Trishuli Flood Rescue

Inspiration

Floods can disrupt roads, isolate communities, and make emergency coordination difficult. Responders need a way to review incidents, examine geographical context, and organise verification priorities in one place.

We built 'Trishuli Flood Rescue' to explore how geospatial technology can support flood-response coordination in the Trishuli River Basin.

What it does

Trishuli Flood Rescue is a geospatial decision-support prototype featuring:

"Interactive map": Explore the analysis region and geographical context.
"Emergency incident management": Review incident reports, severity levels, and statuses.
"Flood monitoring workflow": Configure a Sentinel-1 satellite change-analysis request for a selected date and area.
"Rescue coordination": Review potential infrastructure-access concerns and settlement connectivity.
"Demo scenario": Explore four clearly labelled illustrative incidents and their sample priorities.
"Reporting": Export incident records as CSV and generate a downloadable demonstration situation report.

The prototype is designed to help users organise information and identify what requires further verification. It does not replace emergency services or field assessments.

How we built it

The frontend uses React, TypeScript, and Vite, with a geospatial map interface. The FastAPI backend is built in Python and includes a workflow using Sentinel-1 imagery from Microsoft Planetary Computer, raster processing, and historical OpenStreetMap data through the ohsome API. Supabase supports the incident-management workflow.

Challenges we faced

Working with satellite imagery and geospatial datasets presents challenges in image access, scene selection, coordinate transformations, and distinguishing potential changes from confirmed flooding.

The live satellite raster-processing workflow remains unresolved in this demonstration, and a live river-gauge feed is not connected. The demo incidents are illustrative rather than real reports. We therefore avoid presenting unverified impact figures as real observations.

What we learned

This project helped us explore geospatial processing, full-stack integration, incident-management design, and the importance of communicating uncertainty in safety-critical applications. Our goal is to develop a practical decision-support tool that helps responders review information and prioritise verification.