江西区域多元变量背景误差协方差特征分析

The Analysis of Multivariate Background Error Covariance in Jiangxi Province

江西省暴雨灾害频繁发生,改善同化系统性能是提高数值天气预报水平的有效手段,构建合理的背景误差协方差是做好资料同化的关键工作。基于WRF模式的江西区域一个月控制预报为样本,计算得到多元变量相关的背景误差协方差,分析其平衡约束特征、特征值、特征向量以及特征长度尺度。结果表明,江西区域模式层中的低层和高层风场辐合、辐散分量的作用更大,各个变量对水汽场的贡献也集中在低层和高层,其中温度场起主导作用;模式层高层温度场的模拟效果偏差,各个变量的垂直相关性较大;相较于风场,温度场和水汽场在水平方向影响范围小,具有较强的局地性。

Heavy rainstorms occurred frequently in Jiangxi province.Ameliorated performance in the assimilation system was an effective way to improve the numerical weather prediction（NWP）.A reasonable background error covariance（BE）modeling was significant to obtain a good data assimilation result.The multivariate background error covariance（MBE）was calculated based on the one-month forecast in Jiangxi province,and the major parameters of BE,such as correlation coefficients,eigenvalues,eigen-vectors and eigen-length scales were compared and analyzed.Results indicated that the convergence and divergence of wind field played a more important role at bottom and top of the model,where each variable made a significant contribution to the water vapor field,especially the temperature field.In addition,the temperature field simulation at the top of model was not satisfactory and the vertical correlation among each variable was relatively strong.The influence of temperature and water vapor field was smaller than that of wind field in the horizontal lengthscales,which presented strong local features.