A Study of the Correlativity Between Weather Stations and the Improvement of the Objective Analysis Scheme of MM4 Model

Using the principal component analysis in multivariate statistical analysis, the correlations between 169 weather stations are discussed in the computation domain of the MM4 forecast model. The result shows that there are 89 key stations which play an important role in the forecasting, and the information of the other stations could be obtained from them by multivariate linear regression. It also shows that the loss of the statistical information in the whole domain is less than 2%. A storm rainfall event on 24—25, July 1992 is selected to test the regression method which is used to supplement the missing data. It is found that the objective analysis field and forecast field are improved