PODEn4DVar对松弛系数和局地化半径的敏感性

A parametric sensitivity of the PODEn4DVar to the relaxation coefficient and the localization radii

针对松弛系数和局地化半径的敏感性对PODEn4DVar同化方法性能的影响,以浅水波方程作为预报模型,测试了其对不同松弛系数α和局地化半径R的敏感性,获得了不同模型误差情形下该方法同化时产生的均方根误差变化情况,确定了不同模型误差时的最优松弛系数α和局地化半径R。实验结果表明,在模型有误差时,PODEn4DVar同化方法对参数地选取很敏感。选取最优参数值后,同化精度有了明显提高。

The choices of the relaxation coefficient and the localization radii have an impact on the performance of the PODEn4DVar assimilation method. By applying the PODEn4DVar with a two- dimensional shallow-water equation model, the sensitivity of the PODEn4DVar to the relaxation coefficient a and the localization radii R is examined, and then the root mean square error of the assimilated results is obtained in the case of different model error in the data assimilation process. Optimal relaxation coefficient c~ and the localization radii R of PODEn4DVar in the imperfect model are determined through its root mean square error. The results show that the PODEn4DVar approach is sensitive to the parameters. Its assimilation precision and assimilation performance have been improved significantly after the optimal value of parameters is selected.