As smart buildings, campuses, airports, and shopping malls grow increasingly complex, the need for precise and reliable indoor location-based services has become critical. However, unlike outdoor navigation—which benefits from decades of investment in mature infrastructure such as GPS, OpenStreetMap, and a rich ecosystem of standardized APIs—the indoor spatial data landscape remains fragmented. Addressing this gap requires not only robust data standards capable of representing indoor environments but also practical service interfaces that enable those standards to be easily used in real-world applications.

OGC IndoorGML[1] is the international standard that provides the formal foundation for indoor spatial information modeling. It represents indoor space using two complementary layers. The primal space layer encodes the physical geometry of indoor environments through cell spaces (e.g., rooms, corridors, and open areas) and cell boundaries (e.g., walls and virtual thresholds). In parallel, the dual space layer represents navigable connectivity as a graph, with nodes corresponding to navigable cells and edges representing their connectivity. This dual representation enables both geometric description and topological reasoning for indoor navigation. The latest IndoorGML 2.0 specification further extends this model by introducing support for multi-layered space representations. This capability allows multiple thematic perspectives—such as topographic layout, furniture configuration, accessibility constraints, or emergency routes—to coexist within a single building model. As a result, IndoorGML 2.0 provides a flexible semantic framework capable of supporting diverse indoor service scenarios.

Despite the richness of the IndoorGML data model, most prior work has focused on the construction phase of the indoor spatial data lifecycle, such as developing data generators, editors, and conversion tools. While these efforts have established essential foundations for creating IndoorGML datasets, comparatively less attention has been given to how such data can be efficiently served, queried, validated, and consumed by downstream applications. Consequently, a persistent gap remains between the semantic expressiveness of IndoorGML and its practical accessibility in production systems.

A key challenge in bridging this gap lies in the encoding and delivery of IndoorGML data in web-based environments. IndoorGML was originally designed with an XML-based encoding schema that provides strict schema validation and strong standards compliance. However, in practice, XML-based workflows often introduce considerable operational overhead. XML requires constructing the full document tree before accessing individual elements, involves relatively high parsing costs, and typically depends on domain-specific libraries that complicate integration into modern software stacks. In contrast, JSON has become the (de-facto) standard data exchange format for web-based systems due to its lightweight structure and direct compatibility with modern programming environments. JSON-based data structures map naturally to objects in most programming languages, support incremental parsing, and integrate seamlessly with web technologies such as JavaScript frameworks, mobile platforms, and contemporary GIS clients.

To address this need, IndoorJSON[2] has been proposed as a JSON-based encoding schema for the IndoorGML 2.0 conceptual model and is currently being standardized by the OGC. IndoorJSON preserves the complete semantic structure of IndoorGML—including CellSpaces, CellBoundaries, Nodes, Edges, multi-layered spatial models, and dual-space connectivity graphs—while providing a lightweight and web-friendly representation suitable for modern development environments.

Building upon this encoding approach, API – IndoorFeatures[3] is designed to operationalize IndoorGML-based data in web services. Rather than relying on customized or proprietary encodings, the API directly adopts the upcoming standardized IndoorJSON schema, ensuring interoperability and long-term compatibility within the evolving OGC ecosystem. In this sense, the implementation also serves as a practical proof of concept demonstrating that IndoorJSON-based indoor spatial data exchange can be deployed reliably in operational systems and can fully support real-world navigation and querying tasks.

API – IndoorFeatures is implemented as a RESTful service that exposes IndoorGML 2.0 data encoded as IndoorJSON through HTTP endpoints. The API allows developers to access and utilize indoor spatial data without requiring detailed knowledge of the underlying IndoorGML data structures. The implementation is built on extending pygeoapi[4], a Python-based framework that conforms to OGC API standards, and extends it with indoor-specific functionality while maintaining compatibility with the OGC API – Features[5]. In the system architecture, IndoorGML geometries and topological relationships are stored in PostgreSQL/PostGIS, enabling efficient spatial indexing and complex SQL-based queries. Indoor routing operations are supported through pgRouting, which operates directly on the dual-space connectivity graph stored in the database. To maintain data consistency, all write operations are validated against IndoorGML 2.0 requirements, ensuring that the stored data preserves the semantic integrity of the standard. Additionally, a bundled web application provides interactive visualization, geometric querying, and routing capabilities directly within a browser environment.

This study presents the design requirements and implementation considerations encountered during the development of API – IndoorFeatures and evaluates its effectiveness through practical use cases using real building datasets. By extending the indoor spatial data lifecycle beyond data construction to include serving, validation, sharing, and consumption, this work provides a reusable, extensible open-source platform to support the development of next-generation indoor location-based services.

References:
[1] OGC IndoorGML 2.0 Part1 – Conceptual Model, https://docs.ogc.org/is/22-045r5/22-045r5.html
[2] IndoorJSON Schema and Sample Data, https://github.com/opengeospatial/IndoorGML-SWG/tree/master/IndoorGML2/IndoorGML2_metanorma/Part%20II/JSON
[3] API – IndoorFeatures, https://github.com/STEMLab/API_IndoorFeatures
[4] pygeoapi, https://pygeoapi.io/
[5] OGC API - Features - Part 1: Core corrigendum, https://docs.ogc.org/is/17-069r4/17-069r4.html

While LLMs excel at synthesizing text, they lack geospatial awareness and cannot reason over spatial networks. This talk demonstrates how Geo-GraphRAG and DGGS, grounded in open standards, enable geospatial reasoning with transparent, traceable results.

“Find me sunny buildings near Hiroshima Station.”
This may sound easy for AI, but handling spatial relationships and 3D information such as which way building surfaces face is not straightforward.
This talk presents a portable DuckDB-based approach to 3D city model search that combines spatial, semantic, and geometric cues.

QGIS is translated into many languages, but structured tutorials remain scarce. In 2024, we launched a Japanese QGIS platform now reaching 20,000 monthly users. Our data shows onboarding content — installation guides, first steps — draws the widest audience. We share our approach and how to replicate it in your language.

Re:Earth Flow is an open-source visual ETL engine for 3D city models, shown here last year as an alpha. This year we stopped adding and started reading it back: a new geometry core, a new expression language, and a written standard that all 164 of our workflow actions now have to meet. What we found, what we changed, and which decisions we stopped delegating. With a live demo on Japan's PLATEAU data.

Understanding the Earth system requires observing it: repeated and spatially explicit Earth observation (EO) measurements are what turn “the planet” into variables we can analyze, compare and model. Over the last decades, EO has evolved quickly, and the number and variety of remote sensing instruments available to the community has grown even faster.

Today, EO spans instruments onboard satellite platforms, airborne sensors, and in-situ or terrestrial measurement systems. Passive optical sensors measure reflected (and, for some missions, emitted) radiation in multispectral (e.g. Sentinel-2 and Landsat) and hyperspectral systems (e.g. EnMAP and EMIT), typically across the ultraviolet, visible, near-infrared and short-wave infrared, and sometimes extending into the thermal infrared. Active systems such as synthetic aperture radar (e.g. Sentinel-1) provide illumination-independent observations and enable monitoring through clouds and different atmospheric conditions. Airborne and UAV instruments can push spatial detail to the centimeter scale and offer flexible acquisition timing, while ground-based sensors provide the most direct in-situ reference.

This explosion of platforms and modalities gives rise to a very practical problem: many users simply do not have a single and reliable place to find out what instruments exist and, crucially, what their characteristics are (from governance and mission status to spectral configuration, imaging geometry, spatio-temporal resolution, or where the data can be accessed).

Although data catalogues (including large platforms and STAC catalogues in particular) have made dataset discovery dramatically easier, they are fundamentally designed to describe collections of data products and their assets, not to provide a persistent, curated description of the instruments that generated them.

Here we present Awesome Earth Observation Instruments, an open, community-oriented registry of EO instruments providing machine-readable instrument metadata intended to complement existing data catalogues and support reproducible, automated geospatial workflows.

The catalogue is an open registry: a community-maintained listing of EO instruments hosted on GitHub, where contributors can add instruments and their associated metadata under a shared standard. The specification is designed to be straightforward to read and implement, while remaining strict where it matters. Contributions are validated against a YAML-based JSON Schema.

The core schema requires a minimal set of attributes that identify an instrument and support practical use, including id, name, acronym, start date, and an explicit operational status (e.g. operational, retired, or planned). Instrument and platform type are treated as first-class metadata and contributors indicate whether an instrument is, e.g., multispectral, hyperspectral, radar, lidar, RGB, or other, and whether it operates from a satellite, an aircraft, a UAV, or in a terrestrial or in-situ configuration. Governance information is also captured (operator and responsible institutions, and whether the instrument is public or private), and authoritative references are required to document the provenance of reported properties. The core schema supports optional information such as an end date for retired instruments, notes for additional context, and general data-access links.

A key design goal of the catalogue is modularity and extensibility. The catalogue follows an approach inspired by the STAC specification, using optional extensions that can be included when information is available. We currently provide four extensions: spectral, imaging, and two data-access related extensions (Earth Engine and Planetary Computer).

The spectral extension captures spectral characteristics using spectral bands or a spectral range, and (optionally) spectral response functions. For multispectral instruments, spectral bands represent per-band parameters such as center wavelength and bandwidth, together with band description and ground sampling distance (GSD), and include per-band common names aligned with the eo-stac extension. For hyperspectral instruments, spectral range captures the wavelength interval (minimum and maximum) and the total number of bands. Spectral response functions are an optional component, but can be stored per band when available.

The imaging extension provides additional parameters relevant for interpretation and modelling, including swath width, across- and along-track field of view (FOV), and instantaneous field of view (IFOV). Horizontal and vertical FOV can be added as well. It also supports optical parameters such as entrance pupil diameter, focal length, f-number, and a shared GSD when applicable.

Finally, the data-access extensions record structured links to data distribution points in Google Earth Engine and Planetary Computer (for different processing levels, including raw, top-of-atmosphere, and bottom-of-atmosphere), indicating the reference source links as well as the collection names for data querying.

Having the standard in an open GitHub repository makes it straightforward for the community to use the catalogue in practice, both as a human-readable reference and as a machine-readable registry inside existing workflows and projects. In parallel, the repository itself works as a simple discovery interface since users can browse what instruments are out there, inspect their characteristics, and make more informed choices based on their specific needs.

Because the schema is open, versioned, and simple, it is easy for the community to add new instruments as they appear, and to update or extend the schema (or extensions) as requirements evolve. This creates a clean path for existing EO data catalogues to link data products to the instruments that generated them through shared identifiers. To avoid duplicated instrument entries and ensure quality metadata, every entry must have an unique identifier and links to the sources of the introduced metadata for each instrument entry (links to the original and official instrument sources or operators have first-class priority over links to unofficial sources). Furthermore, an audit will be performed by a maintainer for every added instrument, similar to the audits performed for other open cataloguing systems (e.g. conda-forge or Awesome Spectral Indices).

Looking forward, we expect the catalogue to be used as a community reference point for EO instrument characteristics, with a workflow-friendly input that lowers the barrier for both discovery and integration. To support programmatic use, we anticipate developing a Python package that provides validated access to the registry, enabling direct interoperability in the environments where data access and analysis actually happen.

We also aim to align the catalogue with complementary open initiatives, including the TACO (Transparent Access to Cloud-Optimized datasets) specification and Awesome Spectral Indices (ASI), so that instrument metadata, data access, and band and index semantics can connect directly across open geospatial ecosystems.

A modern vector map of Japan built from GSI open data and OpenStreetMap, focused on cartographic design and customization. Built on vector tiles, it integrates with MapLibre and open-source tools, balancing Japanese mapping conventions with modern usability.

Overview of how the ESA Planetary Science Archive (PSA) uses open-source technologies such as OpenLayers, GeoServer, Three.js, and PostGIS to manage, visualize, and distribute planetary data, particularly for Mars, enabling interactive exploration, 3D visualization, and efficient access for the scientific community.

V-World, Korea’s national geospatial platform by MOLIT, provides 2D/3D data and APIs. Supported by Digital Twin Platform policy, it advances 3D services and analytics. This study reviews operations, recent upgrades, and future plans including improved data management and Geo-AI integration

geo-base is a self-hostable open-source geospatial platform that combines
a tile server, a management dashboard, and an MCP connector, enabling small
teams to manage and query their own spatial data in natural language without
relying on external cloud services.

The future of geospatial analysis lies in combining the scalability of cloud AI services with the flexibility of open source tools. This session showcases architectural patterns for leveraging the best of both worlds, alongside case studies demonstrating how organizations use this combination to achieve their goals.

Using cheap 250 g consumer drones introduces challenges when processing imagery at scale. At HOTOSM we have been refining workflows alongside OpenDroneMap to process areas of <100 km² captured using drones such as the DJI Mini series, combining drone-specific correction techniques with scalable Kubernetes-native processing pipelines.

This study introduces a three-dimensional GIS-based wildfire fireline visualization system developed using open-source GIS technologies. The ultimate goal is to further develop the system into a real-time wildfire monitoring platform that can support rapid and effective responses to large-scale wildfires in the future.

Ever wondered how geospatial data travels from cloud storage to your browser's GPU? This talk walks through the full pipeline: accessing cloud-optimized formats like PMTiles, decoding vector tile formats like MVT and MLT, and rendering on the GPU. We also peek at future directions like WebGPU and WebAssembly.

Source Cooperative's data proxy lets users access open datasets through S3-compatible tools. We rebuilt it from scratch as Multistore, an open source S3 gateway designed to be reusable across the ecosystem. This talk covers why we rebuilt, what we learned, and how others can adopt it.

Re:Earth is an open-source geospatial data platform for government and public sector.
High school students in Hiroshima and Nagasaki have used Re:Earth for a range of initiatives, from peace education to inquiry-based GIS learning.
In this session, held in conjunction with FOSS4G Hiroshima, the students will introduce their own initiatives and what they learned.

This session follows on from the QGIS "Feature Frenzy" talk and is your once-in-a-lifetime opportunity to ask Marco Bernasocchi (QGIS.org Chairperson) and Nyall Dawson (QGIS Core Contributor) anything about QGIS.

Re:Earth is an open source, no-code WebGIS platform built on Cesium, developed by Eukarya Inc. This talk presents two educational case studies — high school students participating in urban planning and elementary school students creating a their town 's map — showing how open source geospatial tools empower non-expert users.

In this talk, I'll try to explain the metadata problem in geospatial data and why things like STAC are useful but cannot always solve your problem.

This work explores how geospatial systems can move beyond map-centric interfaces by transforming complex GIS data into interpretable visual narratives. It introduces the concept of a Spatial Narrative Interface and examines how structured visual explanations can support human understanding and decision-making.

We present i.hyper, a comprehensive hyperspectral processing addon for GRASS. It includes modules for import, metadata handling, atmospheric correction, preprocessing, spectral resampling, indices, albedo, visualization, exploration, export, and more, covering the workflow from raw satellite products to analysis-ready data.

Morining Break sponsored by MIERUNE Inc. & Cesium

Open-source communities keep asking how to sustain what they give away for free. Beyond donations and commercialization, this keynote brings in a third reference point rarely examined in depth: the cooperative. Three speakers explore what a bounded solidarity can teach an unbounded gift.

Retrieval-Augmented Generation (RAG) has become a widely adopted approach for enhancing Large Language Models (LLMs) with external knowledge. Recent studies such as Spatial-RAG (Yu et al., 2025) and MapQA (Li et al., 2025) have proposed frameworks that integrate spatial data with LLMs for geospatial question answering. However, most existing work focuses on proposing individual frameworks rather than systematically comparing multiple architectural approaches under controlled conditions. Consequently, practitioners lack empirical guidance on which RAG architecture best suits geographic Point of Interest (POI) queries—a common use case in location-based services, urban planning, and tourism applications.

This study presents a systematic comparison of six RAG architectures for geographic POI question answering, evaluated under identical conditions to identify which architectural approach best suits spatial queries about real-world POIs. Specifically, we investigate the relative effectiveness of deterministic structured processing, graph-based retrieval, and vector search—and whether these components function as complementary rather than competing strategies.

We constructed a POI dataset by extracting 1,047 points of interest from OpenStreetMap (OSM) within the Shibuya station area of Tokyo, Japan. Each POI was enriched with computed spatial metadata including Haversine distance and compass direction (eight cardinal/intercardinal directions) from the Shibuya station reference point (35.658034°N, 139.701636°E). To validate generalization across different urban contexts, we additionally collected POIs from three other major Tokyo districts—Shinjuku, Ikebukuro, and Tokyo Station areas—totaling approximately 3,600 POIs. All POI embeddings were generated using the multilingual-e5-base model (768 dimensions) and stored in ChromaDB, an open-source vector database.

To address these research questions, we designed and compared six RAG architectures, all sharing the same underlying LLM (Qwen2.5-7B-Instruct, 4-bit quantized) and embedding model: (1) Vector RAG (Baseline), using cosine similarity retrieval with no spatial computation; (2) Structured RAG, introducing four deterministic spatial processing modules—proximity sorting, sensitivity analysis across varying search radii, directional comparison, and category aggregation—triggered by question-type classification and applied in combination rather than as mutually exclusive alternatives, with vector search always contributing as a complementary source; (3) GraphRAG, constructing a knowledge graph with 1,090 nodes and approximately 82,000 edges across seven relationship types for relational context retrieval; (4) Hybrid RAG, combining structured spatial processing with vector retrieval in a fixed pipeline; (5) Adaptive RAG, dynamically selecting between Structured RAG and GraphRAG based on query characteristics; and (6) Agentic RAG, employing a ReAct-style reasoning loop where the LLM autonomously decides which retrieval tools to invoke.

Evaluation relied on a hierarchical five-level test prompt framework designed to capture the diverse reasoning demands inherent in geospatial queries, with a total of 55 prompts (L1: 10, L2: 15, L3: 10, L4: 10, L5: 10) spanning 12 subcategories: L1 Basic Retrieval (factual POI lookup by name or category), L2 Spatial Reasoning (proximity ranking, density estimation, directional comparison), L3 Constraint Satisfaction (queries combining spatial and categorical filters), L4 Decision Support (location selection and business-oriented recommendations requiring multi-criteria reasoning), and L5 Advanced Reasoning (sensitivity analysis under varying search radii, comparative evaluation, and uncertainty quantification). All systems were evaluated using identical test prompts and a multi-dimensional scoring function comprising: keyword success rate, composite score (0–100), reasoning quality (0–5), evidence citation (0–5), constraint satisfaction (0–5), and uncertainty acknowledgment (0–5). All metrics were computed via rule-based automatic evaluation using regular expression and keyword matching—not LLM-as-judge—ensuring deterministic and reproducible scoring.

In the architecture comparison, Structured RAG achieved 89.1% overall accuracy, outperforming GraphRAG at 76.7% and Adaptive RAG at 86.1%. Notably, Structured RAG outperformed GraphRAG even on graph-specific test cases designed to favor relational retrieval (86.9% vs. 80.2%), demonstrating that coordinate-based deterministic computation renders explicit graph edge representations largely redundant for spatial relationship reasoning. Performance differences were most pronounced at L2 and L5, where deterministic spatial computation provided precise distance and directional data that graph traversal could not match.

In the multi-area generalization evaluation across four Tokyo districts, Hybrid RAG achieved the highest keyword success rate of 96.9%, followed by Adaptive RAG (96.9%), Graph RAG (100.0% keyword but lower composite), and Agentic RAG (89.2%). Multi-dimensional composite scoring revealed a persistent quality gap across all systems: Hybrid RAG scored 67.1/100 in composite quality despite near-perfect keyword accuracy, with the gap widening at L4 (Decision Support) and L5 (Advanced Reasoning), where keyword matching alone failed to capture reasoning quality.

We attribute the consistent superiority of Structured RAG over GraphRAG to the inherent computability of spatial relationships: distance and direction between POIs are directly calculable from coordinates, making explicit graph edges redundant for the majority of geographic queries. The quality gap observed in Hybrid RAG—near-perfect keyword success yet composite scores well below that level—indicates that retrieving relevant POIs is necessary but insufficient; generating well-reasoned answers at L4–L5 requires structured context that guides the LLM's reasoning process, not merely correct entity retrieval. This finding empirically demonstrates that single-metric evaluation based on keyword accuracy is insufficient for geospatial QA, underscoring the importance of hierarchical, multi-dimensional evaluation frameworks.

Our systematic comparison yields three principal findings for the geospatial open-source community: (1) deterministic structured processing using coordinate-based computation consistently outperforms graph-based retrieval for POI-centric spatial queries; (2) structured processing and vector search are complementary—not competing—retrieval strategies, and combining them yields the best results; (3) controlled multi-architecture comparison under identical evaluation conditions is essential for advancing geospatial RAG research. All experiments were conducted using exclusively open-source software (ChromaDB, NetworkX, Hugging Face Transformers) and open data (OpenStreetMap), ensuring full reproducibility. Future work will extend this evaluation framework to larger-scale multilingual POI datasets beyond the Tokyo metropolitan area and investigate adaptive weighting between structured and vector retrieval components.

Geospatial analytics and Geospatial queries can be a challenge for Geospatial backend it's can be slow and or might clause a problems. This is architecture design use case on forest and land monitoring platform (LANDX) by GISTDA Thailand. how i deal with large geospatial data and big geospatial query.

Machine learning is widely used to derive spatial data for environmental decision-making, including in digital soil mapping (DSM), where mapped products support carbon accounting, land management, crop modelling, food security, and policy reporting. In DSM, soil organic carbon (SOC) has become a key target variable because many climate- and land-related decisions now depend on spatially explicit estimates of SOC stocks and change. Recent advances in machine learning-based spatial modelling have accelerated SOC mapping, while growing standardisation efforts have improved the consistency of data collection and analysis workflows.

Predictions from machine learning-based environmental models are inherently uncertain, both due to incomplete knowledge of the processes controlling the target variable (epistemic uncertainty) and irreducible variability in the system itself (aleatoric uncertainty). Predictions should therefore always be accompanied by their uncertainty estimates to support decision-making and avoid overconfidence in model outputs. Previous studies show that interpretation and decision-making depend not only on the reported uncertainty values, but also on the type of uncertainty reported and how it is communicated. This makes uncertainty quantification especially valuable in open-source applications, as it improves the reliability and explainability of shared models and datasets.

Uncertainty quantification in DSM has received increasing attention in recent years, although most studies either report a single uncertainty measure or compare uncertainty performance across different modelling algorithms. In SOC modelling, uncertainty is most commonly expressed through prediction intervals, often derived from Quantile Regression Forest (QRF) or related ensemble-based approaches, with 90% prediction intervals widely adopted following the GlobalSoilMap framework (Arrouays et al., 2014).

However, much less attention has been paid to comparing different uncertainty quantification methods within a single model family, such as Random Forest. In applied workflows, the modelling algorithm is often chosen before uncertainty becomes a methodological consideration, usually on the basis of predictive performance. Therefore, uncertainty quantification often needs to be added to an existing workflow rather than drive model choice. This creates a need for systematic comparison of uncertainty quantification methods within a shared model framework, allowing different uncertainty representations to be evaluated under otherwise identical modelling conditions, with attention not only to implementation but also to the sources of uncertainty represented and the quality of the resulting estimates.

We address this need by systematically comparing uncertainty quantification approaches applicable to Random Forest-type models, which remain among the most widely used methods in DSM due to their robustness and compatibility with open-source geospatial workflows. We examine how different approaches influence the spatial pattern of estimated uncertainty, identify the main drivers of these differences, and assess the quality and interpretability of the resulting uncertainty estimates. More broadly, we aim to provide a structured comparison of uncertainty quantification methods that can be integrated into Random Forest-based soil modelling workflows, highlighting their practical differences and implications for use.

We apply the comparison to a baseline Random Forest (RF) model for national-scale SOC prediction in Estonia, largely developed by Kmoch et al. (2021) and Choi et al. (2025), and organise the tested methods into two broad groups reflecting primarily model-related versus data-related sources of uncertainty.

The first group includes RF-based interval methods. Under Quantile Regression Forest (QRF), we consider (a) conventional 90% prediction intervals expressed through interval width, (b) quantile formulations emphasising the tails of the predictive distribution, and (c) formulations emphasising the central region containing most predictions. We also incorporate conformal prediction into the RF pipeline to generate prediction intervals with user-specified confidence levels, including (d) split conformal prediction with a simple scoring function (Singh et al., 2024), (e) conformalized quantile regression, which produces wider intervals for more difficult predictions (Romano et al., 2019), and (f) class-conditional conformal prediction based on land-use or related covariate classes. In addition, we assess model sensitivity through (g) hyperparameter resampling.

These RF-based interval methods describe uncertainty given the available model and training data, but not whether the available data themselves are representative of the prediction domain or free from substantial error. This limitation is especially relevant in DSM, where sample coverage is often sparse or uneven, and where both predictors and observations contain uncertainty. We therefore distinguish a second group of approaches targeting data uncertainty, particularly uncertainty related to representativeness and extrapolation. This group includes the Area of Applicability (AoA) framework of Meyer and Pebesma (2021), from which we consider both (h) the binary AoA mask and the (i) continuous dissimilarity index (DI) as possible uncertainty expressions. We additionally explore related out-of-distribution detection approaches as complementary indicators of covariate-space dissimilarity.

We compare the resulting uncertainty representations through a joint evaluation framework combining interval quality diagnostics and spatial pattern analysis. For RF-based interval methods, this includes empirical coverage of nominal prediction intervals, average interval width, and conditional coverage, allowing under- and over-confidence to be distinguished. For dissimilarity-based approaches, we assess whether areas flagged as dissimilar, outside the AoA, or otherwise weakly supported by the training data are associated with larger residuals and poorer interval calibration. In addition, we examine spatial autocorrelation and clustering of high-uncertainty regions for each method, and compare these with known features of the sampling design and environmental covariates.

Preliminary results show that the choice of uncertainty quantification method can lead to substantially different uncertainty patterns even when applied to the same RF model. Although dissimilarity-based approaches sometimes highlight the same areas as interval-based methods, and high dissimilarity index values often coincide with wider intervals and poorer interval calibration, this relationship is not consistent across all soil types and land cover classes. These findings indicate that method choice influences the uncertainty that is ultimately communicated and interpreted, and that combining multiple quantification methods provides a stronger basis for interpreting model outputs and supporting decision-making. Furthermore, disagreement between methods may itself serve as an additional uncertainty diagnostic.

To support reproducibility and reuse, the full Python workflow and code for the uncertainty quantification and comparison framework will be made openly available after the publication of related work.

Citations:

Arrouays, D., McBratney, A.B., Minasny, B., Hempel, J.W., Heuvelink, G.B.M., MacMillan, R.A., Hartemink, A.E., Lagacherie, P. & McKenzie, N.J. (2014). The GlobalSoilMap project specifications. GlobalSoilMap: Basis of the global spatial soil information system.
Choi, J., Kmoch, A., & Uuemaa, E. (2025). Optimisation of sampling design for multivariate soil mapping with machine learning. In Proceedings of the 2025 conference on Big Data from Space (BiDS’25). Publications Office of the European Union. https://doi.org/10.2760/2119408
Kmoch, A., Kanal, A., Astover, A., Kull, A., Virro, H., Helm, A., Pärtel, M., Ostonen, I. & Uuemaa, E. (2021). EstSoil-EH: a high-resolution eco-hydrological modelling parameters dataset for Estonia. Earth System Science Data, 13(1), 83-97.
Meyer, H., & Pebesma, E. (2021). Predicting into unknown space? Estimating the area of applicability of spatial prediction models. Methods in Ecology and Evolution, 12(9), 1620-1633.
Romano, Y., Patterson, E., & Candes, E. (2019). Conformalized quantile regression. Advances in Neural Information Processing Systems, 32.
Singh, G., Moncrieff, G., Venter, Z., Cawse-Nicholson, K., Slingsby, J., & Robinson, T. B. (2024). Uncertainty quantification for probabilistic machine learning in earth observation using conformal prediction. Scientific Reports, 14(1), 16166.

The Earth System Grid Federation (ESGF) is a global partnership supporting the distribution, archive, and discovery of climate data. Its new architecture introduces STAC catalogues, Kerchunk‑enabled access, and an event‑driven search system synchronising two discovery nodes, improving consistency, reliability, and interoperability across climate and Earth observation communities.

We share GISTDA’s Dragonfly project lessons on national-scale agricultural monitoring. We optimize COG performance to reduce TTFB and ensure data reliability using Great Expectations automated quality gates. Attendees gain actionable DataOps insights for building resilient, high-performance cloud-native ecosystems from real-world operational experience.

A practical look at how the Pacific Community is shifting from short‑term geospatial projects to sustainable, open data products through shared standards, regional governance, and coordinated stewardship.

Under the BIG-Z project, Zanzibar has geospatially enabled its Municipals Revenue (MUTM) software. This talk demonstrates using a 100% open-source geospatial stack including QGIS, PostGIS, and GeoServer to solve municipal finance challenges by integrating the Municipal Revenue Management (MUTM) database with payment gateway for real-time compliance visualization

Mastering QGIS Map Themes streamlines cartographic workflows, automates Atlas production and enhances field data collection using Mergin Maps and QField. This approach ensures consistent, efficient map production and rapid switching between map views for diverse GIS tasks.

This presentation describes a cloud-based approach to managing large-scale maritime data, enabling near real-time visualization alongside historical analysis. It highlights challenges in handling high data volumes, ensuring performance, and controlling access, while supporting efficient and scalable geospatial data publication.

This presentation reports the renewable energy zoning project conducted in Urahoro, Hokkaido. By combining QGIS with various open data sources and utilizing a custom-built participatory mapping system developed with generative AI, we visualized locations valued by residents to achieve an appropriate balance between conservation and renewable energy development.

We present the Geonorge.Forvaltning.Client, a web-based GIS interface designed for the collaborative maintenance of Norway’s national map data. By lowering technical barriers for public administrators, this tool fosters inter-agency transparency and democratizes spatial data management, aligning with the FOSS4G spirit of open, accessible geospatial technology.

We continue to see exponential growth in the volume of data available to understand the planet. Now, with Agentic systems, we are on the cusp of understanding this data as fast as we collect it.

The key to this success will be Cloud-native geospatial. Agents want to discover data, query it, transform it, and hand back an answer. If your data isn't in a format they can reach, you're no longer part of the conversation. This talk unpacks that shift: why predictable machine access patterns favor cloud-native formats like STAC, COGs, GeoParquet, and Zarr; how agentic interfaces are quietly retiring the static dashboard; and what it means for data providers, services companies, and analysts.

National Land Survey of Finland is developing a new system for updating National digital terrain model (DTM). This talk covers the main features of the new solution. Focus is in the process workflow and how the quality management is done with a new system called Pinta.

Maplat is an open-source platform for integrating historical maps, illustrated maps, tourist maps, and other non-georeferenced maps with modern web mapping technologies.

Unlike our technical presentation, this sponsor session focuses on real-world applications, business opportunities, and collaboration rather than implementation details.

We will introduce how Maplat is being applied to cultural heritage, tourism, museums, local governments, and digital transformation projects, and discuss how open-source software can become sustainable through partnerships and commercial services.

Applied Technology Co., Ltd. has been transforming our customers' challenges into value through cutting-edge technology in two core business areas: "Monodzukuri" (manufacturing) and "Machizukuri" (urban development). Specifically in the "Machizukuri" sector, we have delivered highly specialized engineering services centered around four main pillars: "Environment," "Disaster Prevention," "Urban Development Support," and "System Development." In this session, we will introduce these technologies and showcase our newly released WebGIS-based system, "Machi-Space®," which supports the fields of "Environment" and "Urban Development Support."

This is an introduction to the nationwide forest open data sets that JAFTA has been involved in developing, such as the National Forest Inventory data, the National Forest Resource Mesh Map Tiles, and the zoning tool "Morizon".

Accurate land-cover classification maps are essential geospatial resources that support a wide range of societal applications, including urban development, environmental monitoring, and disaster risk management. With the growing availability of satellite remote sensing data, pixel-wise classification methods, which assign a single land-cover class to each image pixel, have become widely adopted. However, when medium-resolution satellite imagery, such as Sentinel-2, is used, a single pixel often contains more than one land-cover type. This phenomenon, commonly referred to as the mixed-pixel problem, leads to the loss of information about minority classes within a pixel, making it difficult to accurately characterize fine-scale surface conditions and limiting the practical utility of classification results. One promising solution to this issue is compositional classification, which estimates the proportion of each land-cover class present within a pixel rather than assigning a single label. This approach retains information about all classes, including those that occupy only a small fraction of a pixel. Compositional data are subject to two mathematical constraints: all values must be non-negative, and the proportions across all classes must sum to one. Yet, the implementation of such compositional classification is challenging, as machine learning or deep learning model architectures do not consider these characteristics properly. This study proposes a deep learning framework for compositional land-cover estimation at 10 m spatial resolution. We explore combinations of input features, model architectures, and loss functions suited to compositional outputs. Rather than seeking a single definitive solution, we aim to provide practical insights into how these design choices interact and influence estimation quality.

As an experiment, we examined two inputs, three models, and two loss functions. Two types of 10-m level input features were examined. The first was spectral reflectance data from 10 multispectral bands (B2-8, B8A, B11, and B12) in Sentinel-2, which directly capture surface characteristics across visible and infrared wavelengths. The second was Embedding V1, a 64-dimensional feature vector generated by AlphaEarth Foundations, a geospatial embedding model that integrates multiple data streams, including optical, radar, and LiDAR observations from multi-temporal satellite imagery. We used OpenEarthMap (OEM), a publicly available global dataset with high-resolution eight land-cover labels at 0.25–0.5 m spatial resolution, as reference data. To generate training targets at medium resolution, we aggregated all OEM pixels within each 10 m pixel and computed the fractional coverage of each land-cover class. This aggregation produced pixel-level composition vectors that served as ground truth for supervised learning, establishing a direct correspondence between medium-resolution inputs and high-resolution reference proportions. Three deep-learning model architectures were tested: a multilayer perceptron (MLP), a two-dimensional convolutional neural network (2D-CNN), and a three-dimensional convolutional neural network (3D-CNN). All models consisted of two fully connected layers followed by a Softmax activation function to ensure outputs formed valid compositional proportions. The CNN-based models were included to capture spatial context from neighboring pixels, while the MLP was used as a simpler baseline that processes each pixel independently. The dataset was split into 70% for training, 10% for validation, and 20% for testing. We compared two loss functions: Mean Absolute Error (MAE), a widely used metric in regression tasks, and the Aitchison distance, a measure derived from compositional data analysis. The Aitchison distance evaluates the relative differences between components rather than their absolute errors by applying a logarithmic transformation and centering to each proportion. To eliminate redundancy associated with the compositional constraint, we applied a linear transformation using an orthonormal basis, projecting the data into a D−1 dimensional space. This metric enables stable distance computation while reflecting the geometric structure and constraints inherent to compositional data. This combination of architectures and loss functions allowed for a systematic comparison across multiple design dimensions. We evaluated the performance of each combination using MAE and the Aitchison distance to identify the most effective framework for compositional land-cover classification mapping. All processing and training are implemented with open-source Python geospatial/ML tools, and we will release code and experiment configurations to support reproducibility.

The experimental results revealed clear differences based on the combination of inputs and model architecture. When using only Sentinel-2 spectral bands as input, 3D-CNN models outperformed the MLP (MAE of 0.1126 versus 0.1185), indicating that spatial context from convolutional operations benefits models when input features are limited to raw reflectance. In contrast, when using Embedding V1 as input, the MLP achieved the best performance (MAE of 0.0989) among all tested models. This suggests that Embedding V1 already encodes rich spatial and semantic information through its pretraining process, making simpler models better suited to leverage these representations without overfitting to their internal structure. Regarding the choice of loss function, the resulting maps showed a meaningful difference between MAE and Aitchison distance. Models trained with MAE tended to produce smoothed predictions, with estimated proportions concentrated near intermediate values and rarely approaching 0 or 1. This smoothing effect is a known limitation of MAE in compositional contexts, where extreme values are physically meaningful. When the Aitchison distance was used instead, the estimated proportion maps showed sharper contrasts and more realistic spatial patterns that better matched the actual distribution of land-cover classes. These results suggest that using a loss function aligned with the mathematical properties of compositional data leads to more credible and interpretable outputs.

The findings demonstrate that the combination of input features, model architecture, and loss function substantially impacts the quality of fractional land-cover estimation. Under our experimental conditions, the MLP model using Embedding V1 and the Aitchison distance achieved the highest accuracy, providing practical insights into the design of compositional land cover classification frameworks in remote sensing.

The Copernicus Data Space Ecosystem gives every user free access to Europe's Earth observation archive — petabytes of satellite imagery, derived products, and land monitoring datasets, all through cloud-native APIs. We showcase how sovereign EU infrastructure and open standards turn this archive into a foundation for geospatial AI.

This presentation introduces a development plan for a web-based map application for "Watashitachi no Hiroshima (Our Hiroshima)", a supplementary social studies textbook for regional studies that has been used in elementary schools in Hiroshima City for over 50 years.

This talk introduces recent development features in Reearth CMS, including an API-first architecture, an interactive API playground for exploring endpoints, and flexible import/export pipelines designed to improve developer experience and simplify integration with external systems.

Decision-makers invest in innovations and programs to create positive social and environmental outcomes. However, there is no consistent or standardized method to evaluate the process or impact of these decisions. A decision support tool provides access to machine learning (ML) features, generating insights revealing core predictors behind key outcomes.

Stop struggling to share QGIS maps. With Kumoy, you can push your QGIS projects into an interactive web map in just a few clicks. You can also work together on the same project, manage roles and permissions, and keep QGIS maps and data in sync.

Topology validation is an essential tool for GIS quality control; however, discrepancies frequently arise between ArcGIS and QGIS results. This presentation compares the outcomes of both tools, analyzes the practical issues stemming from these differences, and proposes potential solutions to ensure consistent data integrity.

Mapping and monitoring agricultural crop health, particularly for sugarcane, is essential for improving yield quality and ensuring sustainable crop management. Sugarcane plays a critical role in agricultural economies and bioenergy production in many tropical and subtropical countries, including Thailand. However, crop productivity and quality are strongly influenced by soil fertility and nutrient availability. Accurate and timely information on soil nutrients is therefore crucial for effective crop management and decision-making. Traditional soil analysis methods rely heavily on labor-intensive field sampling and laboratory testing, which are often time-consuming, costly, and spatially limited. As a result, these approaches are not always suitable for large-scale monitoring of soil conditions across extensive agricultural landscapes. Recent advances in Earth Observation (EO) technologies provide new opportunities to overcome these limitations. Satellite-based EO now offer high spatial, spectral, and temporal resolution data that can be used to monitor crop health and soil conditions at regional scales. In particular, hyperspectral satellite has shown strong potential for estimating soil properties because of its ability to capture detailed spectral signatures associated with soil composition, moisture content, and nutrient levels. Hyperspectral sensors offer continuous spectral information across hundreds of narrow bands, enabling the detection of subtle variations in soil characteristics that may not be captured by multispectral sensors. When combined with machine learning algorithms, hyperspectral data can be used to develop predictive models for estimating soil nutrients and other agronomic parameters at the field levels across large agricultural areas. Despite these technological advancements, several challenges remain for implementing satellite-based soil monitoring in Thailand. The northeastern region of the country is characterized by complex topography, heterogeneous land-use patterns, fragmented smallholder farms, and highly variable weather conditions. These factors complicate the accurate mapping of soil properties and crop conditions using remote sensing data. In addition, sugarcane cultivation in this region often occurs within relatively small and dispersed field plots, making it difficult to capture field-level variability using conventional monitoring approaches. Therefore, the integration of hyperspectral satellite data with advanced machine learning models offers a promising approach to improve soil nutrient estimation and support precision agriculture practices in the region. This study aims to develop predictive models for estimating key soil nutrients in sugarcane fields across Northeast Thailand using PRISMA hyperspectral imagery and the random forest (RF) algorithm. Specifically, the study focuses on estimating six important soil properties: soil pH, soil organic matter (SOM), electrical conductivity (EC), nitrogen (N), phosphorus (P), and potassium (K). These soil parameters are critical indicators of soil fertility and directly influence crop growth, nutrient uptake, and overall yield potential. The analysis was conducted using hyperspectral imagery (239 bands) acquired from the PRISMA satellite platform, which provides high-resolution spectral data suitable for environmental and agricultural monitoring applications. To support model development and validation, field data were collected from sugarcane-growing areas across the study region. A total of 46 soil sampling plots were established within representative sugarcane fields. Field surveys were conducted between 1st and 25th May 2025, during which soil samples were collected at an approximate depth of 20 cm. The collected soil samples were subsequently analyzed to determine the corresponding soil nutrient properties, including pH, SOM, EC, N, P, and K thought laboratory. These ground measurements served as reference data for training and validating the predictive models. In this study, a RF was implemented to estimate soil nutrients from hyperspectral satellite data. The RF algorithm is widely recognized for its robustness, ability to handle nonlinear relationships, and strong performance when working with high-dimensional datasets, like hyperspectral imagery. The model development process included feature extraction and optimization to identify the most relevant spectral variables associated with soil nutrient conditions. To evaluate model performance, the dataset in this study was divided into training and validation subsets. Specifically, 37 soil samples (80%) were used for training the model, while the remaining 9 samples (20%) were reserved for independent validation of the model results.
The experimental results demonstrated that the RF-based predictive models achieved varying levels of accuracy for different soil properties. The coefficient of determination (R²) values ranged from 0.30 to 0.80 across the estimated parameters. Higher predictive performance was generally observed for N and OM variables, which exhibit stronger spectral responses in hyperspectral imagery. In contrast, P and EC parameters showed relatively lower predictive accuracy due to weaker spectral signatures and potential environmental influences. Nevertheless, the resulting spatial distribution maps of soil fertility indicators exhibited consistent and meaningful spatial patterns when compared with the ground datasets. The generated maps provide valuable insights into the spatial variability of soil fertility conditions across sugarcane fields in Northeast Thailand. These spatial patterns can help farmers, agricultural planners, and policymakers identify areas with nutrient deficiencies or potential soil management issues. By integrating remote sensing-based monitoring with field observations, the developed approach offers a scalable framework for large-area soil nutrients assessment and crop monitoring. In addition, this work developed a spatial recommendation framework based on the derived nutrient maps and crop condition indicators. This framework aims to support precision agriculture practices by providing location-specific information for crop management. For example, farmers can use these spatial recommendations to optimize fertilizer application, adjust irrigation strategies, and improve overall crop management during different growth stages. The targeted interventions can help improve resource-use efficiency, reduce environmental impacts, and enhance crop productivity. Overall, this study demonstrates the potential of integrating PRISMA hyperspectral satellite data with the efficient RF model for monitoring soil nutrients at the field levels over the large-scale. The proposed approach provides a cost-effective and scalable alternative to traditional soil sampling methods, enabling continuous monitoring of soil fertility conditions at the field level. By supporting more informed and timely agricultural management decisions, this framework contributes to the advancement of precision agriculture and sustainable sugarcane production in Thailand. Ultimately, the proposed workflow can help improve crop health, optimize fertilizer management, and increase sugarcane yields across diverse agricultural landscapes.

Keywords: Sugarcane crop, random forest, PRISMA hyperspectral, soil nutrients, precision agriculture

OGC API and modern standards have made spatial data integration seamless — like a "super expressway" breaking traditional barriers. Yet this ease lures organizations into a "Convenience Trap," where unchecked connections breed API Sprawl, traffic bottlenecks, and untraceable security vulnerabilities.

pipGIS is a cloud‑native web GIS platform for municipalities and public agencies that need to manage spatial data, publish interactive maps and support everyday operational workflows. Built on an open‑source stack (PostGIS, GeoServer and modern web mapping libraries), pipGIS provides multi‑tenant deployments, configurable user roles and tools for data editing and quality control tailored to transport and land‑use datasets. The talk will briefly present the architecture of pipGIS and show how it supports use cases such as traffic counting, road asset inventories and urban planning analyses.

LocationMind, a University of Tokyo-originated startup founded in 2019, will present an overview of our business activities leveraging geospatial data and location intelligence, focusing primarily on our international expansion. We will also introduce the data and AI products developed through these initiatives, with a special emphasis on our utilization of FOSS4G.

Lunch sponsored by Eukarya/Re:Earth & Geo Technologies

Raster, vector, point cloud: every geospatial format solves the same core problems, including linearization, chunking, compression, and metadata. Let’s build a format from scratch to see these concerns in practice.

This talk addresses some common challenges users face, when consuming data from OGC API. It will map these issues to best practices, offering pointers for data consumers to improve their workflow and providing publishers with answers to common user complaints about the standard or its implementation.

This presentation introduces how GIS is currently used in agriculture in Hokkaido. We also present Hokuren’s initiatives and future plans for utilizing FOSS4G technologies, particularly QGIS, to support agricultural data management and farm operations.

We present Data4Land (https://doi.org/10.1016/j.softx.2025.102226), a flexible and reusable open-source workflow for semi-automatic enrichment of remote sensing products, such as land-use/land-cover (LULC) datasets, with other vector datasets of higher accuracy and consistency, such as OpenStreetMap (OSM). Originally designed for analyses of functional habitat connectivity, where features such as roads and railways dissect natural habitats, the workflow has applications in a range of environmental monitoring and assessment, for example predicting land conversion or analysing land surface temperatures. Data4Land is published on github along with a sample raster dataset and a set of suggested test values for ecological parameters. The sample dataset covers part of England and includes nine category labels representing LULC classes. Samples of input data and folder structure for output data are located in the data subrepository, and the detailed user guide is available in the documentation.
Users are not constrained to use the modular noteboooks for connectivity computation, but can run any of them in combination to create updated LULC datasets for multiple purposes. The input parameters, such as the width of buffer or decay rate, are flexible, and users should define them with reference to their own ecological framework or research question (e.g. species and their migration characteristics, stressors etc.).
Connectivity between natural habitats is an important parameter for mitigating the ‘demographic bottlenecks’ that affect populations’ survival in natural and semi-natural ecosystems, and wildlife corridors are widely recognised as an effective approach to ensure or restore movement of plants and animals between habitats. However, land-use/land-cover datasets produced from remote sensing may not consistently capture small/linear features of interest, including ecological barriers, eg. roads, railways and water objects. These features may boost or limit habitat connectivity (depending on the species being considered), while being too narrow to be detected through widely-used remote sensing products (for example, Landsat and Sentinel) due to relatively small spatial resolution (20–30 m) and shadows cast by vegetation. To overcome this challenge, the open-source Data4Land workflow was developed to enrich the LULC with vector data dynamically retrieved from open APIs (for example, OpenStreetMap or World Database on Protected Areas). Users apply Data4Land through a series of open-source Jupyter notebooks which can be flexibly configured to model different impacts and diffusion of impacts for specific feature classes. A particular strength of the workflow is the capacity to vary the distance at which landscape features such as roads and urban areas exert impact on the surrounding landscape. This allow the creation of suites of landscape maps which capture the sensitivities of different species groups to anthropogenic pressures such as noise, disturbance and pollution.
To illustrate the use of Data4Land outputs, we assess historical trends (1987-2022) and analyse connectivity dynamics for a range of threatened species with different ecological characteristics. Case studies for Catalonia and Northern England and for Albera Natural Park in the Pyrenees have been used to demonstrate the capabilities of the developed technical workflow at the regional and local extent respectively. The regional case studies focus mainly on forest- and shrubland-dwelling species (badger, beech marten, common genet, stoat, and hedgehog), The local case study considers Testudo hermanni (Hermann’s tortoise), which is is particularly vulnerable to habitat fragmentation due to its limited dispersal ability. Connectivity outputs for the local case study are validated with species occurrence records from GBIF, iNaturalist and the administration of Albera park.
Where vector data represented ecological barriers, connectivity indices tended to drop at all scales once the enriched LULC datasets were implemented. Edge effects of biodiversity stressors affected connectivity by up to 2.9-fold, while low spatial resolution underestimates the role of ‘stepping stones’’, and individual filtering of OpenStreetMap features was essential for datasets with spatial resolution > 30 m. Data4Land prevents the calculation of spurious ecological corridors between species’ habitats by applying ecological barriers not covered in input datasets. It results in lower, and more realistic, connectivity values and provide a more sophisticated representation of actual ecological corridors compared to connectivity computations based on non-enriched remote sensing LULC datasets.
This scalable workflow enables automated habitat connectivity assessments for regional and local spatial planning, biodiversity conservation, and ecosystem services evaluations. It demonstrates the interoperability of four open-source technical components and highlights the need for multiple connectivity indices, as their temporal trends may diverge and are not always intercorrelated.
The Data4Land tool is capable of providing users with more accurate reflections of land use and land cover not only for nature conservation studies but also in other fields. The connectivity case study illustrates the value of a transparent and repeatable workflow in which parameters such as dispersal distance and edge effects of stressor features can be systematically varied in order to consistently assess their effect. The developed workflow is flexible and scalable and would be useful to implement at regional and local scales by planning authorities, environmental consultants, and nature conservation experts as a part of environmental impact assessments. We believe that the Data4Land tool will be a useful contribution to analysis not only of habitat connectivity, but of a range of landscape analyses where documentation of provenance and process are important.

Caucaia's story reveals how transitioning from fragmented files to an open spatial database is transforming urban management. By adopting PostGIS and QGIS, we overcame historical challenges and are currently structuring our data architecture to achieve the interoperability required by the national SINTER framework by the end of 2026.

GeoSampa 2.0 modernizes São Paulo’s geospatial platform through the adoption of the OSGeo-based GIFramework Maps, combining technical evolution with international collaboration. The upgrade improves usability and scalability while preserving an open SDI, demonstrating how open-source technologies can deliver high-performance, interoperable solutions under real institutional constraints.

An idle-aware geospatial processing scheduler that auto-scales workers based on queued workloads. The system runs spatial processing tasks opportunistically using available resources and supports cloud-native deployment for scalable geospatial data processing pipelines.

Bare-earth Digital Elevation Models (DEMs) or Digital Terrain Models (DTMs) are fundamental to geospatial applications, from flood modelling and landslide assessment to infrastructure planning and environmental management. However, original publicly accessible global elevation products, such as SRTM (Shuttle Radar Topography Mission), ASTER GDEM (Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Model), and Copernicus DEM, represent Digital Surface Models (DSMs). DSMs are elevation models that include canopy heights and building structures, rather than true bare-earth topography. Their vertical accuracies (how closely elevations match the real ground height) typically range from 4–15 m RMSE (root mean square error), and their spatial resolutions (size of the smallest discernible detail) are constrained to 30 m or coarser. Airborne Light Detection and Ranging (LiDAR) derived DTMs achieve sub-meter vertical accuracy and spatial resolution because lasers/photons penetrate vegetation to measure ground elevation. However, high-cost LiDAR surveys lead to fragmented coverage, even in developed countries. For example, about 16% of New Zealand’s land surface still lacks airborne LiDAR mapping, resulting in critical data gaps in the very remote, rugged, and densely vegetated terrain where elevation information is most important.

Recent deep learning approaches have the potential to enhance the vertical accuracy and spatial resolution of global DEMs through super-resolution techniques. For example, JSPSR (Joint Spatial Propagation Super-Resolution) networks improve Copernicus GLO-30 DSM from 30 m to 3 m spatial resolution by utilising high-resolution remote sensing imagery, reducing elevation RMSE by over 70% across diverse sites. However, their performance drops in dense forest canopies. Optical sensors cannot penetrate vegetation, so models must infer ground elevation from indirect cues (such as estimating canopy height or shape). Spaceborne LiDAR missions, such as ICESat-2’s ATL08 product, provide global terrain measurements that can penetrate vegetation. However, incorporating these measurements poses two main challenges. First, after rigorous data quality filters, ATL08 photons can cover as little as 0.1–0.2% of dense-forest mountain areas. Second, spaceborne LiDAR provides sparse point data, whereas optical imagery and DEMs are dense raster grids, resulting in a data-geometry mismatch that most conventional network architectures cannot inherently accommodate.

This study describes an open-source deep neural network framework that tackles these two challenges using a triple-branch multi-modal fusion network. It processes three complementary data streams: high-resolution remote sensing imagery via a Swin Transformer encoder, interpolated Copernicus GLO-30 DEMs via a parallel Swin encoder, and serialised ATL08 along-track data via a sparse encoder that preserves measurement features without rasterisation or interpolation. All code and trained models will be released under an open-source license to support open-science principles.

The main innovation is a multi-scale deformable cross-attention mechanism that enables effective fusion of sparse LiDAR measurements with dense raster data. At each scale, individual ATL08 photons independently query the surrounding image and DEM features through learned deformable sampling patterns, allowing each photon to adaptively sample contextual information most relevant to elevation prediction at its location. Inspired by deformable DETR, the deformable cross-attention mechanism implements bidirectional information flow: image and DEM features inform the interpretation of photon measurements during downscaling, while photon-derived elevation features are injected back into the feature maps during upscaling. This design provides that sparse yet accurate LiDAR measurements guide feature extraction, while dense image context enriches photon representations, addressing the key challenge of multi-modal feature fusion in open geospatial data science.

Ultimately, the Spatial Propagation Network (SPN) transforms sparse DSM grids into dense predictions by conditioning content-adaptive kernels on fused multi-modal features. Through multiple propagation iterations, corrections propagate along paths guided by image content, following ridgelines, thalwegs, or areas with similar vegetation characteristics, while re-injecting precise photon measurements at each iteration to retain accuracy at known locations.

Sparse-to-dense progressive supervision computes elevation loss only at ATL08-confirmed locations (typically fewer than 64 per 256×256 training patch), whereas multi-scale deep supervision heads propagate gradient information throughout the network, even in areas without direct elevation constraints. This strategy prevents learning invalid correlations in unobserved areas while retaining end-to-end differentiability.

We evaluate our approach using the DFC30 dataset, which we augment with spatially matching ATL08 measurements. The training set contains 12,728 image-DSM-photon tuples, and the test set contains 3,196, both with airborne LiDAR ground truth. Our results show considerable improvements versus the baseline method, JSPSR. Across all test sites at 3 m spatial resolution, elevation accuracy (RMSE) improves by 8% in open terrain (from 1.1 m to 1.01 m) and 28% in dense forest canopies (from 6.7 m to 4.8 m). Overall, we achieve a vertical accuracy of 3.2 m for all vegetation classes. The largest improvements occur where optical-only methods perform weakest. In indigenous forests with dense understory and complex terrain, our method decreases systematic bias from 15.1 m to 8.8 m.

All code, trained weights, and preprocessing tools will be public under an open-source license. This ensures complete repeatability and allows community adaptation. The trained model enables end-to-end 3 m DTM generation for unmapped areas of New Zealand using global DEMs. It will produce a seamless bare-earth elevation product at a national scale, covering about 268,000 square kilometres. This supports applications such as landslide mapping, hydrological analysis for freshwater management, carbon stock assessment in forests, and infrastructure planning in rural areas.

For the FOSS4G community, this work makes three key contributions. First, it delivers a practical, open-source solution for generating high-resolution bare-earth DTMs by fusing global datasets (Copernicus DEM, remote sensing imagery, and ICESat-2 ATL08) using reproducible methods. Second, it provides an architectural strategy that respects the geometry of different data modalities, e.g., constructed tables, sparse point clouds and dense rasters, by using a template for multi-modal fusion in open geospatial science. Third, it shows that open data and software can solve real-world data gaps to produce operational products that benefit communities, environmental management, and disaster resilience. More broadly, our work supports ongoing efforts in the open geospatial community to improve global terrain characterisation by fusing diverse Earth observation assets, making high-quality elevation data accessible to all.

Orbitalnet discusses its journey toward realizing geospatial-Intelligence and what lies ahead.

This talk presents a reusable geospatial platform architecture for operational use in the Global South, and explains how ArkEdge Space Inc. uses MapLibre GL JS, STAC, and TiTiler to build adaptable applications across domains, using an agricultural project with the Paraguayan Space Agency as a case study.

Auspatious designs and delivers cloud-native geospatial solutions that turn complex data into useful, usable products.

We work with governments, research organisations, and mission-driven partners to:

• Build robust, open, and reproducible geospatial platforms
• Lower barriers to accessing and analysing Earth observation data
• Apply open standards, open data, and open source software to problems
• Share knowledge, grow capability, and strengthen the global EO community

This short 5 minute talk will cover the AWS Open Data program, which consists of the open sourced Registry of Open Data (https://registry.opendata.aws/) and the Open Data Sponsorship program.

This presentation will introduce Asia Air Survey's work in remote sensing, geospatial analysis, and consulting, with a focus on disaster risk reduction, infrastructure management, environmental conservation, and sustainable development. Through selected case studies and project experience from Japan and overseas, we will showcase how aerial surveying, LiDAR, satellite imagery, UAVs, AI, GIS, and digital twin technologies are applied to real-world challenges. Our booth will feature examples of these technologies and solutions, demonstrating how an integrated geospatial approach can support better planning, decision-making, and long-term resilience.

As AI agents and LLM-based systems become primary consumers of geospatial data, the weak point in most data infrastructure is no longer access — it's meaning. Open data alone doesn't guarantee interoperability; without a shared semantic layer, every new API becomes another silo an AI agent has to be taught to navigate. This talk shares Geolonia's experience building GeonicDB, a context data platform built on the NGSI-LD standard — including why we chose not to simply adopt an existing broker like FIWARE Orion-LD, and built our own instead. We'll cover the reasoning, the trade-offs, and a live demo of AI agents querying standardized context data directly.

Every day we hear about floods, heat waves, and climate change.

Yet most people still cannot answer a simple question:

"What is happening where I live?"

Over the past several years, I have worked on public-sector GIS projects in Japan, designing products for disaster preparedness, urban planning, environmental visualization, and citizen participation.

What surprised me most was that people were rarely asking for GIS itself. They were asking for a better way to understand their own communities.

This talk shares how these projects gradually changed my perspective as a product designer. Rather than focusing only on maps, layers, or dashboards, I will explore how open geospatial technology can help people observe environmental change, understand local context, make informed decisions, and participate in shaping the future of the places they live.

I believe the next challenge for the FOSS4G community is not only making geospatial data open—but making spatial understanding accessible to everyone.

"Machi-Space" empowers diverse stakeholders to perform environmental simulations and share data effortlessly, lowering barriers to participation in urban development and providing a collaborative platform for evidence-based decision-making.

The global push for Net Zero has turned our eyes toward the horizon—specifically, the vast, untapped energy of our oceans. This session explores using FOSS4G in building a comprehensive ocean renewable energy platform.

What if data formats didn't need their own libraries? The cylf ecosystem leverages WebAssembly to make codecs, format drivers, and storage drivers modular, sandboxed, and fetchable on-demand. Each can be developed independently and on its own lifecycle. We'll demo a working proof of concept.

Floods are becoming more frequent and severe worldwide due to climate change. These disasters cause significant human and economic losses, making the development of rapid and highly accurate flood mapping technologies essential for clearly identifying damage conditions and supporting subsequent rescue activities.

Remote sensing with openly available Synthetic Aperture Radar (SAR) plays an essential role in disaster monitoring. As an active sensor that transmits its own microwave signals, SAR can acquire observations regardless of weather or time of day, enabling reliable monitoring even in extreme conditions.

However, conventional SAR-based flood detection methods face several challenges. Threshold-based approaches applied to a single SAR image struggle to distinguish permanent water bodies from newly inundated areas. To address this, methods that compare pre-flood and flood-time images have been developed. Yet openly available data such as Sentinel-1 have satellite revisit cycles that create temporal gaps of several days to weeks between compared images. During this interval, seasonal vegetation changes and speckle noise differences are frequently misidentified as flood-induced changes, leading to increased false positives. To address these false positives caused by temporal misalignment, this study proposes a flood detection method that synchronizes the temporal axes of the compared images.

In our experiment, we used the openly available "UrbanSARFloods" dataset. This dataset was created for global-scale flood detection and covers 18 flood events from five continents, including flood and non-flood label data. Each image chip has a spatial resolution of 10 m and a size of 512 × 512 pixels, and data from a selected 628 area of interest were used in this study.

We used Sentinel-1 Ground Range Detected (GRD) data as our primary flood detection source. Sentinel-1 GRD is a multi-looked SAR product projected to ground range using an Earth ellipsoid model. In Interferometric Wide (IW) mode, it provides roughly 10-meter resolution with dual-polarization (VV and VH) across a 250 km swath. Sentinel-1 acquires observations regardless of weather or time of day, with a 6–12 day revisit cycle. We collected SAR time series from Google Earth Engine (GEE), covering one year prior to each flood event up to 40 images per location. These data include two polarization modes: VV and VH. To ensure consistency, only data acquired from the same orbit (ascending or descending) as the flood-time observation were used.

GRD data contain inherent speckle noise that requires filtering. To improve data quality, we reduced speckle noise using a median filter (3 × 3) and performed normalization using Min–Max scaling. For datasets with missing time-series images, we interpolated missing data using three-dimensional spline interpolation.

The revisit gap can cause temporal misalignment between compared images. To obtain an image representing non-flood conditions, we simulated imagery from past SAR GRD sequences at the timing of the flood observation. A three-layer ConvLSTM model captured spatio-temporal dependencies and generated predicted "non-flood" SAR images. The model used a hybrid Mean Absolute Error (MAE)-Structural Similarity (SSIM) loss function to preserve edge details and suppress speckle noise. This function prioritizes structural and statistical similarity over standard MSE.

We created multiple image chip pairs by using six consecutive time steps as one set and sliding the window forward by one time step. Each image chip pair consists of five continuous input images and one subsequent image as reference. The model training was performed by minimizing the error between the predicted image generated from the five input images and the reference image. A dedicated ConvLSTM model was built for each location to learn its unique surface characteristics. After training, we input the five most recent image chips into the model to generate a predicted image for the next time step.

Finally, we performed flood detection using a Siamese Network to compare the simulated images with the observed images acquired during flooding. This CNN-based model uses an encoder–decoder architecture with skip connections, extracting high-dimensional features while preserving spatial boundary information. After computing pixel-wise differences between the feature maps, we applied a sigmoid function to produce a flood probability map. The final flooded areas were detected through thresholding (0.5).

Quantitative evaluation using a pre-split test dataset consisting of 30 samples demonstrated considerable performance improvements. The proposed method achieved an F1 score of 0.605 (Precision: 0.556, Recall: 0.661), representing a 42% improvement over the conventional pre-/post-flood comparison method, which achieved only 0.427 (Precision: 0.328, Recall: 0.610). This improvement was primarily attributed to the substantial reduction in false positives caused by temporal misalignment.

The precision improvement from 0.328 to 0.556 indicates that the proposed temporal synchronization approach effectively suppresses misdetections caused by seasonal vegetation changes and speckle noise variations. Meanwhile, the recall score of 0.661 demonstrates the method's capability to detect actual flood events without sacrificing sensitivity.

Qualitative evaluation further confirmed the effectiveness of the proposed approach. Visual inspection revealed that the method successfully suppresses speckle-like noise patterns that frequently lead to false positives in conventional methods. The approach accurately detected both large-scale inundated areas, such as flooded river plains and urban zones, as well as small, isolated flooded regions that are typically difficult to identify. The generated flood probability maps showed clear spatial boundaries between flooded and non-flooded areas, enabling more reliable damage assessment.

While we demonstrated the usefulness of the flood detection method, several challenges were identified. In regions with complex topography, frequent misdetections occurred due to terrain-induced radar shadow, geometric distortions (layover), and changes in vegetation's dielectric properties. In urban areas, flood detection often misses actual flooding because double-bounce scattering reduces the backscatter intensity differences between images. Future work will consider incorporating Digital Elevation Model (DEM) data and coherence data, which quantify the phase correlation of microwave signals between two temporal images, to account for physical terrain constraints and further suppress misdetections in urban areas, enabling more effective flood detection.

In conclusion, this study proposed a flood detection method that eliminates temporal lag between compared datasets and demonstrated improved performance in detecting flood-induced changes. By enabling near-real-time flood detection under extreme weather conditions, this method is expected to strengthen societal response capabilities to flood disasters.

Introduction  
Remote sensing-based spectral diversity has emerged as a scalable proxy for field-measured biodiversity [1-4], grounded in the Spectral Variation Hypothesis, which posits that greater spectral heterogeneity in remotely sensed imagery reflects greater ecological diversity on the ground [5]. While studies have examined methodological choices known to influence spectral diversity mapping performance [6,7], the effect of input feature type, specifically the choice between spectral bands, Principal Component Analysis (PCA) transformed spectral bands, and spectral indices (SI), has not been empirically tested. The biodivMapR package [8] implements the spectral species concept for biodiversity mapping and acknowledges both PCA-transformed bands and SI as valid input options, but provides no firm recommendation, explicitly calling for users to test and compare approaches against ground observations [9]. Within the FOSS4G ecosystem, reproducible biodiversity monitoring workflows based entirely on open data and open-source software remain under-evaluated, particularly with respect to methodological parameterisation choices. To address this gap, we compare three spectral diversity computation approaches across three ecologically distinct hemi-boreal forest sites in  Estonia and validate them against field-measured tree species diversity. We aim to address two research questions: (1) Does PCA-based dimensionality reduction produce stronger correlations with field-measured tree species diversity than non-PCA spectral bands or SI used as biodivMapR inputs? (2) Are results consistent across ecologically distinct hemi-boreal forest sites? 
Methods  
We selected three study sites representing characteristic hemi-boreal landscape types in Estonia: Otepää, a glacial moraine upland characterised by mixed nemoral forests dominated by spruce (n = 3,758 plots); Soomaa, a peatland and floodplain forest complex dominated by pine, birch, and alder with extensive bog-forest mosaics (n = 10,259); and Rapla, a lowland managed forest mosaic representative of intensively managed hemi-boreal production forests (n = 4,322). We filtered forest inventory validation plots using a morphological erosion criterion, retaining only plots with a minimum residual area of 900 m² after applying a 30-metre inward buffer, to exclude geometrically irregular or insufficiently large plots where moving-window diversity estimates would be strongly influenced by non-forest edge pixels. We pre-processed the Sentinel-2 imagery from the Copernicus Open Access Hub using open-source geospatial libraries in Python. We masked forests prior to spectral diversity computation, ensuring analysis was restricted to forested areas across all sites.  
We compared three approaches: (1) PCA applied to ten Sentinel-2 bands, with the first three principal components used as input features for Shannon spectral diversity mapping; (2) bands without dimensionality reduction (non-PCA); and (3) a set of three ecologically relevant SI - Normalised Difference Vegetation Index (NDVI), Normalised Difference Water Index (NDWI), and Leaf Area Index proxy derived from Soil-Adjusted Vegetation Index (LAI-SAVI), derived from the same Sentinel-2 bands. Shannon alpha-diversity was computed using a moving window of three pixels (30 m × 30 m) across tiled processing grids, with mosaicked outputs validated against field-measured tree species Shannon diversity using Spearman rank correlation. The effect of inter-quartile range (IQR) outlier filtering, an internal biodivMapR preprocessing step that masks pixels whose spectral values fall outside Q1 - 4 × IQR or Q3 + 4 × IQR, was additionally assessed by comparing results with and without this filtering applied. The entire workflow was implemented using open Sentinel-2 imagery and the open-source R package biodivMapR, ensuring full methodological transparency and reproducibility within the FOSS4G ecosystem. 
Results  
PCA-based spectral diversity consistently achieved the highest correlation with field-measured tree species diversity across all three sites, outperforming both the SI and spectral band approaches. The performance advantage of PCA was modest but consistent across sites. SI outperformed spectral bands at two of three sites. IQR filtering produced no consistent improvement across sites or methods. Pixel-level analysis confirmed that IQR masked fewer than 1.4% of pixels in the pre-masked forest-only imagery, regardless of site or method. 
Discussion 
Results indicate a moderate relationship between spectral and field-measured diversity across all methods and sites. While remote sensing-derived spectral diversity cannot serve as a standalone biodiversity indicator at this spatial resolution [1,4], it provides a reproducible, scalable, and open-source means of monitoring relative diversity patterns across landscapes. 
The modest but consistent advantage of PCA over spectral indices reflects its ability to condense spectral variance across correlated bands into uncorrelated components, reducing noise while retaining biodiversity-relevant signal. This comes at the cost of additional processing overhead and user-guided component selection, which limits full workflow automation. SI, by contrast, are theoretically grounded, computationally efficient, and directly interpretable, making them a practical alternative for large-area applications where automation and transferability are priorities. Spectral bands alone are not recommended due to band collinearity, increased spectral noise. The negligible effect of IQR filtering suggests that, in pre-masked forest-only imagery, this step may not be necessary. The ecological significance of the pixels removed warrants further attention.  
Conclusions  
We evaluated three input feature approaches for spectral diversity mapping in hemi-boreal forests using Sentinel-2 and biodivMapR, validated against field-measured tree species diversity across three ecologically distinct sites. PCA consistently outperformed spectral bands and SI, though the advantage was modest. SI represent a practical open-source alternative for large-area applications where computational efficiency is a priority. We identify integration of complementary environmental predictors is identified as a priority for future work, as we expect it to strengthen this relationship. Because the workflow relies exclusively on open data and tools, it transfers directly to other forest ecosystems without licensing constraints. All scripts and parameter configurations will be made openly available to support reproducible biodiversity monitoring. 
References 
[1] Fassnacht, F.E. et al., 2022. About the link between biodiversity and spectral variation. Appl. Veg. Sci. 25, e12643.
[2] Rocchini, D. et al., 2021. rasterdiv: An Information Theory tailored R package for measuring ecosystem heterogeneity from space. Methods Ecol. Evol. 12, 1093–1102.
[3] Torresani, M. et al., 2024. Reviewing the Spectral Variation Hypothesis: Twenty years in the tumultuous sea of biodiversity estimation by remote sensing. Ecol. Inform. 82, 102702.
[4] Wang, R., Gamon, J.A., 2019. Remote sensing of terrestrial plant biodiversity. Remote Sens. Environ. 231, 111218.
[5] Palmer, M.W. et al., 2002. Quantitative tools for perfecting species lists. Environmetrics 13, 121–137.
[6] Robertson, K.M. et al., 2023. Effects of Spatial Resolution, Mapping Window Size, and Spectral Species Clustering on Remote Sensing of Plant Beta Diversity. J. Geophys. Res. Biogeosciences 128, e2022JG007350.
[7] Schmidtlein, S., Fassnacht, F.E., 2017. The spectral variability hypothesis does not hold across landscapes. Remote Sens. Environ. 192, 114–125.
[8] Féret, J., De Boissieu, F., 2020. biodivMapR: An r package for α‐ and β‐diversity mapping using remotely sensed images. Methods Ecol. Evol. 11, 64–70.
[9] Féret, J., De Boissieu, F. (n.d.). biodivMapR: Produce diversity maps from optical images. https://jbferet.github.io/biodivMapR/articles/biodivMapR_02.html

GeoSolutions has been involved in several projects, from local administrations to global institutions, involving GeoNode deployments, customizations and enhancements. A gallery of projects and use cases will showcase GeoNode's versatility and effectiveness, both as a standalone application and service component, for building secured geodata catalogs and web mapping services.

Step into the world of mapping superheroes! The Global Open Mapping Guru Network empowers digital volunteers to mentor, validate, and lead open mapping initiatives worldwide. Learn how this global movement combines skills, collaboration, and impact, leveling up the future of open mapping while driving real-world change in communities everywhere.

This study analyzes multi decadal spatiotemporal land cover change in Dammam, Saudi Arabia (1986–2025) to establish a green infrastructure baseline. An open source data, Landsat and Sentinel-2 data classified using Random Forest in QGIS reveal urban expansion patterns and support greening priority planning using NDVI and NDBI.

From our production “KnoWaterLeak” leakage-risk service, we show real-time, on-demand vector tiles for large pipeline datasets. We cover in-database MVT generation, safe DSL-to-JSON translation, routing between tile and aggregation endpoints, and data-model choices that keep latency low and database load predictable.

Waystones is an open-source tool for designing geospatial data models and deploying production-ready OGC APIs. By automating the configuration of pygeoapi and QGIS Server, it enables organizations to meet High-Value Dataset (HVD) requirements and share standards-compliant data directly to the cloud.

In recent years, the widespread use of mobile devices equipped with LiDAR sensors has made it possible to easily capture 3D point clouds. However, these devices suffer from cumulative self-localization errors, limiting accurate 3D measurements to relatively small areas. Existing approaches to wide-area georeferencing of mobile LiDAR point clouds typically rely on GNSS receivers, direct 3D point-cloud matching, or alignment to pre-existing high-precision urban 3D models. These approaches can be costly, computationally demanding, or impractical where such reference data are unavailable.

Therefore, this study proposes a low-cost method for constructing a high-precision wide-area 3D map using only LiDAR-equipped mobile devices and open data. It is intended to help municipalities and citizen communities without expensive equipment or professional surveying skills develop local digital twin infrastructure. The resulting 3D maps can support 3D visualization of potential disaster impacts and community disaster preparedness.

The proposed processing pipeline accepts 3D point clouds from mobile devices and Japanese governmental open geospatial data as input and outputs open point-cloud formats such as LAS/LAZ for seamless use in FOSS4G software such as QGIS. The workflow consists of two stages: registration of adjacent point clouds and georeferencing of the integrated point cloud within a global geodetic coordinate system. Rather than directly estimating rigid transformations in full 3D space, the proposed method separates spatial information into horizontal and vertical components and optimizes them sequentially, improving registration stability for point clouds acquired by mobile devices. Geographic coordinates are assigned through image matching against open geospatial data instead of GNSS-based positioning. All processing, except point cloud acquisition, is implemented using open-source Python libraries widely used in the FOSS4G ecosystem. To support reproducibility, we will release the source code, processing workflow, example configuration files, and links to the open input datasets used in the experiment.

First, rough overlaps are extracted from the approximate positional information of each point cloud. CSF is then applied to separate each point cloud into ground and non-ground points. Height-based slices of the non-ground points are projected into bird’s-eye-view images representing wall surfaces. ORB feature-based image matching in OpenCV provides initial alignment, followed by horizontal refinement on the XY plane using the Point-to-Point ICP algorithm in Open3D. Finally, vertical and tilt errors are corrected using the ground points.

Next, absolute coordinates are assigned to the registered point clouds using Fundamental Geospatial Data provided by the Geospatial Information Authority of Japan. Although validation is performed with Japanese open data, the method is applicable in other regions where road edge vector data and digital elevation models (DEMs) are openly available. For horizontal georeferencing, road edge data is used. Using geopandas and shapely, a rough region around the point cloud is extracted and converted into a road edge image. The integrated point cloud is processed in the same manner to generate a wall surface image. OpenCV’s normalized cross-correlation (NCC) is then used to estimate the optimal translation and rotation parameters. For vertical georeferencing, a 5 m mesh DEM is referenced, and elevations are extracted with rasterio. A correction surface is generated by smoothing the elevation differences between the ground points and the DEM, and the resulting correction values are applied to the entire point cloud. This aligns absolute elevation with the DEM while preserving local terrain variations.

To verify the effectiveness of the proposed method, an experiment was conducted in the Jinaimachi district of Tondabayashi City, Osaka Prefecture. Measurements were conducted using an iPad Pro (11-inch, 4th generation), and the Scaniverse application. Five scan datasets were obtained from these measurements. In addition, ground truth data were prepared using measurements from a high-performance 3D scanner (Matterport Pro3) and a GNSS receiver (Drogger RZX.D). As evaluation metrics, both the RMSE between feature points of adjacent point clouds after registration (relative accuracy) and the RMSE between the final constructed 3D map and the ground truth data (absolute accuracy) were evaluated. The target accuracy was defined as an RMSE of within 0.1 m for relative accuracy. For absolute accuracy, the horizontal accuracy was required to be within 1.75 m, corresponding to “Map Information Level 2500” and suitable as a base map for hazard mapping, while the vertical accuracy was required to be within 0.30 m, suitable for flood simulation.

As a result of the evaluation, the average relative accuracy achieved an RMSE of 0.048 m, sufficiently satisfying the target value. Regarding absolute accuracy, the horizontal RMSE was 0.63 m and the vertical RMSE was 0.09 m, demonstrating favorable results. The results satisfy the requirements for Map Information Level 2500 horizontally and Level 500 vertically.

For the FOSS4G community, the proposed pipeline demonstrates how mobile LiDAR, governmental open data, and open-source geospatial libraries can be combined into a reproducible workflow for practical 3D hazard mapping. It also aligns with the conference’s emphasis on Asian geospatial initiatives by demonstrating a reproducible workflow based on Japanese governmental open data. Experimental results confirmed that the proposed method provides sufficient accuracy for 3D hazard maps that enable intuitive visualization of flood depths and landslide-affected areas. This method facilitates 3D data development in municipalities and citizen communities with limited budgets and can contribute to regional digital transformation. Future work includes improving robustness in wider and more diverse environments and developing a web system that integrates data from multiple devices to generate 3D maps automatically. Unlike workflows that depend on GNSS receivers or pre-existing high-precision 3D city models, the proposed method enables georeferenced wide-area 3D mapping using only mobile LiDAR, open geospatial data, and an open-source processing stack.

This talk presents lessons from QGIS workshops, highlighting hands-on learning, open data and adaptive teaching to build user skills from beginners to advanced practitioners.

1-Introduction and Study Area
Cladophora is a filamentous green alga native to the North American Great Lakes. Its excessive proliferation not only causes foul odors and impairs public beach recreation but also triggers severe ecological issues, including avian botulism outbreaks. Since the 1990s, the filtering effect of invasive species such as dreissenid mussels has significantly increased water clarity, allowing sunlight to penetrate to greater depths. This has led to massive Cladophora blooms even under relatively low nutrient concentrations. The study area of this research focuses on the nearshore waters along the southern shore of Lake Ontario (the United States side). To achieve precise calibration of remote sensing observations, the spatial scope of the study is strictly defined as two independent 6 km × 6 km square regions, centered respectively around two key hydrological and biological monitoring stations established by the United States Geological Survey (USGS): the OIR station (Irondequoit, near Rochester) and the OOL station (Olcott).
These two core USGS stations provide substantial, highly valuable ground-truth data for this study. These comprehensive datasets encompass multi-depth water flow velocities, water turbidity, and various critical chemical constituents in the water column (such as nutrient concentrations). More importantly, the stations provide net weight data of Cladophora samples collected in situ across different depth gradients. These multi-dimensional, high-precision ground truth indicators not only serve as an irreplaceable validation foundation for evaluating and calibrating various spectral remote sensing indices within our open-source computational architecture, but also enable us to deeply investigate the complex mechanisms underlying the relationships between micro-environmental physicochemical variables and nearshore benthic algal outbreaks.

2-Evaluation of Traditional Indices and Experimental Derivation of a Novel Index
In the preliminary remote sensing analysis phase, we developed a Python-based workflow to extract Sentinel-2 image bands and automatically calculated various traditional spectral indices, including NDVI, FAI, NDAVI, and SABI. Statistical analysis of multi-temporal imagery (from May to August 2023) revealed that the mean and median values of these indices were frequently negative or extremely low, accompanied by disproportionately large standard deviations. For instance, across multiple summer observation dates, the median values for NDVI and FAI consistently hovered near zero (ranging from -0.012 to 0.025). At the same time, NDAVI and SABI exhibited even deeper negative medians (often between -0.05 and -0.09). Furthermore, the high standard deviations—frequently exceeding 0.25 for NDVI and 0.50 for SABI—demonstrated massive signal noise. This statistical analysis demonstrates that vegetation indices based on the Near-Infrared (NIR) band exhibit severe absorption failures in aquatic environments, rendering them inadequate for precise mapping of submerged benthic Cladophora.
To address this optical challenge and identify the optimal spectral response, we designed a controlled physical experiment. A 3m × 3m water tank was used, with an incandescent light source simulating solar irradiance. A receiver simulated the satellite sensor to capture reflectance from a green surrogate representing benthic algae. Strikingly, the experimental results revealed that the strongest reflectance signals emerged in the Blue and Short-Wave Infrared (SWIR) bands, significantly diverging from the band selections of traditional vegetation indices. Based on these empirical findings, we are currently conducting rigorous mathematical derivations utilizing the Blue and SWIR bands to formulate a novel, water-penetrating spectral index specifically optimized for Cladophora detection.

3-Automated Open-Source Cloud-Masking Algorithm to Bypass API Limitations
To achieve high-frequency monitoring of Cladophora, we aimed to build a fully open-source, automated data acquisition architecture. However, querying the Copernicus Data Space API inevitably encounters strict request frequency limits and download volume quotas. Furthermore, the official API only provides the average cloud cover percentage at the full-scene level. For our small 6 km × 6 km Region of Interest (ROI), this macroscopic cloud assessment is highly inaccurate. A scene with a low average cloud percentage might still have dense clouds completely obscuring our study area, leading to massive invalid downloads and wasted bandwidth. Additionally, a single remote sensing image rarely covers the target area perfectly without clouds, necessitating the seamless mosaicking of multiple images and stricter screening for high-quality data.
To overcome this core bottleneck, we designed and implemented a regional cloud-masking algorithm based on image Quicklooks (previews) within our workflow. Since Quicklook files are extremely small and consume negligible download bandwidth, the program automatically prioritizes retrieving them. Given that Quicklooks do not inherently contain geographic coordinates, the algorithm first extracts the boundary coordinates of the scene's footprint polygon from the metadata. Subsequently, it correlates and standardizes the ROI's geographic coordinates against this footprint boundary. Based on this geometric translation, the system can precisely reverse-engineer the specific pixel rectangle corresponding to the study area on the unreferenced Quicklook image. Ultimately, the algorithm computes the proportion of white pixels exclusively within this localized bounding box to accurately assess the true cloud cover rate within the ROI. Only when the ROI's cloud cover meets strict clear-sky thresholds does the system automatically trigger the API to download the heavy, high-resolution original imagery. This algorithm successfully achieves precise "on-demand downloading," effectively circumventing API bandwidth restrictions while dramatically improving the efficiency of acquiring the cloud-free data required for subsequent image mosaicking.

4- Conclusion and Future Works
This study successfully established a highly efficient, Python-based open-source remote sensing download architecture that practically circumvents API limitations. It also highlighted the severe shortcomings of traditional vegetation indices through both satellite data statistics and controlled physical experiments. Future research will focus on advancing two primary tasks:
First, further refining the Quicklook-based cloud-masking algorithm to automate the acquisition of extensive multi-temporal imagery for seamless spatial mosaicking. To ensure complete reproducibility, this process will be integrated into an end-to-end Python pipeline, with the full source code made freely available on GitHub.
Second, finalizing the mathematical formulation of our novel Blue-SWIR spectral index based on the water tank experiment, and deploying it within our open-source pipeline to precisely map the spatial distribution and evolutionary dynamics of Cladophora during peak summer blooms.

References:
[1] Howell, E. T. (2018). A decadal-scale perspective on the occurrence of Cladophora on the north shore of Lake Ontario. Environmental Monitoring and Assessment.
[2] Wright, N., et al. (2024). CloudS2Mask: A novel deep learning approach for improved cloud and cloud shadow masking in Sentinel-2 imagery. Remote Sensing of Environment, 306, 114122.
[3] Copernicus Data Space Ecosystem. (2024). Quotas and Limitations Documentation.

This presentation covers our experience with the migration of GEODES (the Earth observation catalog of the French space agency, CNES) from proprietary systems to an Open Source, cloud-optimized architecture, highlighting the strategic 'how' and 'why' behind the shift.

—An Observation-Aware, Offline-First System for Repeatable Field-to-Tree Production—

This talk presents MIMAR, an observation-aware, offline-first production architecture built on OpenDroneMap and OpenSfM. It connects controlled UAV acquisition, high-recall reconstruction in repetitive canopies, GPU-accelerated matching and bundle adjustment, recoverable local execution, spatial quality gates, and decision-ready tree records.

The focus is not a faster component, but a repeatable field-to-tree system whose performance is accepted only when matching recall, reproducibility, mission geometry, and operational economics survive together.

Rekichizu is a web service for exploring historical maps of Japan in a modern digital map style. Through data creation with QGIS and collaborative open data production with CODH, it makes historical landscapes accessible to everyone, preserving cultural memory through open-source technology.

pygeoapi is one of the most popular open source solutions for deploying OGC compliant geospatial APIs. This session will explain best practices for deploying pygeoapi and scaling it within a cloud native, containerized, and horizontally scalable deployment.

STAC has evolved into core geospatial infrastructure. This talk covers its current state, real-world adoption, key challenges, and how it fits into emerging architectures for scalable discovery and analysis.

WoSIS is a global soil information service that safeguards, standardises and shares soil profile data from contributors worldwide. Built on PostgreSQL/PostGIS, GraphQL and OGC services, it covers the full cycle from ingestion to dissemination. This presentation discusses the architecture, soil data workflows, and latest developments.

iRIC is a free and open-source software platform for river flow, flood inundation, sediment transport, and rainfall-runoff simulation. It provides a common graphical interface for a wide range of numerical models developed by researchers and engineers.

This sponsor session briefly introduces the iRIC platform, its international user and developer community, and recent developments in river and flood simulation. We also invite researchers, engineers, developers, educators, and organizations to visit our booth and explore opportunities for collaboration.

Data gets collected in the field, processed on the desktop, and consumed by people who never open a GIS. OPENGIS.ch works across that whole path. We build QField for field work, we're the second biggest contributor to QGIS core, and we've just launched Georama (georama.io) for publishing maps and geospatial stories on the web. Eight minutes on how the pieces connect, and how the commercial side pays for the open-source side.

Large companies often have established processes, dedicated roles, and structured ways of working. Small open source companies operate differently, with deep expertise, flexibility, and a close connection to their communities and customers. After more than 12 years at HERE Technologies, I spent 5 months at GeoCat exploring this other side of the industry. In this session, I’ll share what I learned from working with a small, open source-focused team: how priorities are set, how community contributions and commercial needs are balanced, what challenged my assumptions, and why some lessons from open source teams can be valuable for organizations of all sizes.

Eliminate idle costs with a scale-to-zero, cloud-native GIS architecture. This session explores high-performance analytics using DuckDB and serverless mapping with PMTiles to remove dedicated servers. Learn to build sustainable, fast GIS applications across any cloud provider while paying only for active usage.

TBD

Seagrasses are flowering plants (Angiosperms) that have secondarily colonised marine environments, analogous to the evolutionary return of whales to the sea. Seagrass ecosystems provide essential functions and ecosystem services to coastal areas by introducing structural complexity through plant tissue in intertidal zones. These services include habitats for small intertidal organisms, nursery and feeding grounds for fishery species such as squid spawning aggregations, shoreline protection through sediment stabilization, and water purification.

Recently, seagrasses have been gaining broad attention as a major component of blue carbon. In Japan in particular—where the world's first voluntary blue carbon credits targeting seagrass beds and macroalgal beds were issued—private-sector restoration activities have been increasing (Yamakita, 2025; Kuwae et al., 2026). Yet many seagrass beds worldwide are in decline or unknown status (Waycott et al., 2009; McKenzie et al., 2020). Adaptive management of these ecosystems requires the ability to detect both long-term trends (Yamakita et al., 2011) and abrupt collapses across spatial and temporal scales relevant to policy and local management.

Continuous monitoring of seagrass faces practical limitations. High-resolution commercial imagery and intensive field surveys provide high accuracy, but they tend to be prohibitively expensive, particularly when long-term time-series data or extensive coastal areas are involved (Duffy et al., 2025). On the other hand, combining historical aerial photographs and recently available low-cost or free satellite data with open-source software (OSS) has the potential to enable accessible monitoring by local citizens, experts from other fields, or even countries and organizations with limited budgets. However, this approach remains limited to case studies and has not yet been utilized as an integrated monitoring method.

This study analyses the Ako tidal flat in the Seto Inland Sea, Japan, where nearly all Zostera marina disappeared within a single year in 2025. Using aerial photographs from the 1940s onward, high-resolution satellite imagery, GRUS images (2.5–5 m), and monthly Sentinel‑2 composites (10 m), we reconstructed approximately 80 years of seagrass distribution. YOLO-based segmentation using deep learning achieved high accuracy (overall accuracy ≥ 0.9) across these datasets; although species could not be discriminated, the models captured the major temporal dynamics in vegetation area.

The long-term mean seagrass area was 6.8 ha, but values fluctuated widely, from 3.5 ha in 1974 to 41.3 ha in 1989 except 0.2 ha in 2025. The highest total area was recorded in 1989 (41.3 ha), followed by 1999 (14.0 ha) and 1966 (13.6 ha), indicating that vegetation was most extensively distributed around the 1990s. In contrast, detections in 1974 and 2019 showed considerably smaller values, with the total area declining to 3.5 ha in 1974, 4.5 ha in 2019, and reaching a minimum of 0.2 ha in 2025.

Sentinel‑2 composites from 2019 to 2026 revealed clear seasonality, with vegetation increasing in early summer and declining from autumn. In 2025, however, the area decreased sharply after summer and remained anomalously low throughout the winter of 2025–2026.
A distinct seasonal pattern was observed, with peaks occurring annually from May to July (monthly averages for the entire period: June: 16.0 ha, July: 17.6 ha), followed by a repeated pattern of rapid decline from late summer through autumn. However, observations for August and September were limited in many years due to cloud cover.

Field surveys confirmed the absence of living Z. marina shoots, while Zostera japonica persisted locally, indicating that the 2025 event was not a normal fluctuation in this area but a rapid ecosystem shift involving loss of the dominant canopy-forming species, most plausibly driven by regionally elevated summer water temperatures.

The findings also have implications for seagrass Essential Ocean Variables (EOVs) and the State of Nature (SoN) metrics used in TNFD-aligned nature-related disclosures. Unlike forests, seagrass meadows require finer temporal resolution because both pronounced seasonality and abrupt collapse strongly influence area-based indicators. Therefore, in addition to previously noted issues such as species-level classification accuracy, we recommend that (1) baselines be defined over the longest available record and justified ecologically, (2) seasonal standardization be applied before inter-annual comparisons, and (3) years with extreme area anomalies be flagged rather than used as reference points.

This study demonstrated that, by integrating historical aerial photographs, high-resolution satellite imagery, and satellite constellations using open-source analytical methods, it is possible to reconstruct both long-term and seasonal variations in seagrass vegetation with high accuracy, and to detect sudden ecosystem collapse. Both long-term and seasonal variations were significant, and while the recent disappearance of eelgrass fell within the range of these variations, it was qualitatively different; the event was likely influenced by elevated sea surface temperatures across a broad region, rather than by site-specific factors alone. For reporting frameworks such as TNFD and EOVs, using seagrass area as an indicator requires not only verification using sparse high-resolution images and annual field surveys but also the establishment of an appropriate time scale with high-frequency observations and standardized seasons.

This session shares the 16-year evolution of FOSS4G Hokkaido, Japan’s first local community. Moving beyond a simple success story, we offer a practical "Survival Toolkit" to overcome organizer burnout and organizational challenges, providing actionable insights to help communities foster Geospatial Sovereignty through long-term resilience.

The rapid development of high-resolution topographic information obtained through unmanned aerial vehicle (UAV) LiDAR systems has significantly improved the capability to analyse terrain morphology, detect geomorphologic processes, and support spatial hazard analysis at centimetre-scale resolutions. At the same time, the maturity of open-source geospatial software frameworks, including QGIS, GRASS GIS, SAGA GIS, PDAL, ESA SNAP, and the Python scientific stack, has enabled the development of transparent and reproducible analytical methods outside closed software environments. Despite these developments, a considerable number of terrain hazard modelling studies still rely on closed or only partially documented processing chains, which limits transparency, reproducibility, and scientific reuse of methods. In addition, while machine learning-based landslide and terrain instability analysis methods are well established, only a small number of studies present fully reproducible end-to-end workflows based on free and open-source geospatial software, including UAV LiDAR-derived geomorphometry.
In this study, we present an open, reproducible, and fully documented machine learning-based framework for high-resolution terrain classification and hazard susceptibility mapping using UAV LiDAR-derived geomorphometry. The main purpose of this study is to demonstrate that robust scientific analysis of terrain morphology, including terrain hazard modelling, can be achieved using an open geospatial software stack, while at the same time ensuring that the predictive accuracy of the proposed framework is comparable to closed software environments. A secondary purpose of this study is to quantify the relative importance of key geomorphometric factors controlling terrain instability at very high spatial resolutions, while at the same time exploring the reproducibility of the proposed framework within the broader context of open geospatial science and FAIR data principles.
The research is based on a dense UAV LiDAR survey that results in a centimetre-scale digital terrain model after classification of ground points, removal of outliers, and interpolation via an openly scripted PDAL processing pipeline. All the data were acquired in Croatia (Europe), on different geomorphology terrain characteristics. UAV LiDAR data were collected using DJI Matrice 350 RTK with L2 payload. To ensure centimetre level accuracy, UAV was paired with GNSS receiver Emlid Reach RS+ with the connection on Croatian Positioning System (CROPOS). From the normalized terrain surface, a full range of geomorphometric derivatives was computed via GRASS GIS and SAGA GIS plugins integrated into QGIS. The computed derivatives include slope gradient, aspect, plan and profile curvature, terrain position index, surface roughness, flow accumulation, and the LS erosion factor. These parameters are recognized as key indicators of morpho dynamic processes linked to slope instability and erosion. The raster layers were normalized to a consistent spatial resolution and extent and a consistent coordinate reference system to obtain a multivariate predictor stack for machine learning-based research.
The reference data for the supervised models were derived from detailed geomorphological interpretation of the LiDAR terrain surface and existing engineering-geological information in the study area. The stable and unstable terrain units were identified to develop the training and validation datasets. To guarantee methodological reproducibility and computational reproducibility, all the steps involved in the research were implemented via openly accessible Python scripts and QGIS models.
The machine learning classification was performed via open-source Python libraries. The most dominant libraries utilized were scikit-learn. The classification results were evaluated via spatially independent validation. The models were tested via confusion matrix, accuracy, F1 score, and receiver operating characteristic score. Feature importance was also computed to evaluate the relative contribution of individual geomorphometric variables to terrain instability hazard. The feature importance results provided physically interpretable results on terrain instability via LiDAR terrain morphology.

The results show that fully open-source machine learning workflows, when applied to UAV LiDAR-derived geomorphometric parameters, can attain high predictive reliability in differentiating stable from potentially unstable terrain conditions. Ensemble-based methods, particularly Random Forest, show the most balanced performance and robustness against predictor multicollinearity. For all models, slope gradient, curvature, and flow accumulation are consistently identified as dominant predictors, which is consistent with geomorphological theory and further underscores the analytical value of centimetre-scale terrain representation. Beyond predictive accuracy, the principal scientific contribution of this work is its operational reproducibility within the open geospatial community. The entire workflow, from point cloud pre-processing to final susceptibility map generation, can be executed using exclusively free and open-source tools, along with fully disclosed parameters and computational steps. Moreover, all processing scripts, derived data products permitted under data sharing restrictions, and workflow documentation will be made openly available under an open-source license via an open repository. The proposed framework is inherently transferable and applicable to the rapidly developing and expanding UAV LiDAR acquisition initiatives across the globe. By demonstrating the capability of high-end terrain hazard modelling without recourse to proprietary tools, this work makes a significant contribution to a community-based methodological blueprint applicable to environmental monitoring, infrastructure planning, and risk management under resource-constrained or open science-based scenarios. By integrating UAV LiDAR geomorphometry, interpretable machine learning, and fully reproducible open-source processing, this work advances the scientific and operational role of free and open geospatial technologies in high-resolution Earth surface science.

The greatest gap in geospatial technology is not data-to-code — it is complexity-to-usability. Through two real-world platforms, this session explores how to turn powerful open-source spatial tools into decisions anyone can act on — by giving users less to think about, not more data to interpret.

The NISAR mission launched in July 2025, and data is now publicly available. This talk presents a range of open source tools available for accessing and transforming L-band data from the mission.

Learn to overcome geospatial scaling bottlenecks by moving from monolithic designs to a composable architecture. We introduce Meros, an open-source feature data service, demonstrating how decoupling APIs—like OGC API - Features—from databases creates highly flexible, resilient, and easily deployable services.

Watershed delineation is essential for flood hazard mapping to minimize disaster impacts. Traditional approaches require time-consuming DEM preparation and preprocessing. This presentation introduces a serverless function that extracts watersheds from nationwide flow direction data hosted on cloud storage, comparing performance between COG and Zarr formats.

This presentation concerns migrating an SDI from a single-machine, multi-container Docker architecture to a cluster-based solution. The authors will share their experience, including performance testing results and an evaluation of potential bottlenecks. They will also discuss the critical human factors involved in such a migration.

This presentation highlights how GeoSolutions leverages cloud infrastructure and geospatial technologies to enable precision farming at scale. Drawing on 10 years of experience, it covers data ingestion, optimization, modeling, and GeoServer deployment strategies, along with real-time visualization techniques for designing scalable, high-performance agricultural data solutions.

Terra Draw is a drawing library having unified interfaces for most of mapping libraries. This talk updates the state of TerraDraw and maplibre-gl-terradraw this year.

This talk traces my path from an undergraduate Geomatics Engineering student to an open-mapping leader who applies OpenStreetMap (OSM) data to real engineering problems in Nepal. Since 2023, OSM has served as working engineering data across projects spanning flood hazard modelling, disaster-preparedness mapping, public health, and urban planning — including a flood susceptibility study of the Sunkoshi River Basin where OSM-derived hydrography, combined with AHP-weighted terrain, rainfall, and land-cover data, showed that community-mapped rivers can substitute for missing authoritative datasets in mountainous terrain. Alongside this technical practice, I founded KU YouthMappers to open geospatial work to students across faculties and put women into leadership, lifting female participation in flagship mapping activities to 62% and earning recognition from the Annapurna Post. This talk presents both threads together open data used for real engineering solutions, and the community that makes that work sustainable as a model for women-led open-mapping practice.

National-scale geospatial analysis using conventional geospatial RDBMS often faces trade-offs between spatial resolution and computational cost. It makes extracting insights from geospatial big data time-consuming or expensive. This proposal demonstrates two national-scale WebGIS applications built with open source pipelines to mitigate them.

A real-world case from Paraguay showing how a combination of open-source geospatial tools supports biodiversity monitoring, trail management, wildlife tracking, and conservation. The system integrates field data, fire alerts, deforestation monitoring (GLAD), and lightweight drone data into a single, scalable workflow for protected area management.

How JR West Democratized Geospatial Data to Tackle the Technical Succession Crisis

This presentation introduces an open-source Geospatial Observation Stack using OpenTelemetry, Prometheus, Grafana, and Loki. We demonstrate how end-to-end tracing and SLOs reduce MTTD and MTTR in cloud-native architectures, enabling developers to pinpoint bottlenecks across complex spatial data pipelines and move toward predictive maintenance.

Geoconnex links hydrologic data across many government agencies in the United States as a knowledge graph, published in accordance with W3C Spatial Data on the Web best practices. This session will provide an overview of Geoconnex and how to access its data using either SPARQL, OGC API Features, or GeoParquet.

Terrain traversability mapping is an important geospatial task in off-road mobility assessment, route planning, environmental analysis, and spatial decision support. In recent years, AI and machine learning methods have increasingly been used to estimate terrain-related indicators from heterogeneous spatial inputs, offering a scalable alternative to purely rule-based GIS procedures. However, model performance in this domain is often summarized through aggregate regression metrics alone, while the spatial structure of prediction error and the influence of incomplete geospatial reference data remain underexplored. This paper addresses that gap by focusing not only on predictive accuracy, but also on uncertainty in AI-based terrain traversability modelling. Rather than asking only whether a model predicts well on average, we ask where it fails, why those failures occur, and how limitations of source geodata propagate into mapped analytical outputs.

The study investigates the estimation of a terrain passability coefficient (IOP), expressed as a continuous value in the range from 0 to 1 and interpreted as the difficulty of traversing a given terrain unit. The modelling workflow is built around regular 100 m × 100 m grid cells, which serve as primary spatial units for analysis. Within each cell, vector-based topographic and thematic data are transformed into a structured set of quantitative attributes describing surface features, linear features, point objects, and terrain morphology. This conversion from a discrete vector data model to a continuous feature space enables the use of machine learning methods for prediction of terrain passability. The final dataset contains 236,617 records and 115 non-empty explanatory variables selected from a larger attribute structure affected by missing values. A compact artificial neural network is used for regression-based estimation of IOP, with 20% of the observations reserved for model testing.

The architecture of the model is intentionally simple. It consists of an input layer, one hidden layer with 57 neurons, and a single-neuron output layer. The hidden layer uses the ReLU activation function and L2 regularization, while model training is guided by mean absolute error as the loss function and a decreasing learning rate. This design choice is deliberate: the goal of the study is not to claim novelty through increasingly complex network architectures, but to examine how geospatial data quality constrains the reliability of AI-based prediction. In this sense, the model is treated as an analytical instrument for exposing data-driven uncertainty rather than as an end in itself.

At the aggregate level, the model achieves good performance. On the test set, the results reach BIAS = 0.0024, MAE = 0.0097, and RMSE = 0.015, with training and validation behaviour suggesting neither severe overfitting nor underfitting. These values indicate that machine learning can successfully approximate the terrain passability coefficient from structured geospatial attributes. Yet the study shows that such global statistics do not fully capture model behaviour in space. The distribution of errors is not random, and the most important discrepancies emerge in value ranges and terrain contexts that are weakly represented or incompletely encoded in the source data. In particular, low IOP values tend to be underestimated, while the upper extreme of the coefficient range is also less stable. This means that the model may overestimate traversability in some difficult areas and underestimate it in highly passable terrain, despite apparently good overall metrics.

The most significant contribution of the paper lies in the spatial interpretation of these discrepancies. Local analysis reveals that some of the strongest deviations are not primarily caused by the neural model itself, but by incompleteness in the geospatial reference base used to construct training and validation labels. Two examples are especially illustrative. A very large discrepancy was identified over a mining area, where the lack of sufficiently detailed source information caused the feature representation of the terrain cell to differ substantially from actual conditions. Another important discrepancy appeared in wetland terrain, where missing swamp-related information in VMAP led to a clear mismatch between predicted and reference passability. These cases show that a model may appear statistically robust while still producing spatially misleading outputs in areas where key terrain categories are absent, weakly represented, or overly simplified in the underlying geodata. From the perspective of GeoAI, uncertainty is therefore not only an algorithmic issue; it is also inherited from the thematic completeness and semantic fidelity of the input spatial database.

A second major source of uncertainty concerns representativeness. The training data originate from a specific study area and do not include explicit temporal labels, even though terrain traversability is inherently sensitive to seasonal and environmental variability. A model that performs well in one region may not preserve the same level of accuracy when transferred to another region or to similar terrain observed under different weather or seasonal conditions. This issue is particularly important for traversability analysis, where the same land cover or soil-related feature may behave differently across time due to moisture, freeze-thaw effects, vegetation changes, or local environmental context. The paper therefore argues that evaluation of geospatial machine learning models should explicitly consider both thematic incompleteness and spatiotemporal representativeness, rather than treating prediction error as a purely technical property of model architecture.

The contribution of the paper is threefold. First, it presents a practical AI-based workflow for terrain traversability prediction from structured geospatial attributes derived from open spatial analysis procedures. Second, it demonstrates that spatially explicit error analysis provides insights that remain invisible in global regression statistics alone. Third, it frames incomplete and unevenly representative geospatial reference data as a primary source of uncertainty in traversability prediction. By shifting attention from model novelty to data-driven reliability, the paper contributes a realistic and methodologically transparent perspective to GeoAI research in the open geospatial domain. The broader implication is that future work on terrain traversability prediction should invest not only in better models, but also in richer, more complete, and more temporally explicit geospatial reference datasets.

Forest vertical structural diversity is a key indicator of ecosystem complexity, habitat heterogeneity, and biodiversity potential. Foliage height diversity (FHD), derived from vertical vegetation profiles, is widely used to quantify this structural heterogeneity. Airborne laser scanning (ALS) provides accurate three-dimensional forest structure information, but its limited spatial coverage and high cost hinder large-scale monitoring. The Global Ecosystem Dynamics Investigation (GEDI) mission has enabled global sampling of forest vertical structure using spaceborne LiDAR. However, its footprint-based sampling produces spatially discontinuous observations. As a result, continuous regional-scale mapping of forest vertical structural diversity remains a major challenge. Previous studies have focused primarily on height-related metrics or canopy cover estimation. Regional-scale mapping of foliage height diversity remains limited, especially in cool-temperate and boreal forest ecosystems like northern Japan, where complex terrain, climate gradients, and forest management regimes interact.
This study proposes a large-scale framework for estimating forest vertical structural diversity across Hokkaido, Japan, by integrating GEDI-derived FHD with multi-source satellite remote sensing data and machine learning. The novelty of this work lies in two key contributions: (i) integrating a wide range of complementary satellite data sources including multi-frequency SAR, optical imagery, climate variables, land cover products, topography, and nighttime light data, for FHD estimation, and (ii) explicitly interpreting model behavior using Shapley additive explanations (SHAP) to identify physically meaningful drivers of forest vertical structural diversity.
We used diverse openly available geospatial data. GEDI Level 2B observations acquired in 2023 served as reference data. After quality filtering, GEDI footprints were spatially matched with satellite-derived features, including Sentinel-1 SAR, Sentinel-2 spectral bands, ALOS L-band SAR metrics, a digital elevation model (DEM) from JAXA, TerraClimate climate variables, forest type classes from Copernicus Global Land Service (CGLS), Dynamic World-derived forest probability, and VIIRS nighttime light intensity. These variables were selected to represent complementary aspects of canopy structure, vegetation condition, terrain-driven environmental gradients, climatic constraints, and anthropogenic disturbance. A Light Gradient Boosting Machine (LightGBM) regression model was trained to predict GEDI-derived FHD from the multi-source feature set. Model evaluation was conducted using out-of-fold (OOF) predictions to reduce optimistic bias and to provide a robust estimate of generalization performance. The resulting predictions were used to produce a spatially continuous wall-to-wall map of FHD for the entire region.
The model achieved an RMSE of 0.360 and an R² of 0.306 in predicting GEDI-derived FHD. This indicates that multi-source satellite observations capture part of the spatial variability in forest vertical structural diversity across heterogeneous landscapes.
The predicted wall-to-wall FHD map reveals spatially coherent patterns across Hokkaido. Higher values appear in mountainous forested regions, while lower values occur in flatter or more human-influenced areas. Extremely high-elevation areas show lower FHD values, likely reflecting harsher climatic conditions and simplified forest structures near the treeline.
These spatial patterns align with known ecological gradients in forest composition and management intensity. This suggests the model captures meaningful large-scale structural variability rather than random noise. To interpret the model beyond simple prediction, we used SHAP analysis to understand how individual features contribute to predicted FHD values.
SHAP-based global feature importance showed that tree cover probability (from land cover products) was the most influential predictor. This highlights the fundamental role of forest presence and canopy continuity in determining vertical structural diversity. Optical spectral bands from Sentinel-2, particularly visible and red-edge bands, contributed strongly, reflecting how spectral responses vary with canopy density and vegetation condition. SAR backscatter from ALOS L-band and Sentinel-1 also showed high importance, indicating that longer-wavelength radar signals capture canopy structure and woody biomass information relevant to vertical heterogeneity. Topographic variables (DEM) contributed significantly, suggesting that elevation-related environmental gradients influence forest structure through climate and disturbance patterns. Nighttime light intensity and its distance-based metrics showed measurable but secondary contributions, implying that human pressure is associated with reduced vertical structural diversity in more developed areas. SHAP summary plots further revealed nonlinear and asymmetric relationships between key predictors and FHD. Higher tree cover probability and stronger L-band HV backscatter were associated with positive contributions to predicted FHD, while increasing nighttime light intensity tended to reduce it. Elevation showed both positive and negative contributions depending on context, reflecting complex interactions between topography, forest type, and management practices. These results demonstrate that the machine learning model captures ecologically meaningful relationships rather than purely statistical correlations.
This study's key contribution is demonstrating that foliage height diversity can be estimated at regional scale by combining spaceborne LiDAR data with multi-modal satellite observations. The resulting model can be meaningfully interpreted in ecological terms using SHAP analysis. This interpretability distinguishes our framework from purely predictive approaches and reveals how canopy cover, radar-derived structure, topography, and human influence shape forest vertical structure.
While predictive accuracy remains moderate, reflecting the inherent challenge of inferring three-dimensional forest structure from two-dimensional satellite observations, the results show that broad regional patterns of forest vertical structural diversity can be captured consistently.

Nominatim struggles with CJK place name search. Searching for "広島平和記念資料館" (Hiroshima Peace Memorial Museum) returns no results.
Its token-based search requires exact token matches and depends on complete alternative name data - which is often missing in OSM.
I demonstrate how PostgreSQL extensions can fix this with minimal code changes.

Coffee Break sponsored by MIERUNE Inc. & Cesium

Sustainability challenges are deeply intertwined and complex. Maps made with open-source software and FAIR spatial data can be used as leverage points to promote global sustainability understanding while simultaneously integrating localized values and identifying solutions. Cartography can be utilized to bridge the Policy – Data nexus globally and locally.

Speakers from various United Nations entities share how technology actually works in the field — and what is still missing — in the closing keynote of FOSS4G 2026 Hiroshima.