Partition of soil moisture profiles based on sequential clustering method | 基于有序聚类法的土壤水分剖面划分

Chinese. 研究有序聚类法在土壤水分剖面划分中的应用。在黄土高原丘陵区陕西省安塞县，以乔木林地、灌木林地、果园用地和天然草地等4种土地利用类型为研究对象，测定4种用地类型0～500 cm土层（每20 cm为1层，共25层）的土壤水分含量，然后采用有序聚类法对4种土地利用类型土壤水分剖面进行分层。从专业角度无法确定最优分类数K时，可以利用碎石图来确定，对全部样本进行分类并作碎石图可知，当K=4时，有序分割达到最优；对土地利用类型而言，土壤水分剖面总体可划分为水分弱利用层、根系集中层、强烈耗水层、水分调节层等4层，中间2层又可简单地归结为土壤水分利用层。基于有序聚类法划分的土壤水分剖面分层结果符合黄土高原丘陵沟壑区的土壤水分垂直分布实际，表明了该分层方法的合理性。

English. The research was conducted to study the application of sequential clustering method in the vertical profiles of soil moisture classification. In the hilly Loess Plateau of Ansai Shaanxi, the research object was 4 land use types such as tree land, shrub land, orchard and grassland land use types. The soil water profile of 4 land use types was classified by the orderly clustering method according to the water content of 0-500 cm (every 20 cm for 1 layer, altogether 25 layer). When K values cannot be predicted from the angle of profession, it can be fixed by the stone graph. As a result, through all the samples and stone map, it is known that when K=4, there is an optimal ordered partition. According to various land use types, the overall vertical soil moisture profile in this area can be divided into 4 types: weak moisture utilizing layer, root concentration layer, strong water consumption layer, and moisture regulating layer. Meanwhile, the middle two layers can be merged into a layer of moisture utilizing. Based on the orderly clustering method the result of soil moisture profile distribution fits for Loess Plateau area. So the hierarchical method is reasonable.