Impact of different satellite radiance data sets using 3D-Var and hybrid variational/EnKF data assimilation systems in the Rapid Refresh

As part of the NOAA operational Rapid Refresh (RAP) hourly updated prediction system, satellite radiance data are assimilated using standard procedures within the Gridpoint Statistical Interpolation (GSI) 3D-VAR analysis package. Within the community-based GSI code, a hybrid variational/Ensemble Kalman Filter (EnKF) data assimilation system has been developed. This more advanced assimilation procedure has been implemented within a real-time experimental version of the RAP, using ensemble information from the 80-member global ensemble data assimilation system, with operational implementation at NCEP expected early in 2014

In this presentation, we will report on an ongoing investigation of the impact from the assimilation of different satellite radiance data sets on short-range RAP forecasts, using both the old 3D-VAR RAP model system and the new hybrid RAP model system. Our previous studies have been demonstrated that satellite radiance data have positive impact for the short-range forecasts within the RAP model system. Systematic tests of the impact of radiance data assimilation for various instruments and channels within RAP are being conducted. Using a retrospective testing environment, the impact from a variety of satellite radiance data sets (AMSU-A, HIRS, MHS, AIRS, GOES, and possibly data from NPP) are being examined (individually and in combination) relative to a control run assimilating only conventional observations. Regional radiance bias correction in the hybrid system will also be evaluated. The impact differences for the satellite data between the 3D-VAR RAP and the hybrid EnKF RAP will be documented through these retrospective runs.

The RAP has a very short observation cut-off time of about +35 minutes after the initial time for given model run. For satellite data, the very short observation cut-off time and the long data latency pose significant challenges for real-time data coverage. The retrospective runs are conducted in an ideal condition (no data latency and cut-off issues). Additional experiments will examine the impact from the satellite radiance data when assimilated in a real-time configuration (less coverage than the idealized configuration). In addition, in order to improve the real-time satellite data coverage, we are examining the possible use of short latency direct readout data for the real-time RAP. As part of this process, the Regional ATOVS Retransmission Services (RARS) real-time data files for the RAP will be assimilated and the impact will be assessed.

At the conference, we will report on progress to date in all aspects of this work, including assessing the impact from various data sets in the RAP hybrid system, documenting the impact differences between the 3D-VAR and hybrid EnKF systems, and evaluating the impact from the real-time direct readout data vs. the current operationally used data.