Radar and Satellite Data Assimilation in the RAP and HRRR Models

The 13-km Rapid Refresh (RAP) and 3-km convective-allowing High-Resolution Rapid Refresh (HRRR) are hourly updating models that utilize a specially configured version of the Advanced Research WRF (ARW). The RAP and the HRRR are run in real-time operationally at the National Center for Environmental Prediction (NCEP) and also experimentally at the Earth Systems Research Laboratory's Global Systems Division (ESRL/GSD). The successful assimilation of radar and satellite observations is of paramount importance for initializing the RAP and the HRRR models. Radar and satellite observations are incorporated into the analysis through two components of the analysis system, the GSD cloud and precipitation hydrometeor analysis (hereafter referred to as the HM-analysis) and through the specification of latent heating temperature tendencies.

The HM-analysis is part of the Gridpoint Statistical Interpolation (GSI) data assimilation system that is used to assimilate many observation types in operational models. Currently, the HM-analysis is an analysis component that can be run after the GSI variational/hybrid minimization steps. The HM-analysis is a non-variational update that explicitly builds or removes cloud and precipitating hydrometeors, and correspondingly adjusts moisture and temperature fields to support the cloud hydrometeor changes. In order to improve the analysis and prediction of clouds and precipitation, the HM-analysis is being transitioned to a variational and then hybrid ensemble-variational framework within GSI. This presentation will demonstrate initial work on the formation of new hydrometeor observation operators, the development of cloud control variables, and establishing static background covariances for cloud hydrometeors. The new HM-analysis will be tested on both the RAP and the HRRR.

This presentation will also discuss recent research focused on improving the latent heating specification in the HRRR, based on the use of radar reflectivity observations. Finally, retrospective case studies with the HRRR will be presented to compare reflectivity forecasts.