提为反问题的数值预报方法与试验 I. 三类反问题及数值解法

提为初值问题的数值预报在通过改进数值模式、观测手段及分析方法而改进预报的同时，仍然面临着两大困难，即模式误差和初值不完整。然而我们有大量的气候演变的历史观测资料，其中蕴含着关于气候系统的信息。本文针对这两个困难，系统地提出充分利用历史资料反演订正模式和初值进而改进数值预报的三类反问题，并给出数值解法。最后将三类反问题应用于一个简单模式进行反演预报的数值试验，其数值试验结果将在本文的第二部分给出。

Methods and Experiments of Numerical Prediction Raised as Inverse Problem. Part I: Three Kinds of Inverse Problems and Numerical Solutions

Although numerical prediction can be improved through improving numerical model, observation method and analysis method, etc., the numerical prediction which is raised as an initial value problem still faces two difficulties of model error and incomplete initials. Nevertheless, there is a quantity of historical observed data of climate evolution in which the information of climate system is contained. In view of the two difficulties, three kinds of inverse problems of numerical prediction are put forward in this paper to improve prediction through retrieving initial value and model parameters to emend numerical model by using the historical data. To solve the three inverse problems, three kinds of numerical solutions are put forward simultaneously. Finally, through using the three kinds of inverse problems in a simple model, some numerical experiments of retrieval prediction are carried out and their results are to be given in the second part of this paper.