This paper shows a geostatistical performance of estimating the modeling errors for two-dimensional groundwater flow with best model selected by AIC. A variation in the water levels are formulated from the series solution of the FE equation. A hypothetical aquifer into twenty constant-permeability zones is designed to allow assessing the errors and the level variations. Spatial estimation of the error complements and model selection are herein carried out. It can be proven that such estimation and selection are greatly helpful in perfectly modeling at the subsequent time.
We forcused on the prediction of precipitation that is important information for disaster prevent, the relevance between distribution of MSM-GPV and distribution of actual precipitation were studied.As a result, there were the relevance between distribution of MSM-GPV and distribution of actual precipitation.