集合卡尔曼滤波同化多普勒雷达资料的观测系统模拟试验

本文将集合卡尔曼滤波同化技术应用到对流尺度系统中,实施了基于WRF模式的同化单部多普勒雷达径向风和反射率因子的观测系统模拟试验,验证了其在对流尺度中应用的可行性和有效性,并对同化系统的特性进行了探讨。试验表明:WRF-EnKF雷达资料同化系统能较准确分析模式风暴的流场、热力场、微物理量场的细致特征;几乎所有变量的预报和分析误差经过同化循环后都能显著下降,同化分析基本上能使预报场在各层上都有所改进,对预报场误差较大层次的更正更为显著;约8个同化循环后,EnKF能在雷达反射率、径向风观测与背景场间建立较可靠的相关关系,使模式各变量场能被准确分析更新,背景场误差协方差在水平方向和垂直方向都有着复杂的结构,是高度非均匀、各项异性和流依赖的;集合平均分析场做的确定性预报在短时间内能较好保持真值场风暴的细节结构,但预报误差增长较快。

Assimilation of Doppler Radar Observations with an Ensemble Kalman Filter: OSS Experiments

The feasibility and availability of applying ensemble Kalman filter (EnKF) technique in convective scale systems were demonstrated by an observation system simulation experiment (OSSE) in this paper, which assimilates simulated radial velocity and reflectivity of one Doppler radar with an EnKF assimilation system based on WRF model. The experiment shows: the assimilation system has the ability to accurately analyze the detailed characters of flow fields, thermodynamic and microphysical fields of the storm, forecast and analysis errors of almost all variables decrease significantly after assimilation cycles, and forecast fields of all levels can be improved by assimilation, especially on levels of larger errors. Reliable correlativity between radar reflectivity, radial velocity observations and forecast fields can be established after 8 assimilation cycles, and the background error covariance has a complex structure and is highly inhomogeneous, anisotropic and flow dependent. Determinate forecast of ensemble mean field can keep the detail characters of truth storm in short period but errors grow fast.