风暴尺度天气下利用集合卡尔曼滤波模拟多普勒雷达资料同化试验I.不考虑模式误差的情形

本文在假定模式无偏差的情况下, 利用一次风暴过程的模拟多普勒雷达资料进行一系列风暴天气尺度的集合卡尔曼滤波资料同化试验, 检验集合卡尔曼滤波在风暴天气尺度资料同化方面的效果, 并验证各集合卡尔曼滤波参数对同化效果的影响。试验结果表明, 集合卡尔曼滤波能有效地应用于风暴尺度的资料同化; 40个集合成员以及6 km的局地化尺度能较好地滤除采样误差造成的虚假相关, 同时可以将观测信息传递到无观测的模式格点; 利用背景场加上空间平滑的高斯型随机扰动生成初始成员的方式较未经过平滑的方式有更好的分析效果; 背景场扰动方法能够提高样本的离散度; 只同化反射率的同化试验表明, 反射率的同化效果较明显, 也证明了集合卡尔曼滤波在非常规资料同化中的作用; 增加径向风资料同化的效果优于只进行反射率同化的结果。

Storm-Scale Ensemble Kalman Filter Data Assimilation Experiments Using Simulated Doppler Radar Data. Part I: Perfect Model Tests

This study tested the performance of ensemble Kalman filter (EnKF) under the perfect model assumption for storm-scale data assimilation of simulated Doppler radar observations. A series of observation system simulation experiments have been carried out to check the filter performances of EnKF in different configurations. It is found that the EnKF is very effective in the assimilation of storm-scale under the perfect model assumption if the configuration is properly set. An ensemble of 40 members and localization scale of 6 km are suitable to our experiments. To initialize the ensemble, adding smoothed Gaussian initial perturbations is superior to non-smoothed ones.Inflation that augments ensemble spread is necessary.Although the observation operator for reflectivity is strong nonlinear, assimilation reflectivity data only can bring positive effect to analysis. If both the radial velocity and reflectivity observations are assimilated, analysis will be better than that from single reflectivity assimilation.