Geostatistical data are obtained only at selected sites even though they are potentially available at any location In a continuous surface. Therefore it is necessary to estimate the unknown values at unsampled locations based on observations. In this study we compared the accuracy of 5 spatial interpolation methods: local trend surface, IDW, RBF, ordinary kriging, universal kriging. These interpolation methods were applied to annual rainfall data. As the results of validation tests, universal kriging with gaussian variogram model showed the best accuracy in comparison with other interpolation methods. In the case of kriging, the predicted values were more accurate and within a more narrow range than other methods. In contrast with kriging, local trend surface analysis, IDW and RBF showed the wide range of predicted values and abrupt changes between neighbors.