Flood-Risk Agent Pipeline
A self-contained example of how an agentic geospatial data pipeline can gather spatial evidence, compare risk factors, and recommend safer alternatives without manual layer stitching.
Inputs
5 layers
Agents
6 steps
Output
ranked sites
This v1 demo uses simulated spatial evidence so it can run without paid APIs. The same interface pattern can be connected to real satellite imagery, elevation models, rainfall products, FEMA flood layers, permitting data, PostGIS, and vector search.
Find lower-risk growth zones with access to existing roads and enough buildable land.
Click a candidate area to inspect the agent risk summary.
Selected site
Riverbend Infill
98
High Risk
Low terrace near river confluence
Evidence profile
Agent reasoning
- Historic flood polygons overlap the western edge.
- Drainage channels converge before leaving the site.
- Urban growth pressure is high, but mitigation cost is likely elevated.
Ridgeview Plateau is currently ranked best for residential expansion with a risk score of 42.
Safer alternatives to compare
Next checks
- Run a parcel-level hydraulic review before acquisition.
- Price detention storage and freeboard requirements.
Ranked candidates
Public demo mode
Real map context for candidate markers, flood-risk overlays, and safer-site comparison.
Relative elevation, low points, flow direction, and drainage stress.
Storm severity adjustment used to lift or lower site risk.
Historic inundation and known flood exposure signal.
Likelihood that a location will face near-term development demand.
Sites ranked by scenario-specific spatial evidence.
The agent can rank alternatives, but a production workflow still needs source URLs, query parameters, timestamps, confidence notes, and human review before a site decision.
Agent 1
Data Collector Agent
Satellite, elevation, rainfall, parcel, and flood-history layers
Agent 2
Terrain + Drainage Agent
Slope breaks, low points, flow paths, and drainage bottlenecks
Agent 3
Rainfall/Flood History Agent
Storm intensity, known flood polygons, and recurrence signals
Agent 4
Urban Expansion Agent
Growth pressure from roads, parcels, utilities, and nearby projects
Agent 5
Risk Scoring Agent
Weighted spatial evidence ranked by the current planning scenario
Agent 6
Recommendation Agent
Safer alternatives, tradeoffs, and next field checks