Evaluating Potential Future Configurations of the High Resolution Ensemble Forecast System

The High Resolution Ensemble Forecast version 2 (HREFv2) became the first fully operational convection-allowing model (CAM) ensemble run at the National Centers for Environmental Prediction (NCEP) in November 2017. The HREFv2 ensemble membership and configuration is based on that of the Storm Scale Ensemble of Opportunity (SSEO), which was utilized in Storm Prediction Center (SPC) operations from 2011 to 2017. Because the HREFv2 members have diverse model cores, physics, and initial conditions, individual deterministic CAMs can be added to or removed from the membership without requiring changes in other members’ configurations. Currently, the HREFv2 consists of four time-lagged members and four non-time-lagged members.

During the summer of 2018, NCEP’s operational version of the High Resolution Rapid Refresh (HRRR) began producing forecasts out to 36 hours for four of its daily cycles (0000, 0600, 1200, and 1800 UTC). This presents an opportunity to add HRRR members to the HREF, which currently also generates forecasts out to 36 hours. The increased ensemble spread obtained by adding one or more HRRR members could also potentially justify removing some or all of the time-lagged members from the current HREFv2 configuration. These 36-hour HRRR forecasts were run experimentally in conjunction with the 2018 Hazardous Weather Testbed (HWT) Spring Forecasting Experiment (SFE), allowing for testing of several new candidate HREF configurations during the peak of the spring convection season. Objective verification metrics and subjective evaluations from the 2018 HWT SFE for model fields relevant to forecasting severe convective storms are presented for these candidate HREF configurations, illuminating the practical and statistical significance of configuration differences.