摘要
针对数值预报中产生误差的两个来源提出了数值预报中存在着两类反问题。并在一维的非线性平流扩散方程上，用共轭方程的解法对提出的两类反问题作出了理想场的数值试验。试验结果表明，这种解反问题的方法非常有效。它利用＂观测资料＂所包含的时间演变的信息确定出了方程的初值或方程中误差订正项的空间分布状况。而且无论对＂观测资料＂的超定还是欠定都能得出较有意义的结果。因而有很大的利用前景。
According to the two sources of errors in NWP, two kinds of inverse problems are put forward, then some numerical expriments of ideal fields are performed on these two inverse problems using a nonlinear advective-diffusion equation and its ad joint equation.The result shows that this method for solving the inverse problems is fairly effective. It can determine the initial spatial state of meteorological elements (or the error correctlve term of model equation) through the time-varition information involved in the observa-nonal data. This method can be used to obtain significant results whether the observation is over-determined or underdetermined.
出处
《气象学报》
CSCD
北大核心
1994年第2期129-137,共9页
Acta Meteorologica Sinica
基金
国家教委高等学校博士点专项科研基金
关键词
天气预报
数值预报
反问题
数值解
Adjoint equation, Inverse problem, Optimum control techniques.
