Testing of a Hybrid 3DEnVar Analysis System by Assimilating Radar and Satellite Data with a Severe Convective Weather Event

Sijie Pan* and Jidong Gao, Xuguang Wang, David J. Stensrud, Thomas A. Jones
University of Oklahoma, University of Oklahoma, National Severe Storm Laboratory, Pennsylvania State University

In the past decade, many studies have been performed aiming to use Doppler radar and satellite data to initialize convective-scale numerical weather prediction (NWP) models. However it still remains a challenging problem. The radar can only observe radial velocity and reflectivity which are related to wind and hydrometeor variables. Other variables including water vapor and temperature which are critical for initializing convective NWP, cannot be provided by radar. In addition, uncertainties in the relationship between observed radiance and the structure of clouds make direct assimilation of satellite radiance data difficult for convective scale. Recently, Jones et al. (2014) demonstrates that assimilating both radar data and retrieved Cloud Water Path (CWP) from satellite data generates reasonable initial conditions for most model variables with an EnKF system. To further investigate this problem, in this study, a hybrid ensemble and variational data assimilation method for convective scale weather events has been developed to assimilate both radar data and retrieved CWP data with the hope of improving storm scale NWP. Because the background error covariances are derived from ensemble forecasts, it is called 3DEnVar method. Frist, the system is tested with simulated radar radial velocity, reflectivity and satellite derived CWP data from a supercell storm. It is shown that assimilating both radar data and CWP data can significantly improve the analyses and forecast results than that by assimilating radar data, or CWP data only. The analysis results can be effectively used for predicting the track and intensity of the supercell thunderstorm. The usefulness of this system for the storm scale NWP will also be evaluated with several real data cases about tornadic severe weather events collected during NOAA Hazardous Weather Testbed Spring experiments.

*email: sijie.pan@ou.edu
*Preference: Poster