2026-09-02 –, Himawari
We introduce the first global dataset of road surface type and width for 9 million km of arterial roads using open-source GeoAI workflows. Supporting a Humanitarian Passability Index, it is openly available via HDX (by UN OCHA), enabling scalable infrastructure intelligence for development and climate resilience.
Road infrastructure is the physical backbone of development, disaster response, and humanitarian access. Yet a consistent, global, and open baseline of road surface conditions has long been missing.
We present a first-of-its-kind open dataset mapping road surface type, width, and passability across more than 9 million kilometers of arterial roads worldwide, derived from PlanetScope satellite imagery (2020 & 2024) and enriched with Mapillary street-level imagery.
Our fully open-source pipeline integrates scalable remote sensing workflows, spatial databases, and deep learning models to generate multi-scale infrastructure intelligence:
• Planetary scale: Road pavedness trends correlate strongly with national development indicators (HDI correlation = 0.65), offering a measurable proxy for infrastructure inequality.
• National scale: Mapping unpaved corridors reveals systemic vulnerabilities in trade connectivity and humanitarian access.
• Local scale: Case studies from Northeast India, Ghana, and Pakistan show how governance, conflict, and political prioritization shape road development — directly affecting disaster logistics and climate resilience.
We also introduce a Humanitarian Passability Index, combining surface type and road width to support operational logistics planning in disaster-prone and conflict-affected regions.
The methodology builds on two peer-reviewed studies (2024, 2025), including one forthcoming in Nature Communications, while translating that research into openly accessible, reproducible geospatial workflows for the global community.
All outputs are released under an open license, enabling practitioners, researchers, and humanitarian actors to reuse, validate, and extend the dataset for applications in disaster risk reduction, sustainable development monitoring, and infrastructure equity analysis.
The data is available on HDX (Humanitarian Data Exchange), helping make it more accessible to humanitarian actors and analysts worldwide. Road network and access data are essential in humanitarian settings, helping responders assess logistics routes, identify access constraints, and better plan the movement of people, aid, and supplies.
This talk demonstrates how open-source geospatial technology can transform multi-source imagery into actionable infrastructure intelligence — and how collaboration between the FOSS4G and humanitarian data communities can scale impact globally.
https://arxiv.org/abs/2512.04092
https://www.sciencedirect.com/science/article/pii/S0924271625000784
GeoAI workflows – end-to-end open-source pipeline for processing satellite and street-view imagery, integrating deep learning models and spatial databases.
PostGIS, QGIS, and GDAL for geospatial data management, analysis, and visualization.
I make my conference contribution available under the CC BY 4.0 license. The conference contribution comprises the abstract, the text contribution for the conference proceedings, the presentation materials as well as the video recording and live transmission of the presentation:
GeoAI expert advancing open-source geospatial tools. She develops reproducible AI workflows combining satellite imagery and GIS data to generate accessible datasets and actionable insights for humanitarian aid, climate resilience, and global infrastructure intelligence, bridging gaps between societies, research communities, and state-of-the-art technologies.