On A Four-Dimensional Variational Method And An Ensemble Kalman Filter To Assimilate Doppler Radar Data And Their Applications In Rainstorm Mesoscale Structure Retrieval

Two methods: the four-dimensional variational (4D-VAR) and the ensemble Kalman filter (EnKF), are applied to assimilate Doppler radar data in a cloud model. The quality of retrieval and performances of the two techniques in obtaining quantitatively mesoscale information of hazardous weather systems, have been tested using simulated data and real observations. The major contents and conclusions are presented as follows:Firstly, the 4D-VAR technique and the cloud model developed by Sun and Crook (1997) , are used to assimilate data from a simulated convective storm. Experiments demonstrate that the 4D-VAR assimilation approach is able to retrieve the detailed structure of wind, thermodynamics, and microphysics using either dual-Doppler or single-Doppler radar data. The quality of retrieval depends strongly on the magnitude of constraint with respect to the variable.Secondly, the 4D-VAR method is applied to a Meiyu heavy precipitation in the middle Yangtze River reach, observed by the YICHANG- JINGZHOU dual-Doppler radars. The study has clearly demonstrated the ability of the 4D-VAR assimilation approach to drive the quantitative information of rainstorm mesoscale structure from real radar observations. It has been revealed in the assimilation results that the torrential rainfall is caused by the wind shear line at low level. The convective rainfall is often related to lower level convergence and upper level divergence coupled with an updraft, characterized by high pressure at lower level and low pressure at upper level. The temperature structure of convective system shows warmer at middle level and colder at lower and upper level than the environment. The water vapor, could water and rainwater are associated with the convective cloud, the updraft and the reflectivity, respectively.Thirdly, the mesoscale feature of a Meiyu front rainstorm over the lower Yangtze River basin is investigated by the 4D-VAR assimilation of data from the HEFEI- WUWEI -MAANSHAN dual / three -Doppler radars. The assimilations of data from three groups of dual radars, or single radar, produce generally similar mesoscale structures. The convergence line is primary dynamical characteristic of the Meiyu heavy precipitation. The convective system is characterized by lower level convergence and upper leveldivergence associated with the vertical motion. A region of cooling and high pressure occurs at lower and middle levels compared to warming and low pressure at upper level. The convective cloud is dry and cooling at lower and middle levels. The cloud water and rainwater are associated with the updraft and the reflectivity, respectively.Fourthly, Doppler radar radial velocity is 4D-VAR assimilated in a dry version of cloud model which excludes the moist process. A few case studies demonstrate the retrieved low-level wind fields are reliable through the assimilation of Doppler radar data in a dry version of cloud model.Fifthly, we explore the use of an ensemble Kalman filter (EnKF) to assimilate simulated Doppler radar data in a cloud model. It is found that the EnKF assimilation method is able to produce analyses that accurately approximate the true state after several assimilation cycles. Thus, the ability and feasibility of EnKF in assimilating radar data are demonstrated.Lastly, we adopt the EnKF assimilation technique to analyze the mesoscale structures of two rainstorms, which are then compared with those obtained by the 4D-VAR approach. The result indicates that the 3D wind, thermodynamical, and microphysical structures of rainstorm mesoscale convective system can be obtained from Doppler radar data using the EnKF technique. The mesoscale features of convective system revealed by the EnKF analyses, are also basically reflected in the 4D-VAR results although discrepancies existed in the detailed structure of the retrieval results.