The advanced singular value decomposition(ASVD) method is the singular value decomposition(SVD) method of homogeneity adjustment for the grid and station network data.Based on the relationship between SVD and EOF(empirical orthogonal function),it gives a homogeneity adjustment method of the grid and station network data in SVD method,in such a way that the advanced singular value decomposition(ASVD) method is achieved.Using the ASVD and SVD methods,this paper analyzes the correlation coefficient matrix C. between temperature and precipitation of 160 stations over China in winter during 1951—2010.Results show that both the squarefitting rate and the accumulated squarefitting rate of the preceding four principal modes by ASVD method are obviously higher than those by SVD method.Comparing with SVD method,the relationship between the high value area in C squarefitting figure and the high absolute value area in the preceding two typical singular vector fields by using the ASVD method is more reasonable.It is verified that the homogeneity adjustment is necessary by using the SVD method,and there is more reasonable effect in practical analysis by using the ASVD method.