在分布式土壤侵蚀过程模型中 ,承载数据以及进行运算的最小单元 ,即基本地块的选取是非常关键的 ,它直接关系到模型的模拟精度和运算数据量。目前大多数分布式模型都采用平均布设矩形网格的方法 ,这种方法在基本网格的大小选取上存在着盲目性和不统一性。本文以黄土高原丘陵沟壑区为例 ,利用GIS和SPSS分析了黄土高原丘陵沟壑区属性均一的基本地块在面积上的统计规律 ,给出了基本地块选取的合理依据 ,以促进建立更好的分布式模型

Distributed erosion model is one of the effective methods to evaluate soil loss. The land patch is a key componentin establishing adistributed erosion model. Now many researchers prefer rectangle as cellsin the model. However, they had not made an agreement about the cell size. Furthermore, when the rectanglecells are employed, it is somehow difficult to keep a balance between the precision and the load of the model, because both of them are decided from the size of the cell. The larger the cellis, the more laboriousthe work is, but it is not always the case that the model will be more precise. Actually, thecellsize should be determined according tothe uniformity of the land in the region where the model is to be applied and the accuracy of the model needed. However, there is a chance that the user of the model, and even the designer of the model, have no actual image of the region, and then have no idea about what size the cell should be. Similarly, when a cell is chosen, the accuracy is not clearly presented. Zhifanggou catchment is a typical area of the hilly loess region in the Loess Plateau. Supported by GIS and SPSS, the authors got the patches of the catchment which haveuniform characters respectively, and analyzedthe statistic result of the patches' areas. It was indicated that the areas rangedfrom less than 1 m 2to almost 200,000 m 2. On the other hand, nearly 67% of the patches congregate into two groups: either below 500 m 2or between 1,000 m 2to 5,000 m 2. Accordingto the results, one can get anapproximate image of the hilly loess region and then be more reasonable to establish or usea distributed erosion model forthat region.