Study On Three Dimensional Variational Data Assimilation Of Radar Data

A Doppler weather radar data assimilation scheme with three dimensional variational algorithms is proposed and tested with real observational radar data of both Meiyu torrential rain and typhoon events. The observational operators, commonly used in literatures, describing the physical connection between the fundamental observational elements, i.e. the radial velocity and the intensity of reflectivity, and the model state variables are introduced in this paper. Two approaches dealing with the vertical velocity, which is crucial for meso scale weather system and appears in the observational operators, are compared. One of them introduces the Richardson's equation to compute diagnostically the vertical velocity with horizontal wind, temperature, pressure and its tendency. The other one retrieves vertical velocity from the reflectivity using statistical algorithm and assimilates the vertical velocity taking Richardson's equation as observational operator. It is found that the vertical velocity derived by the former is usually too weak while the later gets the distinct three dimensional convective structure of meso-scale system with strong vertical velocity. It is also found that lower terrain and the diabatic term in the Richardson's equation should not be neglected in assimilating vertical velocity. Vapor and cloud are diagnosticated in order to perfect the convective system. The results of assimilation and prediction experiments with this scheme using operational radar data for a torrential rain event in the Meiyu season and a Typhoon case are analyzed in detail. Both cases show the positive contribution to relief Spin-up phenomenon and improvements of prediction of precipitation caused by meso-P scale convective systems.The innovations and primary conclusions are summarized as follows.(1) The mesoscale 3DVAR radar data assimilation system created in this paper is basically reasonable and effective, single observational datum experiment demonstrates the relations among variables are rational, and the weather cases experiments show assimilating radar data can improve prediction results.(2) Taking Richardson's equation as observation factor to assimilate vertical velocity can advance the dynamical and thermal structure of mesoscale weather system, and the heating term of Richardson's equation has great impact on the results.(3) The weather case experiment of Meiyu torrential precipitation indicates assimilating radar data can ameliorate the prediction of meso-P scale convective weather system on Meiyu front or shear line, relief Spinup phenomenon, and hot-start NWP model.(4) Simulation of a typhoon case presents assimilating radar data advanced prediction of typhoon precipitation, intensity and track, enhanced the meso-P scale system on typhoon spiral rain-band.Because of the fatal differs between large-scale and mesoscale weather systems, such problems as relevant model of background error, representativeness of observation data and observation error structure still need to carry on deeper research.