Addressing convective-scale forecast challenges with the High Resolution Rapid Refresh (HRRR) Model

The High Resolution Rapid Refresh (HRRR) model is being run hourly in real-time at the Global System Division (GSD) of the Earth System Research Laboratory (ESRL). The Weather Research and Forecasting (WRF) based model is run out to fifteen forecast hours over a domain covering the entire continental United States (CONUS) at a spatial resolution of three kilometers, allowing the use of explicit convection. Initial and boundary conditions are derived from the Rapid Update Cycle (RUC) which performs radar data assimilation using a diabatic digital filter initialization.

The production of real-time HRRR forecasts during the spring , summer and fall of 2010 highlighted several challenges with mesoscale and convective-scale forecasts. These forecast problems included “false alarms” where the model generates widespread spurious convection during the first few forecast hours in regions where none is actually observed, and the inability to realistically represent MCS propagation and/or sustain some classic leading-line/trailing-stratiform MCS structures.

Retrospective case-study model experiments will be discussed that attempt to address these and other forecast deficiencies including changes to the radar assimilation procedure, modification of initial conditions and/or changes to the WRF model dynamics/physics including parameterization schemes.