Examination of STEPS Nowcast Parameter Configurations for Use with High-Resolution MRMS Instantaneous Precipitation Rates

Abstract New efforts are under way to study the feasibility of a high-resolution, low-latency ensemble-based precipitation nowcast component with the Multi-Radar Multi-Sensor system (MRMS) to help improve flash flood warnings. The Short-Term Ensemble Prediction System (STEPS) was utilized to test the extrapolation of MRMS instantaneous precipitation rates. The STEPS methodology has a series of parameters that are adjustable; thus, it is imperative to understand how modifying each parameter influences the precipitation nowcast. This study evaluated the performance of the STEPS nowcasting methodology based on varying the following parameters: the rate tracking threshold used to generate motion vectors for advecting the instantaneous rate field, the use of a velocity perturbation scheme, and the number of ensemble members. Testing was conducted using the MRMS instantaneous precipitation rate field at two different horizontal resolutions (1 km and 500 m). Findings from the study showed that the use versus nonuse of a velocity perturbation scheme had the greatest influence on the probabilistic fields and the visual appearance of the nowcast rates, especially with the creation and frequency of greater probability values. Altering the rate tracking threshold was shown to favor different settings across a 0–1-h nowcast period. There were negligible differences with the number of ensemble members applied. The application of different horizontal resolutions of the MRMS data was also shown to influence both the deterministic and probabilistic results. Significance Statement The short-term nowcast of high-resolution radar-based features, such as precipitation rates, can provide a low-latency means of providing precipitation forecasts for up to 60 min by extrapolating the data forward in time based on the current movement. A nowcast system can have a number of tunable parameters that can impact the performance of the nowcast. This study examines the influence of these tunable parameters to help improve the accuracy of the nowcast performance. The findings showed that changing these parameters with different horizontal resolutions of data can influence both the statistical measures as well as the probabilistic information derived from it.