The Study Of Assimilating Doppler Radar Data With Ensemble Kalman Filter

With the continuous development of numerical weather forecast, current mesoscale numerical model has initially had the ability to forecast strong convective weather process, and the effect largely depends on whether the model can provide a correct initial and boundary condition. Because of its high temporal and spatial resolution, Doppler weather radar has become the most effective method in monitoring and warning severe convective weather. It also become a hot direction in mesoscale data assimilation that how to effectively integrate Doppler radar detection into numerical model and enrich the model’s mesoscale information and improve the quality of initial condition.The research, based on the previous studies on WRF-EnSRF, proposes a comprehensive quality control program of Doppler radar data which on the basis of multipass automatic dealiasing and "super observation" algorithm, meanwhile, the WRF-EnSRF’s ensemble forecast module is developed into a two-way nested, mixed microphysical process.On the foundation of the new system, the research verifies a new initial disturbance scheme with controllable scale and related Variables. To solve the problems existing in the current EnKF gain estimate method, a new gain estimate method which on the basis of Gateaux differentiation is also put forward. The results show that:(1) Under the same Spin-up time, the physical constraint perturbation scheme does better than RandomCV and stochastic perturbation scheme in gaining a better background error covariance structure, its prior RMSE of radial velocity and reflectivity is significantly smaller than the other two schemes. In typhoon numerical experiment, after the radar data assimilation, all three initial perturbation schemes can improve typhoon structure and play a positive effect on the adjustment of typhoon temperature and pressure structure, the typhoon path and precipitation forecast, but the new scheme’s effect is more distinct in forecasting typhoon track, center pressure, maximum wind speed and precipitation.(2) Kalman gain estimate method which based on Gateaux differentiation displays a similar effect to that based on tangent linear operator, but the former can avoid the complex problem in constructing tangent linear operator. The results also show that EnKF gain estimate method has problems in higher ensemble dispersion and lower space error covariance in early analysis circulation because of the deficiencies of linearization, thus resulting in the significant difference in final analysis effect when contrasting with the new method.