地層中常由於物理過程、化學作用與生物活動而形成介質之異質性，此異質性常產生流場中水流與污染物偏好的路徑，如介質之大孔隙、岩體之破裂而與高傳導係數連續區，其爲地下水流與污染物快速移動的區域。本研究以一假設案例進行地質統計模式之建構，探討不同採樣方法對地質統計模式建構之影響。由研究結果顯示，階層式隨機採樣較均勻採樣法能獲得較多現地統計資訊，使用逐步高斯模擬法與模擬退火法進行圖象模擬，探討何者較能掌握真實場異質性之空間分佈特徵顯示，模擬退火隨機場能掌握較多統計資訊，故較能展現現地水文地質異質性特徵。 Heterogeneity is often formed by physics process, chemical action and biological activity under ground surface. Soil macropores, such as cracks and channels that formed by roots and soil animals, are recognized as important pathways for water and pollutant and leads to the heterogeneity. In this research, two geostatistical models arc constructed by hypothetical image include uniformly sampling method and stratified random sampling method. The result shows that the stratified random sampling method is better than the uniformly sampling method in that it offers more geostatistical information. Then sequential Gaussian simulation method und simulated annealing method are used to reconstruct the images. The result shows that simulated annealing is superior in better reconstructing the image because of its ability of using more statistical information.