Real-time flood prediction

Real-time flood prediction (marketed earlier as FloodCube Realtime) is 7Analytics' early-warning service for flooding in the coming 72 hours. Flood impacts are rising, and conventional hydraulic models are often too slow and calibration-heavy to give site-level warnings while an event unfolds.

The service combines live rainfall from gauges and radar and calibrated numerical weather forecasts with river-level sensor readings, 1×1 m terrain, satellite-derived land-use classifications, building footprints and drainage data. A rapid 2D inundation model based on the Weighted Cellular Automata 2D (WCA2D) framework, machine-learning rainfall thresholds and neural-network discharge forecasts for gauged and ungauged basins produce flood depth, extent and probability at hourly time steps for every building, asset and user-defined warning point. For major US rivers and coastal storm surge it builds on NOAA's National Water Model, so pluvial, flash and river flooding and storm surge are handled in one framework. Forecasts are updated at least hourly.

Outputs reach users through dashboards, APIs, alert systems and maps. Alerts fire when predicted depth exceeds user-defined thresholds and give the estimated timing, depth and duration. The weather side was developed with StormGeo. Since May 2023 the service has provided 24/7 flood prediction for the TotalEnergies Tower in Houston, with hourly forecasts shown as warnings in TotalEnergies' StormGeo risk dashboard. It is built for insurers, brokers and reinsurers, infrastructure networks such as roads, railways, utilities and hydropower, asset owners and public-sector response teams.