2026-09-02 –, Himawari
OpenDR 1.0 is a cloud-native framework designed to close the disaster response "latency gap." It automates multi-sensor fusion and GeoAI hazard detection while integrating humanitarian exposure modeling and real-time field validation via OGC API standards, empowering regional agencies with geospatial sovereignty.
How can we close the "latency gap" during the first golden hours of a disaster? Traditional disaster response workflows are often desktop-bound, siloed, and dependent on a "download-then-analyze" paradigm that fails in resource-constrained environments.
This presentation introduces OpenDR 1.0, a fully open-source, event-driven framework built to automate the journey from satellite acquisition to actionable humanitarian intelligence. We demonstrate how we operationalized complex scientific logic—including Adaptive Otsu Thresholding for flood extraction and Harmonic Regression for rangeland monitoring—by porting them from platform-locked environments into a modular, independent FOSS4G stack.
Attendees will explore a functional five-tier architecture:
Ingestion: Automated discovery of STAC endpoints for multi-sensor data (Sentinel-1 SAR, Sentinel-2, Landsat, GOES-16, and GRACE) managed as Cloud-Optimized GeoTIFFs (COGs).
Orchestration: Managed by Apache Airflow, utilizing Directed Acyclic Graphs (DAGs) to trigger specialized analytical pipelines the moment new data is acquired.
Big Data Compute: Distributed processing using Dask-Geo on Kubernetes, executing PyTorch U-Net models for hazard segmentation and high-speed imagery unmixing.
Mediation: Serving results via pygeoapi and OGC API standards to ensure seamless interoperability with command centers and mobile tools.
Feedback: A unique "Loop-in-the-Citizen" cycle where field responders using KoboToolbox provide real-time validation to iteratively refine model weights and reduce false positives.
We will share practical implementation lessons from three cross-continental case studies: simulating urban flood resilience with Project PLATEAU 3D CityGML in Tokyo; transboundary health and flood monitoring in East Africa; and 15-minute tactical wildfire tracking.
OpenDR 1.0 is a call to action for building reproducible, platform-independent infrastructure. It ensures that regional agencies can own their analytical logic, maintain data privacy, and achieve true Geospatial Sovereignty through the power of the FOSS4G ecosystem.
To get up to speed on the technical and scientific foundation of OpenDR 1.0, attendees are encouraged to review:
OpenDR Framework Repository: https://github.com/bayillag/OpenDR-Framework (Technical architecture and Jupyter Notebooks).
OGC API Standards Overview: https://ogcapi.ogc.org/ (Focus on OGC API - Features and Processes).
Cloud-Native Geospatial Foundation: https://cloudnativegeo.org/ (Understanding COGs and STAC).
Project PLATEAU (Japan): https://www.mlit.go.jp/plateau/ (Background on 3D CityGML models used in Case Study A).
Dask-Geo Methodology: Fang, B., et al., 2022. Dask-Geo: A library for scalable and interactive analysis of big geospatial data.
PostGIS, pygeoapi, Dask-Geo, Apache Airflow, GDAL/OGR, Apache Sedona, QGIS, KoboToolbox, Docker, Kubernetes.
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:
Bayilla Geda is a geospatial researcher based in Ethiopia, specializing in the development of cloud-native frameworks for real-time disaster response. His work focuses on integrating GeoAI, multi-sensor fusion, and distributed computing to address the unique humanitarian challenges of the Global South.
Bayilla is a dedicated advocate for Geospatial Sovereignty, championing the use of open-source software to empower regional agencies with independent analytical logic and data privacy. Through the OpenDR 1.0 framework, he bridges the technological gap in global hazard monitoring to build multi-dimensional resilience for vulnerable populations worldwide.