Impact and Performance Evaluation of Hyperspectral (CrIS, IASI, and GIIRS) Sounding Retrieval in Numerical Weather Prediction (NWP) Systems: Study of Short-Term and Short-Range Quantitative Precipitation Forecasts (QPFs) in the CONUS and Southeast China

High-spatial (2km horizontally, 56 levels below 100hPa) resolution atmosphere temperature and water vapor content profiles are generated using Hyper-spectral LEO and GEO satellite radiance (CrIS, IASI, GIIRS) and multi-spectral satellite radiance data (AHI for Himawari-8, ABI for GOES-16). By assimilating sounding retrievals of temperature and water vapor into a data assimilation and prediction system with 9km horizontal resolution using GSI data software (3DVAR) and WRF mesoscale forecast model via three cases (two tornado cases in CONUS, one severe precipitation in China), aspiring impacts are detected in quantitative precipitation forecast (QPF) and severe weather prediction: (1) precipitation’s spatial correlation and Critical Success Index (CSI) are improved by 0.1 compared to the control experiment; (2) high-impact zone of severe convective incidents (tornado, hail, etc.) can be labeled clearly via Significant Tornado Parameter (STP).

Based on the results from previous experiments, a new quality control scheme is generated with the aim of using the newly-generated sounding retrievals more effectively. Moreover, an operational Numerical Weather Prediction (NWP) system was set up in order to conduct further reliability tests of the sounding retrievals. By analyzing NWP’s output and logs, positive impact of assimilating sounding retrieval is confirmed and NWP system’s adaptability to these data is also promising : (1) compared to observations, initial condition’s root mean square error (RMSE) of temperature acquires a decrease by approximately 60% after data assimilation with reference to the one before data assimilation; (2) Average accept ratio of sounding retrieval is around 40% with a maximum of 48.2%.