纬度依赖背景场误差协方差及其对台风同化和预报的影响研究

A STUDY OF LATITUDE-DEPENDENT BACKGROUND ERROR COVARIANCE AND ITS IMPACTS FOR TYPHOON DATA ASSIMILATION AND FORCAST

合理的背景场误差协方差矩阵是做好变分同化的关键。为进一步认识背景场误差协方差矩阵在中尺度预报中的作用,对背景场误差协方差中平衡关系的纬度依赖特征及其对中尺度同化及预报效果的影响进行了研究。为使得平衡关系既包含纬度依赖特征,又降低纬度依赖带来的非平衡的噪音信息,在背景场误差协方差矩阵构造的平衡变换过程中引入了递归滤波的方法,对平衡变换中的回归系数进行了滤波平滑。同化单点观测的理想试验表明纬度依赖的背景场误差协方差具有更合理的动力关系,且滤波后的背景场误差协方差既保持了纬度依赖特征且明显平滑了噪音信息。台风"莫拉克"同化数值模拟试验表明纬度依赖的背景场误差协方差明显提高了36 h及以上的中心海平面最低气压和路径预报的预报效果。

The background error formulation plays a key role in a variational data assimilation system.To evaluate the impact of background error formulation on mesoscale weather forecast,this work investigates the features of latitude dependent Background Error Covariance matrix （B）,and their influence on mesoscale data assimilation and forecast.To keep the latitude dependence and avoid the noise,the regression coefficients were smoothed using a recursive filter.Single observation tests indicate that the latitude-dependent background error matrix B shows more reasonable dynamic characteristics; and the matrix B with filtered regression coefficients not only keeps the characteristics of latitude dependence and but also greatly reduces the noise in analysis increments.The results from typhoon “Morakot” suggest that the 36-h and longer forecast skill was improved evidently using the latitude-dependent matrix B.