Assessing Extreme Precipitation Forecasts through In-Situ Soil Measurements: Verification and Validation of the RRFS Ensemble Model

Accurate soil moisture forecasting is vital for predicting near-surface temperature and humidity, the uptake of moisture into developing storm systems, cloud formation, and assessing drought and fire risks. A persistent warm daytime bias during the warm season has been identified as an ongoing issue for many climate and weather models, often attributed to soil moisture bias. Incorporating soil moisture as a control state in Numerical Weather Prediction systems has shown to reduce temperature and humidity forecast errors. This research project evaluates the performance of the experimental Rapid Refresh Forecast System (RRFS) ensemble model developed as part of NOAA’s Unified Forecast System, or UFS. Through comprehensive statistical model verification, this study aims to provide insights into the reliability and precision of the RRFS ensemble, specifically for extreme precipitation events. Three time periods are evaluated: Summer 2023, Winter 2024, and Spring 2024. The analysis includes an ensemble of the model with five members and a control, as well the currently-operational High Resolution Rapid Refresh as a member. Variables assessed include soil moisture and soil temperature at 5 cm, 10 cm, and 50 cm depths, as well as 2-m air temperature and 1-hour precipitation accumulations. In-situ data from the International Soil Moisture Network (ISMN), including 5 different data networks, is used to compare to model values to evaluate the skill of individual ensemble members and overall model biases. This project emphasizes the importance of soil moisture data assimilation to improve land surface condition predictions in order to create and maintain a weather-ready nation.