Research On Data Assimilation Of Satellite Wind For Typhoon Strong Convection Weather

Accurate forecasts of typhoon is of great importance to disaster reduction and relief for coastal cities and can mimimize losses due to typhoon preparation and evacuation as well as to destruction.Sea surface wind vector is one indispensable factor for numerical simulation and prediction of typhoon,while the wind observations coming from remote sensing observation of satellites are the major sources of the sea surface wind.The assimilation of these wind observations is crucial to improve the precision of typhoon forecast.However,due to defects in the assimilation process,a great deal of wind observations are not fully used.Besides,imprecise background,corvariance that lack of flow-dependent information and inappropriate definition of observation error of high wind observations all will adversely affect the analysis.This thesis focuses on the data assimilation of satellite wind observation in servere convection weather,aims at the problems during the process of thinning,quality control,definition of observation error and chose of the scheme of data assimilation,and presents our solutions to these problems which have improved the utilization of satellite wind observations and reduced the forecast error of typhoon track and intensity predictions.Traditional thinning process may cause a loss of wind observation in key areas such as vortex center of typhoon.In terms of SAR wind observed which could acquire fine structure of typhoon,the thinning process will lead to discard of lot of valuation information.This thesis proposed a new thinning method which considerates the structure of wind field.We also modified the superobbing with a feature box.The simulation experiment has shown the efficiency of these two methods.Although there are numerous methods which could acquire high wind speed observations for typhoons,these wind observation were not effectively used due to the rejection of observations with great departure to the background,which somehow may be caused by bad background and failure of quality control process of data assimilation.We introduced two quality control schemes,one is the quality control based on Huber norm distribution and the other is an adaptive quality control scheme.These two schemes all can make good use of high wind speed observations while ensure the stability of the minimization of data assimilation.The adaptive quality control scheme make sure that wind observation with big innovation will have a minor weight than that of Huber norm distribution scheme,avoiding impact to the model caused by big analysis increment,and it has shown obvious improvement on typhoon intensity predictions.3DVAR could not present the rapid evolution of error of servere convection weather due to the use of climatological background error covariance.Short-term ensemble forecasts can provide flow-dependent corvariance and adjusted background for data assimilation.However,ensemble data assimilate takes some cycles of assimilation process to make it more consistent with the model.We designed and implemented a complete hybrid data assimilation specially for the assimilation of wind observation under typhoon strong convection weather based on the WRFDA system.We made a detail discussion on how data assimilation schemes such as variantional data assimilation,ensemble data assimilation and hybrid data assimilation would impact the forecast of typhoon.And we found a new way to definite the error and correlation of the wind components based on the law of error propagation,and made it possible for direct assimilation of correlated wind components.The simulation experiment of typhoon has verified the efficiency of this new method,and has shown a improvement of 14% and 17% for 24 hours forecasts of minimum pressure and maximum wind speed of typhoon,respectively.