A Study On The Application Of Genetic Algorithms To Variational Data Assimilation For Meteorology

The technique of variational data assimilation has been considered as a effective one in improving the quality of initial fields of numeric forecast model. However, In practice, It needs higher condition by using local optimization methods and depending the computer resources.This paper presents an application of genetic algorithms(GAs) to variational data assimilation for the initial fields. The principle of the GAs is introduced, the method of solving the problem of minimization in variational assimilation by using the GAs is put forword, and the key technique of the construction of fitness function, the coding of chromosome, and the selection of genetic operators and genetic parameters are discussed. As an example, we applied it on variational assimilation of the ideal initial fields, the barotropic primitive equation model was employed to illustration, the problem of the superiority of some kinds of genetic operators, the selection of the genetic parameters and weight coefficient are analyzed though numerical simulation. Results show that the aspects are obviously improved in the requires of functional qualities, computational accuracy, convengence and computational speed of algorithms. By this, the new technique of variational data assimilation maybe has wilder application in practice.