Demo · AI for Hydrology

CREST AI Flood Dashboard

Describe a flood in plain language — “flood in Kerrville, Texas, July 2025” — or pick USGS gauges straight from the map, and the dashboard sets up and runs the CREST/EF5 hydrologic model live: 2-D streamflow animates over the map, hydrographs stream in as the model computes, and an AI assistant explains the event, reports skill scores, and can even calibrate the model parameters when the fit is poor. Sign in with Hugging Face for a personal profile.

Open in full screen ↗ Source on GitHub

The demo may take a moment to wake if it has been idle. A simulation with warm-up typically runs 2–4 minutes on the free hardware; repeat runs of the same basin are much faster thanks to state and result caching. If the embed doesn't load, open it in a new tab.