Method

AquaRoute is a predictive access-risk system, not a chatbot and not a generic map. It combines NASA GIBS satellite overlays, OpenStreetMap road geometry, field reports, and deterministic AI scoring functions to turn flood access evidence into response priorities. If a Gemini API key is configured, the same evidence can be transformed into coordinator-ready narrative explanations through a server API route.

AI Backbone

Road, community, route, alert, and brief outputs are driven by scoring functions over the same evidence graph.

Operational Map

Flood extent, predicted spread, road status, communities, shelters, field reports, and uncertainty are visible together.

Human Control

Reports and overrides can change status, but dispatch decisions remain with trained responders.

Architecture Pipeline

  1. 1Ingest satellite flood masks and forecast layers
  2. 2Join flood risk with roads, bridges, shelters, and population
  3. 3Score road passability and forecast access degradation
  4. 4Score community isolation, vulnerability, and aid urgency
  5. 5Compare truck, boat, and mixed response routes
  6. 6Accept field reports and human overrides
  7. 7Generate alerts and incident briefs for human decision-makers

Responsible AI Rules

  • - Never claim a route is guaranteed safe.
  • - Never dispatch responders automatically.
  • - Always distinguish detected, predicted, and field-confirmed status.
  • - Always show confidence, data freshness, and known limitations.
  • - Rank communities by access, vulnerability, shelter reachability, and population.
  • - Keep human verification and override visible.