2026/07/27 by Samuel C. Dahl, Xubin Zeng, Michael A. Brunke +2
Earth and Planetary Sciences · Environmental Science · #Climate variability and models #Meteorological Phenomena and Simulations #Tropical and Extratropical Cyclones Research
paper · doi:10.1175/waf-d-25-0109.1
openalex publication_date 2026/07/27 · openalex created_date 2026/07/28 · openalex updated_date 2026/07/28
Abstract The state of California experiences numerous weather extremes, from flooding rains to devastating wildfires. West-WRF is a customized version of the Weather Research and Forecasting model tailored for hydrometeorological prediction in the Western U.S. This work presents an evaluation of West-WRF hydrometeorological prediction in December 2021 for California against in-situ measurements, ERA5 reanalysis, and other observational datasets. Its performance is compared with those of the global models that drive it as initial and lateral boundary conditions (GFS and ECMWF). As novel AI weather prediction models are increasingly being used, we also evaluate two such models trained using ERA5 data. We show that West-WRF driven by the GFS improves precipitation prediction at short lead times over California and reduces mean absolute error (MAE) over the GFS. The improvement in the snow water equivalent prediction is more substantial, with a larger reduction in MAE of ~70% over the GFS, partly due to an improved snow initialization in West-WRF. West-WRF also improves the prediction of 10-m wind speed in the coastal region of Southern California. Its performance with other quantities including 2-m temperature and specific humidity is more variable. The AI models perform well in predicting the spatial pattern of daytime 10-m wind compared with ERA5 reanalysis, but their performance against in-situ measurements in Southern California is comparable to that of GFS and ECMWF physics-based models, which are much worse than that of West-WRF. These evaluations also provide insights on further improvements of West-WRF.