4D Envar Data Assimilation For The High Resolution Rapid Refresh (HRRR) System

Abstract

The inclusion of the ensemble based background error covariance in the traditional 3D variational (3DVar) data assimilation system has resulted in improved weather forecasts as observed from the implementation of the 3D hybrid Ensemble-Variational (EnVar) technique for NOAA operational forecast systems (e.g., the Global Forecast System (GFS), Rapid Refresh (RAP) system). Since then, significant progress has been made to extend this hybrid capability to include the time dimension. The 4D hybrid EnVar technique applies the ensembles at different forecast times within the analysis time window and therefore provides time-variant ensemble covariance information. This paper describes the efforts at the Developmental Testbed Center (DTC) to apply the 4D hybrid EnVar to high-resolution regional weather forecasts. In this work, the 4D hybrid EnVar technique is applied for regional applications based on NOAA’s 3-km High Resolution Rapid Refresh (HRRR) forecast system. Currently, HRRR is a real-time 3-km resolution, hourly updated, cloud-resolving, convection-allowing atmospheric model, initialized by 3km grids with radar data assimilation. This paper describes an experimental 4D hybrid EnVar system built for HRRR. The results from various sensitivity experiments and the comparison between the experimental 4D and current 3D hybrid EnVar will be presented. In addition, working areas discovered during this effort will also be presented to further improve the application of this technique in mesocale and convective scale weather forecasts.