Cooperative Institute for Research in the Atmosphere

Satellite radiance data assimilation within the NOAA hourly Rapid Refresh and High-Resolution Rapid Refresh

Satellite radiance data have been shown to have consistent positive impact with statistical significance within the NOAA Rapid Refresh (RAP) hourly updated model system.  RAP version 4 (RAPv4) was implemented operationally at the National Centers for Environmental Prediction (NCEP) in July 2018.  This implementation included a significant radiance upgrade package with greater use of direct broadcast/readout data as well as assimilation of data from new sensors.  The RAP uses the Gridpoint Statistical Interpolation (GSI) hybrid variational/Ensemble Kalman Filter (EnKF) data assimilation system, with ensemble information for the regional assimilation coming from the 80-member global ensemble data assimilation system.

The next satellite assimilation upgrade package has been recently finalized for the coming RAP version 5 (RAPv5) NCEP operation upgrade (planned for spring 2020).  This RAPv5 radiance upgrade package includes the assimilation of the GOES-16 Advanced Baseline Imager (ABI) infrared radiance data, the Cross-track Infrared Sounder Full-Spectral-Resolution (CrIS-FSR) data both from S-NPP and NOAA-20, and the Advanced Technology Microwave Sounder (ATMS) data from NOAA-20.  The initial ABI radiance assimilation work started from the code development to read in ABI radiance data into the GSI and single case GSI study to identify the ABI data impact from the three water vapor channels, then ABI O-B bias and standard deviation evaluation for different cloud masks and different surface types within the RAP domain was performed.  Research work associated with ABI error tuning and quality control was conducted through hourly RAP retrospective runs with forecast verification to maximize the ABI data impact.  In addition, a series of RAP retrospective runs were performed to evaluate the forecast impact from these new instruments/data separately and/or combined together.

Use of polar orbiter satellite radiance data in rapidly updated regional models has traditionally been limited by data latency issues combined with the very short data cutoff window for these models.  For the hourly RAP mesoscale model system, the data cutoff time is ~ 30 min. resulting in limited data usage for standard data delivery methods.  The availability of direct broadcast data offers the potential for improvements in the percent of total polar orbiter data being assimilated into rapidly updated regional models.  The data impact from assimilation of the direct broadcast/readout radiance data has been evaluated for the RAP model through extensive retrospective experiments with the short-term forecast verification against the traditional radiosonde observations as well as the CrIS observed radiance observations.

At the seminar, I will present the overall and individual radiance data impact within the RAPv4, and then will focus on the recent radiance updates for the coming RAPv5, including GOES-16 ABI, NOAA-20 CrIS-FSR/ATMS data assimilating.  The impact from direct broadcast radiance data within RAP will also be shown as well as some preliminary work on radiance assimilation within the High-Resolution Rapid Refresh (HRRR)/Alaska.