Quality Control For Surface Observation Data And Data Assimilation

As a means of forecast, Numerical Weather Prediction (NWP) plays an increasingly important role in the weather forecast. But NWP still has some problems, such as numerical model initialization and physical parameterization problems. Data assimilation is to make full use of the available information to determine as accurately as possible the model initial field. It will help to improve the predicted and simulated results choosing proper parameterization scheme. With shortening of the data assimination updating cycle, Surface Observation Data Assimilation becomes more and more important in data assimilation system. As the quality of data affects it's assimilation effect directly, the quality control (QC) methods for surface observation data are very important.In this thesis, a new quality control technology is established on the basis of traditional QC methods. The new QC methods, which are put forward in different regions using the surface data of the past 25 years, are applied to the surface data of 2005 in South China, middle and lower reaches Yangtze River and North China. And then the heavy rain occurred in early June of 2006 in the northern part of Fujian Province is simulated using mesoscale numerical model WRF, to compares and analyses the simulated results before and after data assimilation in different boundary layer parameterization schemes. Results show that: (1) the new QC scheme for surface meteorological data can mark out the suspectable data effectively and provide promising quality assurance for surface observation data. (2) It is good to check surface data in different QC regulations because the data in different climate regions have different features. (3) Tests in different parameterization schemes have different simulated results of the maximum intensity and distribution of precipitation as well as the physical fields. After data assimilation, tests using boundary parameterization schemes enhance the total precipitation notablely and adjust the location imperceptibly, while tests without any boundary parameterization schemes remarkablely change the total precipitation, and adjust the location southward. (4) Better results can be obtained using data assimilation and choosing proper boundary layer parameterization scheme in numerical model.