NEWS-e: Evaluation of Real-Time Warn-on-Forecast Precipitation Forecasts

Reliable numerical guidance for heavy convective precipitation can be difficult to produce due to numerous factors. For example, there is the challenge of either predicting convective initiation of thunderstorms or initializing active storms in convection-allowing models. In addition, there is inherent uncertainty in predicting the development of thunderstorms once they form. Deterministic convection-allowing forecasts such as those available from the High-Resolution Rapid Refresh (HRRR) model are useful, but are not initialized with detailed cloud-scale structures, and hence lack information about forecast uncertainty. One of the ways that the Warn-on-Forecast (WoF) project is advancing the prediction of heavy convective precipitation is to use an ensemble-based, convective-scale assimilation, analysis, and prediction system.

This study investigates application of a 36-member 3-km ensemble system — the NSSL Experimental WoF System for Ensembles (NEWS-e) — for the heavy-precipitation challenge. We examine output from multiple Spring 2016 events, during which the NEWS-e system ran in real-time, embedded within an experimental HRRR ensemble system. The NEWS-e forecasts are verified using quantitative precipitation estimates from the Multi-Remote Multi-Sensor (MRMS) dataset, and are compared to forecasts from both operational and experimental versions of the HRRR.