The 3-km High-Resolution Rapid Refresh (HRRR) hourly updated weather model is a cornerstone of the new Wind Forecast Improvement Project (WFIP) sponsored by Department of Energy and NOAA, promising to improve detailed forecast guidance for both 50-150 wind and solar energy over that currently available from operational NOAA weather forecast models.

In this talk, we will present modifications made to improve forecasts to the HRRR model and its parent model/assimilation system, the 13km Rapid Refresh (RR). HRRR model forecasts are in initial testing from a next-generation version with an improved land-surface model with more accurate land-use data from the MODIS satellite and other improved physical parameterizations for boundary-layer processes and cloud microphysics.

Experiments are also underway for improved data assimilation for the initial conditions for the HRRR with addition of nacelle winds, better QCed mesonet winds, radial winds, and boundary-layer-based pseudo-observations. In addition, enhancements to the cloud/hydrometeor assimilation are also in testing to more faithfully represent GOES-based cloud imagery. These assimilation modifications are within the 13km Rapid Refresh, the HRRR parent model, and the RR-specific version of the Gridpoint Statistical Interpolation (GSI).

Recent validation results on HRRR forecasts for turbine-level winds (WFIP) and solar GHI will be presented, including effects of these latest modifications.