The HRRR relies on the RUC13 data assimilation, which includes radar reflectivity assimilation based on a digital filter initialization (DFI) technique. Use of the forward (diabatic) DFI inside the RUC (and in the future 13km Rapid Refresh) is shown to dramatically improve reflectivity forecasts from the HRRR.
The HRRR has considerable promise for short-range thunderstorm prediction and future warn-on-forecast applications. The HRRR has shown particular skill at accurately depicting storm mode (structure) and location. Also, the hourly output and hourly update frequency of the HRRR provide a large number predictors for the creation of a HRRR-based convective probability guidance product known as the HRRR convective probabilistic forecast (HCPF).
A description of the HRRR configuration will be provided along with case studies including 15-hour forecasts run hourly during the spring and summer of 2010 over the CONUS demonstrating the predictive skill of HRRR individual (deterministic) and time-lagged ensemble (probabilistic) forecasts. Future enhancements including radar assimilation at 3-km within the HRRR will also be discussed.