Impact of adaptively thinned GOES-16 all-sky radiances in an ensemble Kalman filter based WoFS

Abstract

The ability to forecast severe weather events can be improved by assimilating GOES-16 satellite observations into high-resolution convective scale numerical weather prediction models. In this study, multiple sensitivity experiments were performed for four severe weather events that occurred on 7 May, 21 June 2020; and 7, 23 May 2021. Our goal is to assess how the application of data thinning techniques on the GOES-16 all-sky radiance data assimilated by the National Severe Storms Laboratory's ensemble-based Warn-on-Forecast System (WoFS) affect subsequent weather forecasts. A total of three experimental configurations are considered for this analysis: (1) ABI-TH, an adaptive-based data thinning method used in the cloudy region; (2) ABI-ALL, in which all available all-sky radiances were assimilated; and (3) CNTL, a control experiment that used cloud water path (CWP) data but assimilated no radiance data. All three experiments assimilated conventional, radar reflectivity, and radial velocity observations. For each experiment, multiple sensitivity tests are performed to assess the effect of thinning the GOES-16 all-sky radiance data assimilated by the WoFS using an object-based verification method in which model-simulated composite reflectivity and 2–5 km Updraft Helicity (UH) fields were compared to Multi-Radar Multi-Sensor (MRMS) reflectivity and rotation objects. In addition, synthetic satellite images from the WoFS output are generated and compared with the synthetic visible and infrared images from GOES-16. Results show that the ABI-TH approach produced better forecasts relative to ABI-ALL for high-impact weather events. The results from ABI-TH are also similar to results from CWP assimilation (CNTL).