ESRL/PSL Seminar Series

Abstract

GSI-based hybrid ensemble-variational data assimilation (DA) system improved US NWS global forecast significantly since it was implemented at NCEP in May 2012 through multi-institutional collaboration. The 3-dimensional hybrid (3DEnVar) system was also being extended to 4-dimension (4DEnVar). Different from traditional 4DVar, 4DEnVar conveniently avoids tangent linear and adjoint of the forecast model. Experiments suggested 4DEnVar can further improve the global forecasts compared to 3DEnVar.

In the meantime, research and development have also been made to extend the GSI-based hybrid DA system for regional modeling systems. The hybrid DA system is extended with the operational Hurricane Weather Research and Forecast (HWRF) modeling system to improve high-resolution, convection-allowing tropical cyclone prediction. The performance of the newly extended GSI based hybrid DA system for HWRF was studied by conducting experiments with retrospective hurricane cases assimilating airborne radar observations. The performance of the HWRF hybrid DA system ingesting HWRF's own EnKF ensemble (denoted as "hybrid") was compared with that ingesting GFS ensemble (denoted as "hybrid-GFSENS"). Verification against independent in situ flight level data and remotely sensed observations such as SFMR wind speed and HRD radar wind composite shows that the analyses provided by the "hybrid" captures the hurricane structure much better than the hybrid-GFSENS and GSI 3DVar. Forecasts initialized from the analysis of the "hybrid" produce smaller track errors and better MSLP and max wind relationship than that initialized from hybrid-GFSENS and GSI 3DVar.

The GSI-based hybrid DA system is also extended with the WRF ARW model for multi-scale assimilation of both conventional in-situ observations and radar observations to improve convective scale hazardous weather forecasts over the Continental US (CONUS). We first focus on systematic comparison of GSI based 3DVar and EnKF in the context of multiscale data assimilation where scales ranging from convective scales to synoptic scales were resolved by both the model and the observations. Diverse cases are used to obtain robust averaged results. The cases include many examples of both discrete cellular convection and organized Mesoscale Convective Systems (MCSs). Averaged over 10 cases, convective scale precipitation forecasts initialized by GSI-based EnKF were much more skillful than GSI-based 3DVar. The positive impact of assimilating radar data lasted for 5 hours when assimilated by the EnKF. In comparison the impact only lasted for 1 hour when assimilated by 3DVar. In addition, experiments were also conducted with the newly extended GSI based hybrid DA system for the May 8th 2003 Tornadic supercell. It was found that using the GSI based hybrid DA, the tornadic supercell was maintained during the one hour forecast period and the predicted track was consistent with the actual tornado track. In comparison, for GSI 3DVar, the tornadic supercell was not maintained.