Investigation of Ensemble-Based Nowcasting for Predicting Excessive Rainfall within the MRMS System

The need is there to improve the very short-term prediction of precipitation for increasing flash flood warning lead times and help protect life and property. Efforts are now underway to implement a high-resolution ensemble-based precipitation nowcast into the Multi-Radar Multi-Sensor (MRMS) system. This study explores the first phase of the nowcasting project within the MRMS system. The Short-Term Ensemble Prediction System (STEPS) was used to test the nowcasting of MRMS instantaneous precipitation rates out to 60 minutes at a two-minute timestep. The focus of this first phase is to test various parameter settings to find the optimal performance of the nowcast rates. Parameters tested include the horizontal resolution of the MRMS data (notably 1-km versus 500-m), number of ensemble members, the advection tracking rate threshold, and perturbations of the advection field. Results were generated from the evaluation of eight different flash flood events, which ranged from training storms to monsoonal convection to tropical cyclones. The results show which parameters had a more significant influence on the nowcast rates and its performance.

This presentation will describe the use of STEPS with MRMS precipitation data, how the STEPS nowcasts performed across the various settings and flash flood event cases, and discuss the future design and goals for operational nowcasting with the MRMS and Flooded Locations and Simulated Hydrographs (FLASH) systems to help with flash flood prediction and warning lead times. This includes the concept of operations, the creation of hydrologic modeling output using nowcast rates as an additional forcing, and the generation of exceedance probabilities for various rain rate values.