Abstract: Meteorological data sets were widely used in ecological system modeling and zonation of agricultural resources.However,without geographical knowledge being considered,most present interpolation models were purely mathematical ways,so the result were usually very poor,especially in the district where just has less or no meteorological observation stations.Based on the relationships between mean temperature and geographical factors,a geographical knowledge aided spatial interpolation model was introduced to improve the spatial interpolation result.Monthly mean temperature data from 67 meteorological stations in Heilongjiang Province from 1957 to 2004 were interpolated by using Inverse Distance Weighting(IDW),Ordinal Kriging(OK),Spline and Knowledge Based Spatial Interpolation(KBSI)3 Cross-validation was applied to evaluate the four interpolation methods,from which the Mean Absolute Error(MAE), Mean Relative Error(MRE) and Root Mean Squared Error(RMSE) were calculated.The resulted showed that KBSI had the minimal MAE and RMSE.The spatial distributions of monthly mean temperature were also reasonable.

Key words: DEM, DEM, Spatial interpolation, Mean temperature, Heilongjiang Province