Application Of Physical Filter Initialization In 4DVar

Generally,the results of data assimilation are not well balanced dynamically due to the uncertainty of background errors and the errors of observation or the model itself.So,initialization methods have been introduced to remove spurious gravity waves from the analysis.One of the initialization methods is digital filter initialization(DFI),which has been used in operational forecast systems,though its physical meaning is not well understood.Other methods eliminate high-frequency noises in optimized initial conditions by introducing physical constraints,such as model constraint scheme,which minimizes time tendency of model variables in order to achieve an optimized initial condition.In this study,a physical filter initialization(PFI)scheme,based on model constraint scheme,is implemented in the fourdimensional variational data assimilation(4DVar)system of the Weather Research and Forecasting(WRF)Model.The impacts of the PFI scheme are examined by both singleobservation and real-data experiments.The rationality and efficiency of PFI scheme have been verified through the contrast of the two methods.The conclusions are as followed:1 In the process of the traditional 4DVar,the high-frequency noises exist in the analysis field due to the uncertainty of background errors and other factors after assimilating observation information.Such high-frequency noises have certain impact on simulation and forecast.However,in the process of PFI-4DVar,well balanced variables could be achieved in the analysis,thanks to the weak constraint of PFI scheme,which is the dynamic and physics process of numerical model.And the PFI scheme can eliminate high-frequency noises effectively.2 PFI scheme can obtain flow-dependent analysis increments in the analysis,and keep this structure during integration process.Well balanced variables in the analysis retain their compatibility and flow-dependent pattern in the process of integration,without the outward propagation of high-frequency oscillation.3 It could be concluded that precipitation forecast is improved with both traditional 4DVar and PFI-4DVar.What is worth mentioning,the precipitation forecast during the first few hours is improved significantly with PFI-4DVar in terms of precipitation distribution and TS scores thanks to shorter spin-up time.4 By weakening the function and interference radius of BE matrix,the traditional BE matrix would still pass the information of background errors and observation in large scale,and lead to an unbalanced analysis field.PFI-4DVar passes the information more reasonably,and make the variables in analysis field follow physics and dynamics model.Thus,the initial condition achieved from PFI-4DVar is well-balanced and the spin-up time would be shorten.It is preliminarily confirmed that the PFI scheme may play a role of BE matrix.