OrbitalEye

Inspiration

Floods can damage roads, bridges, buildings, and essential infrastructure while leaving communities isolated from hospitals and emergency services. During a disaster, responders need clear and timely information to understand where help may be needed most.

We created OrbitalEye — Satellite Intelligence for Flood Response to explore how satellite imagery and geospatial analysis can turn complex flood information into useful insights for emergency planning.

What it does

OrbitalEye is designed to bring flood mapping and response planning into one dashboard.

• Flood Mapping: Analyse satellite observations to identify potential flood-affected areas.
• Infrastructure Exposure: Overlay geographic information about buildings, roads, and bridges to assess potential exposure.
• Cut-Off Settlements: Analyse road connectivity to identify settlements that may be disconnected from nearby towns or hospitals.
• AI Situation Reports: Present analysis in understandable language to support emergency-response planning.
• Flood Path Exploration: Use elevation and terrain information to explore potential downstream flood pathways.
• Future Risk Exploration: Support scenario-based exploration of areas that may face increased flood risk.

The objective is to help users move from viewing flood information to understanding its potential impact on communities and infrastructure.

How we built it

We designed OrbitalEye around a dashboard that brings together satellite observations, geospatial layers, and flood-response analysis.

The planned data workflow uses Sentinel-1 radar and Sentinel-2 optical imagery where available, along with pre-event OpenStreetMap data for infrastructure and elevation information for terrain analysis.

The system is organized around a workflow of observing affected areas, assessing infrastructure exposure, analysing connectivity, explaining findings, and preparing situation reports.

The AI-assisted component is intended to explain computed results rather than invent statistics. Outputs should be grounded in the underlying analysis, and unavailable or unverified data must be clearly identified.

Challenges we ran into

One major challenge is combining geographic datasets with different acquisition dates, spatial resolutions, and levels of completeness. Satellite observations also have limitations caused by clouds, terrain, and the timing of image acquisition.

Another challenge is distinguishing infrastructure exposure from confirmed damage. A road intersecting a potential flood zone does not automatically mean that the road is impassable, and a settlement identified as potentially isolated requires further connectivity analysis.

Ensuring that AI-generated explanations remain consistent with computed map results is another important challenge. We also need independent validation before reporting quantitative accuracy.

Accomplishments that we're proud of

We are proud of developing the concept of an integrated flood-response platform rather than treating flood mapping as an isolated task.

OrbitalEye brings together several important disaster-management questions: Where might flooding have occurred? Which assets may be exposed? Which communities could face access difficulties? How can the findings be communicated clearly?

We are also proud of designing the project around transparent data sources, independent validation, and responsible communication of uncertainty.

What we learned

This project helped us understand how satellite remote sensing, geographic information systems, terrain analysis, and road-network connectivity can contribute to disaster response.

We learned that reliable flood intelligence requires more than an attractive map. Data quality, geographic alignment, appropriate validation, and clear explanations are equally important.

We also learned that AI should support evidence-based analysis, not replace it. Every reported number should come from computed results, and limitations should be communicated honestly.

What's next for OrbitalEye

Our next steps are to strengthen the integration of real satellite datasets, test flood-extent analysis, improve infrastructure exposure and settlement-connectivity assessments, and develop evidence-grounded situation reports.

We also plan to evaluate the Trishuli flood case study against the official Copernicus Emergency Management Service reference product EMSR927. This reference will be reserved for independent validation and comparison, not used as an input to our analysis.

Future improvements may include more robust downstream flood-path exploration, scenario-based risk assessment, and clearer uncertainty indicators.

Our long-term vision is to make satellite-derived flood intelligence easier to understand and use, helping emergency responders identify potential impacts and prioritize communities that may need assistance.

Project status: Prototype under development. Features and results should be interpreted according to the data sources and analysis modules actually implemented and tested.

Built With

• analysis
• artificial
• data
• disaster
• flood
• geospatial
• gis
• imagery
• intelligence
• learning
• machine
• management
• mapping
• openstreetmap
• python
• react
• remote
• satellite
• sensing
• typescript
• visualization
• vite