优化模式物理参数的扩展四维变分同化方法

数值模拟的一个重要误差来源是模式物理参数,为提高模拟准确率,如何改进模式物理参数是亟需解决的问题.本文对经典四维变分同化技术进行了改进,提出了一种新的利用观测资料来同时优化模式初始场和物理参数的扩展四维变分同化方法,并以Ekman边界层模式和Lorenz模式为例进行了数值试验.结果表明,利用本文提出的新方法,通过对观测资料的变分同化,可以在实现对模式初始场进行优化的同时,纠正了模式物理参数中的误差,从而有效提高了模式的模拟准确率.该方法对于改进数值模式物理参数有着重要的促进意义.

Expanded four-dimensional variatiaonal data assimilation method to optimize model physical parameters

A critical error in numerical simulation stems from the physical parameters in the model. To better assess the accuracy of the numerical stimulation, a mthod of impoving physical parameters is urgently desired. By modifying the four-dimensional variatiaonal data assimilation 4DVAR technique, in this paper a new method is proposed based on the use of observational data to optimize initial field and subsequent physical model. Ekman boundary layer model and Lorenz model are taken for example to conduct numerical experiment. The results show that through the variations in observational data, physical parameters and initial field are improved, thus effectively enhancing the accuracy of the model. This method improves the numerical model and physical parameters.
Keywords： four-dimensional variatiaonal data assimilation observational data model parameters optimize