Improved precipitation forecasts extend hourly prediction and use seasonal modeling
Methodological Changes to the Attention ResUNet Hourly Precipitation Postprocessor
Machine Learning
Summary
Weather forecasts often estimate rainfall for upcoming hours. The authors improved a computer model that turns one kind of weather forecast into many possible rainfall outcomes. Their new method trains fewer models by combining months and times, looks further ahead by 72 hours instead of 48, and adds extra information like local solar time and average monthly rainfall. These changes make the forecasts slightly better and more reliable.
What this means in practice
- •For weather forecast engineers: Generate hourly probabilistic rainfall forecasts up to 72 hours ahead with fewer models and improved skill by using seasonal conditioning and new input data.
- •For urban flood management teams: Use more reliable hourly precipitation probabilities to better plan and respond to flood risks over a three-day forecast window.
Authors
Thomas M. Hamill
Abstract
This note is a technical companion to a previously published preprint describing an Attention Residual U-Net that postprocesses deterministic forecasts from The Weather Company's Global and Regional Atmospheric Forecast (GRAF) model into probabilistic hourly precipitation forecasts. It documents what has changed in that method since publication. Feature-wise Linear Modulation conditioning on calendar season and forecast lead time is used to produce a single trained model for each season, replacing 192 separately trained per-month, per-lead checkpoints. Lead time is extended from 48 to 72 h. Two new input channels are used, per-pixel local solar hour and a static, monthly-varying precipitation climatology. During verification, the climatological reference against which the Brier Skill Score is computed now has an added diurnal dimension, on top of the monthly resolution it already had. Brier Skill Score and reliability are compared between the new vs. the previous training. Forecasts generated with the new training show a modest, consistent improvement of the current training over the original.