Application And Study Of Genetic Algorithm Assimilation System In Modifying Model's Terrain

At present, data errors and model errors are the key factors to the improvement of the numerical weather prediction (NWP) results. In four-dimension data assimilation, the model is often assumed to be precise, only correcting the initial data and neglecting the forecast errors caused by the model errors. In this paper, Genetic Algorithm Assimilation System (GAAS) is used in modifying terrain errors which is also the model errors as terrain is an important parameter of model. It provides a new effective method of improving model. This method is able to retrieve the terrain field, which is matched with Initial meteorological element field; optimize model; and improve the forecast results varying with the range and intensity of the meso-scale heavy-rain center. The application of GAAS is feasible to correct model errors.The development, characteristics, applications of genetic algorithm and the theory of modifying terrain are introduced in this paper. Then establishing MM5 GAAS and using it to modify the terrain parameter of model. Two kinds of experiments based on the uncorrected model and the corrected model are compared, the results are as followed: the modified terrain is more consistent with the meteorological observation elements. Once again shows that the impact of terrain on the precipitation is complicated. Model consistented with observation data redounds to forecasting the better moisture, better wind and vertical structure, better thermodynamical structure and so on, thus enhances the forecast of the model to the heavy rainfall. It provides a new way for the extensive application of GA assimilation system.