修正数值天气预报非系统性误差的方法研究

项目名称： 修正数值天气预报非系统性误差的方法研究

项目编号： No.40875063

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 邵爱梅

作者单位： 兰州大学

项目金额： 40万元

中文摘要： 通过对预报场的后处理来修正预报是提高数值预报精度的一种有效途径。本项目研究如何通过变分方法实现对中短期数值天气预报的非系统性误差的修正。基本的技术路线是通过对过去的预报场和分析场样本的统计分析确定用于推断预报误差的预报场形态（它可以是预报场的某种函数），利用奇异值分解的方法在误差场和预报场（包括它的变形）之间建立映射关系，利用变分方法给出预报误差的最有估计达到修正预报的目的。主要研究内容包括有两个代表性的全球和区域中短期预报误差的统计特征与预报场的关系；寻求用于推断误差场的最佳的预报场变型；用于估计预报误差的变分方法。

中文关键词： 预报误差；订正；变分方法；奇异值分解

英文摘要： Post-processing of forecasts is an effective approach to correct the forecasts and to further improve the accuracy of numerical predication. The project will focus on how to use variational method to correct the non-systematic error in the middle-short-term weather precition. The main technical routine is designed as follows: first the forecasting forms, maybe as a kind of function of forecasts, is determined by the statiscal relationship between the forecast error and forecasts (including their deformation) is established by SVD techinique. Then variational method is employed to obtain the optimal estimation of forecast error to correct the forecasts. In virtue of a representational global model, the destination of this study is to obtain the characteristics of forecast error and its relationship to forecasts, and to seek the optimal forecast deformation and to develope a adaptive variational method to estimate the forecast error.

英文关键词： forecast error; correction; variational method; single value decomposition