The moat · §3.3
Were we right?
Every forecast scored against reality, in public. Nobody else in the dev-API weather space publishes this.
Real-time radar, predictive nowcasting with calibrated confidence, and geospatial queries — for applications that need to understand and react to weather in real time. Every forecast scored against reality, in public.
The moat · §3.3
Every forecast scored against reality, in public. Nobody else in the dev-API weather space publishes this.
MRMS reflectivity grids materialised in seconds, NWS alerts streamed as Server-Sent Events, and a Web-Mercator tile route ready to drop into MapLibre or Leaflet.
GET /v1/mrms/tiles/{z}/{x}/{y}.pngSample a point, reduce a polygon (max / mean / min / count above a threshold). Lat-lng in, value out — no GIS toolchain required.
GET /v1/mrms/grids/polygonPredicted reflectivity 10 / 30 / 60 minutes out, scored against the matching observation as soon as truth lands. Public MAE / bias / RMSE per algorithm × horizon — persistence baseline today, pySTEPS / NowcastNet next.
GET /v1/calibrationThe dev console runs every public endpoint against a local FastAPI instance backed by NEXRAD CONUS data. Spin it up with make start — one command, one terminal.