高阶递归滤波在一次暴雨预报中的数值试验

Numerical experiment of high-order recursive filters in the forecast of a heavy rain

为了考察具有多种特征尺度四阶递归滤波器在极端暴雨事件预报中的应用效果和能力,利用GRAPES三维变分同化系统,对2007年7月16—19日的川渝暴雨过程进行了模式分析。共开展了3组模拟试验:第一组试验在模式中使用单一特征尺度的一阶递归滤波;第二组试验在模式中使用由3种不同特征尺度的四阶滤波拟合而成的四阶递归滤波;第三组试验在模式中使用由3种不同特征尺度的一阶滤波拟合而成的一阶递归滤波。结果表明,具有多种特征尺度的递归滤波的预报效果要好于单一尺度的递归滤波。相比而言,由3种不同特征尺度的四阶递归滤波对预报的改进最大,不论是降水的落区还是降水的强度都有很大程度的改善,对水汽的调节更为理想。

In order to examine the application capability of four-order recursive filters on heavy rain forecast,by using GRAPES-3D Var system,the paper numerically simulate the heavy rainfall in Sichuan Basin during July 16—19,2007.Three experiments were carried out in this study during different periods.The first experiment used one-order recursive filter with one intrinsic scale;the second used four-order recursive filters,which was synthesized from three four-order recursive filters with different intrinsic scale respective;the last experiment used one-order recursive filter,which was synthesized from three one-order recursive filter with different intrinsic scale respective.The results show that the recursive filter with multiple intrinsic scales is better than the one with only one intrinsic scale.The heavy rain process is closer to the reality in the second experiment.The forecast of precipitation region and amplitude has been greatly improved.Its adjustment of vapor is more ideal.