Evolution of HRRR Warm Season Convective Precipitation Forecasts Over Three Years of Model Development

The High Resolution Rapid Refresh (HRRR) model is an hourly, 3km, convection allowing NWP model that has been used operationally in the United States since 2014, with the main goal of improving forecasts of convective precipitation. It is currently the only US NWP model assimilating radar data. Yearly updates to the model assimilation, physics, and microphysics are expected to have resulted in improved quantitative precipitation forecasts (QPFs). Features-based verification of QPFs of warm season convective precipitation is performed for the 2013, 2014, and 2015 versions of the model using both ground and space-based precipitation reference datasets. The verification is performed separately over domains in the US Great Plains and Mountain West in order to infer terrain impacts on QPF skill, and reductions in QPF biases due to model updates will be sought.  Additional observations, including reflectivity profiles and satellite radiances, will be used to infer potential relationships between QPF biases and model processes.