Based on the average temperature from 1961 to 2005, longitude and latitude, height above sea level of 109 meteorological stations in the Yellow River Basin (YRB), and China DEM, accumulated temperature data were rasterized using different methods, viz, Cokriging, accumulated temperature vertically descending and “model based computation result plus spatialized residues”. The results show that the magnitudes of the correlation coefficients can be ranked in the following order, “model based computation result plus spatialized residues”＞ accumulated temperature vertically descending＞Cokriging. But also, the magnitudes of the T\|test Sig. can be ranked in the following order, “model based computation result plus spatialized residues”＜“model based computation result plus spatialized residues” ＜0.05＜Cokriging. Therefore, “model based computation result plus spatialized residues” has the highest accuracy.