集合卡尔曼滤波同化多普勒雷达资料的数值试验

Numerical Experiment of Assimilation of Doppler Radar Data with an Ensemble Kalman Filter

利用集合卡尔曼滤波(EnKF)在云数值模式中同化模拟多普勒雷达资料,并考察了不同条件下EnKF同化方法的性能。结果显示,经过几个同化周期后,EnKF分析结果非常接近真值。单多普勒雷达资料EnKF同化对雷达位置不太敏感,双雷达资料同化结果在同化的初期阶段比单雷达资料同化结果准确。同化由反射率导出的雨水比直接同化反射率资料更有效,联合同化径向速度和雨水有利于提高同化分析效果。协方差对EnKF同化效果起着非常重要的作用,考虑模式全部预报变量与径向速度协方差的同化效果比仅考虑速度场与径向速度协方差的同化效果好。雷达资料缺值降低了同化效果,此时增加地面常规观测资料的同化可以明显提高同化分析效果。EnKF同化技术对雷达观测资料误差不太敏感。初始集合对同化分析有较大影响。EnKF同化受集合大小和观测资料影响半径。同化对模式误差较敏感。利用EnKF同化双多普勒雷达资料,分析了一次梅雨锋暴雨过程的中尺度结构。结果表明,EnKF同化技术能够从双多普勒雷达资料反演暴雨中尺度系统的动力场、热力场和微物理场,反演的风场是较准确的,反演的热力场和微物理场分布也是基本合理的。中低层切变线是此次暴雨的主要动力特征,对流云表现为低层辐合、高层辐散并有垂直上升运动伴随,其热力特征表现为低层是低压区,高层为高压区,中部为暖区而上、下部为冷区,水汽、云水和雨水分别集中在对流云体内、上升气流区和强回波区。

The ensemble Kalman filter （EnKF） is applied to assimilation of simulated Doppler radar data in a cloud model and its performances under different conditions are investigated. The results demonstrate that the EnKF assimilation method is able to produce analyses that accurately approximate the true state after several assimilation cycles. The EnKF assimilation of single radar data is slightly influenced by the radar location. More accurate analyses are obtained during the earlier period when dual-Doppler data are assimilated. It is also found that assimilating the rainwater mixing ratio obtained from the reflectivity results in a better performance of EnKF than directly assimilating the reflectivity. When both radial velocity and rainwater mixing ratio are assimilated, the quality of the EnKF analyses is improved. The covariances between the observed variables and the state variables are important to the quality of the analyses. The analysis error increases when only the covariances of radial velocity with velocities are estimated. As the amount of the observations decreases, the performance of the EnKF analyses is degraded. However, the EnKF can again provide accurate estimates by adding assimilation of the hypothetical surface wind and temperature observations. The EnKF technique is not especially sensitive to the radar observation errors. The initialization of the ensemble has an effect on the quality of the analyses, as do the ensemble size and the radius of influence for the observations. The assimilation is sensitive to the model errors. The EnKF is applied to dual-Doppler radar data of a Meiyu rainstorm. Results demonstrate that the EnKF assimilation method is able to retrieve the detailed structure of wind, thermodynamics and microphysics from dual-Doppler radar observations. The retrieved wind fields agree with the dual-Doppler synthesized winds and are accurate. The distributions of the retrieved perturbation pressures, perturbation temperature and microphysics are also reasonable through the examination of their physical consistency. The wind shear at middle and lower levels is the primary dynamical characteristics of the Meiyu heavy precipitatiorL The convective rainfall is often related to lower level convergence and upper level divergence coupled with the updraft. The convective system is characterized by high pressure at lower level and low pressure at upper level, associated with warmer at middle level and colder at lower and upper levels than the environment. The water vapor, cloud water and rainwater are associated with the convective cloud, the updraft and the reflectivity, respectively.