The 13-km Rapid Refresh (RAP) and 3-km convection-allowing High-Resolution Rapid Refresh (HRRR) are hourly updating weather forecast models that use a specially configured version of the Advanced Research WRF (ARW) model and assimilate many novel and most conventional observation types on an hourly basis using the Gridpoint Statistical Interpolation (GSI) hybrid ensemble analysis package. Included in this assimilation is a procedure for initializing ongoing precipitation systems from observed radar reflectivity data, a cloud analysis to initialize stable layer clouds from METAR and satellite observations, and special techniques to enhance retention of surface observation information. Each hour, the HRRR is run out to fifteen forecast hours over a domain covering the entire conterminous United States using initial and boundary conditions from the hourly-cycled RAP. A key feature of the hourly HRRR system is the very short model latency, with 15-hr forecast output grids available by 90 minutes after the initial time and made available to operational forecasters in both the private and public sectors. In September 2014, the HRRR was transitioned, for the first time, from a real-time experimental research model into the operational NCEP production suite.

Recent development of the HRRR has focused on (1) refining the 3-km HRRR data assimilation to further improve the synthesis of storm-scale information toward improving the early hours of the forecast, and (2) enhancing the RAP and HRRR model physics for better land surface and boundary layer prediction using the updated Mellor-Yamada-Nakanishi-Niino (MYNN) parameterization scheme, Grell-Freitas-Olson (GFO) shallow and deep convective parameterization, aerosol-aware Thompson microphysics and upgraded Rapid Update Cycle (RUC) land-surface model.

This presentation will include a description of the changes to the real-time HRRR configuration, including sub-hourly 3-km radar data assimilation during a pre-forecast hour, hybrid variational-ensemble assimilation of conventional observations and a 3-km non-variational cloud and precipitating hydrometeor analysis using radar reflectivity observations to retrieve rain and snow mixing ratios. Special attention will be given to continuity of convective-scale structures, originating from an accurate storm-scale analysis (initial condition), during much of the free forecast period (several hours). The presentation will also highlight improvements in the RAP and HRRR model physics to reduce certain systematic forecast biases including a warm and dry daytime bias over the central and eastern CONUS during the warm season. Finally, a preview of convection-allowing regional ensembles from current time-lagged HRRR forecasts to more formal ensemble system designs will be discussed in the context of providing both deterministic and probabilistic high-resolution forecasts in the near future.