ANN-CSGD Real-Time Precipitation Forecast

Using GEFS ensemble model data and an Artificial Neural Network (ANN), we generate
probabilistic precipitation forecasts across the contiguous United States.
The model predicts a full Censored Shifted Gamma Distribution (CSGD)
at every land grid point, producing calibrated exceedance probabilities for rolling
12-hour precipitation windows. Animations below show the full forecast horizon for the
latest model cycle.

Loading latest model cycle…

P > 1 inch · rolling 12-hour exceedance probability

Forecast Animation (CONUS)

Click the animation to open it full size.

Excessive Rainfall Outlook (ERO) Comparison

Each panel compares the ANN-CSGD 24-hour product, the GEFS member-derived 24-hour
product, and the WPC Excessive Rainfall Outlook for the selected forecast day.

Forecast day

Day 1 · P > 1 inch (24h)


ANN-CSGD ERO 2x2 comparison

Click the panel to open it full size.