Forecast Skill and Spread of Three Convection-Allowing Ensembles during the 2024 Spring Forecast Experiment

NOAA has developed the RRFS Ensemble Forecast System (REFS) as a replacement / upgrade for the operational High-Resolution Ensemble Forecast (HREF) system. During the April 29 - May 31, 2024 Hazardous Weather Testbed Spring Forecast Experiment (SFE), participants assigned subjective ratings to forecasts from 4 experimental REFS configurations and to the operational HREF. For day 1 forecasts (12Z to 12Z), the subjective ratings averaged over the 5-week evaluation period were similar for all of these ensembles.

Following the subjective evaluation during the SFE, we objectively verify the HREF and two of the REFS configurations. We compute bias, fractions skill score, and ensemble reliability for accumulated precipitation during 6-h periods. We analyze ensemble spread spatially and temporally for the precipitation forecasts and for environmental fields. By computing statistics for subsets of the 10-member ensembles, we diagnose relative contributions to the forecast performance by initial-condition diversity, physics diversity, dynamic-core diversity, and time lagging. By identifying contributors to ensemble spread and forecast skill for each ensemble, we aim to inform refinements of REFS configurations in the near future.