EnKF协方差膨胀算法对雷达资料同化的影响研究

Impact of inflation methods on radar data assimilation using EnKF

基于集合卡尔曼滤波(EnKF)方法同化模拟雷达径向风和回波,引入具有时空自适应理论优势的贝叶斯膨胀算法,通过与常数膨胀算法的对比,分析了两种协方差膨胀算法对EnKF同化效果的影响。结果表明:在对流区域的北侧,由贝叶斯膨胀算法分析得到的回波在水平和垂直结构上均增强;在对流区域,由贝叶斯膨胀算法分析得到的各变量的集合离散度增大,均方根误差减小,水平和垂直速度增大,冷池强度减弱;模拟还发现贝叶斯膨胀算法提高了强对流系统的模拟效果,回波强度增强,阵风锋区内水平和垂直风速增大。这表明贝叶斯膨胀算法有效地改进了基于常数膨胀算法的EnKF同化雷达资料的效果。

Based on the simulated radar reflectivity and radial velocity data assimilated by EnKF, through comparing the introduced Bayes inflation method with the advantage of space-time adaptive theory to constant inflation method, the assimilation effects of two covariance inflation methods on EnKF were analyzed. Results show that horizontal and vertical structures of analysis reflectivity by Bayes inflation ex- periment are stronger in northern convective branch. In convective region, the root mean square error is lower in Bayes inflation experiments while the spread is higher; Bayes inflation experiment has bigger horizontal and vertical wind while has weaker strength of the cold pool. Results of simulations indicate that the Bayes inflation experiment improves largely the condition of the convective system in the respect of magnitude and position. The horizontal and vertical wind speed is bigger. These evidences mean that Bayes inflation helps to improve the effects of EnKF radar data assimilation.