Improving Tropical Cyclone Forecasts Using The GSI Based Hybrid3DEnsVar Data Assimilation Method:a Study With Hurricane ISSAC2012and Airborne Radar Data

As better initial conditions have become a key problem that constrains the predictability of the Numerical Weather Forecast systems today, the importance of the Data Assimilation has been increased. While the most popular data assimilation methods are still Variational methods and Ensemble Filter methods, a new generation of data assimilation called Hybrid, which combines the advantages of both Variational methods and Ensemble Filter methods has been proposed and quickly became a new popular. In the summer of2012, National Centers for Environmental Prediction (NCEP) has made the GSI based Hybrid Ens Variational data assimilation operational in the Global Forecast System (GFS). However, there is no research into the HWRF regional model using this new hybrid data assimilation method to the author’s knowledge.Meanwhile, in the study fields of Tropical Cyclone, researchers have gained a lot of improvements in the track predictions but little on intensity forecasts for recent decades. In order to improve the intensity forecasts, the researches using the inner core data have become more and more popular. The inner core data is supposed to be responsible for improving the structure of the TC and also improving the intensity forecast skills. There have been a lot of researches on the using of Airborne Tailed Doppler Radar data, but mainly focused on the EnKF or3DVAR method.Thus, this paper is going to use the GSI based hybrid3DEnsVar data assimilation method to do some experiments in a2012Hurricane level Topical Cyclone-Isaac with TDR data. And a series of sensitivity tests have been done. Here are the main conclusions,(1) The GSI based hybrid3DEnsVar data assimilation method can assimilate the TDR data effectively in the HWRF regional model, it can greatly improve the TC track and intensity forecast compared to the official forecasts from the operational HWRF;(2) The GSI based hybrid3DEnsVar data assimilation method is obviously better than the operational3DVAR data assimilation method when assimilating the TDR data, and a little more accurate than the EnSRF;(3) Hourly assimilation is almost equal to the penetrate leg only assimilation method when assimilating the TDR radial velocity data, but the6-hourly assimilation shows less improvements compared with the forecast skills of the Hybrid methods.