对中尺度遥感资料进行四维同化的共轭方法及其数值研究

THE CONJUGATE METHOD AND ITS NUMERICAL RESEARCH FOR FOUR DIMENSIONAL ASSIMILATION OF MESOSCALE REMOTE SENSING DATA

本文探讨了数值天气预报过程中对非定时观测的常规遥感气象资料进行四维同化，形成预报初始场的新途径。把数值预报初始场的形成提为数学上的一类反问题，运用数值模式及其共轭方程对气象资料进行变分同化的共轭方法，使众多观测资料的四维同化与时变的动力模型在初始场的形成过程中统一考察，克服了以往一些方法的局限性。从理论和数值研究角度证明了该方法的优点和可行性，表明该方法有可能作为发展一个新的初值方案的雏形，有潜在的应用价值。

In this paper,we find a new way to solve these problems:the mesoscale remote sensing data assimilation and the formation of the initial field in numerical weather prediction. We put them forward as a kind of mathematical inverse problem,then using the theory of conjugate equations of the model to solve it. In the process of the forming initial field,the multiple-time data assimilation and time-variation dynamic model have been unified. In this way,we overcome the defects in the traditional method. Theory and numerical research indicate the method is useful. It is evident that this method has latent capacity to solve practical problems.