Assimilation of Pseudo Water Vapor and Radar Data for Convective-Scale NWP in a Variational Framework

In the past decade, many studies have been demonstrated that using Doppler radar data to initialize convective-scale numerical weather prediction (NWP) models could help improve severe weather forecast. But the assimilation of radar data into convective scale NWP is still challenging because radar cannot directly observe many model variables, especially moisture and temperature variables which are fundamentally important for convective scale NWP. To improve the severe thunderstorm prediction, a new scheme is developed to better initialize convective scale NWP model. The first step is to identify the deep moisture convections by using the Vertically Integrated Liquid water (VIL) calculated from radar reflectivity (Zhang and Qi 2010). Then pseudo water vapor observations are derived based on VIL thresholds inside the deep moisture convective columns. The final step is to assimilate the derived pseudo water vapor observations into convective scale NWP model along with radar radial velocity and reflectivity in a 3DVAR framework. The performance of the scheme is examined, with or without the pseudo water vapor assimilation, for two high impact severe thunderstorm events: the 24 May 2011 Oklahoma tornadic supercells and the 16 May 2017 Texas and Oklahoma tornadic supercells. The analyses and forecasts of these supercells are quantitatively improved in both cases, including more realistic forecast of reflectivity patterns, updraft helicity tracks, which better match the tornado observation tracks when the pseudo water observations are assimilated. Some sensitivity experiments are done with assimilation cycles (5 minutes, 10 minutes, 15 minutes, 30 minutes), and VIL threshold values. The results will be quantitatively evaluated by equitable threat scores and other verification tools.