数字土壤制图作为获取土壤属性空间分布的一种新方法,目前已成为土壤科学的研究热点。探究土壤有机质在复杂地形区的空间变异特征及其空间分布制图,可为土地利用合理规划与管理、精准农业及生态环境模拟等提供有价值的数据。复杂地形区由于其地形起伏大,地貌类型多样以及采样点获取难度大等,对土壤属性制图带来一定的困难,也使得获取的属性空间分布图很难用于实际的使用。基于湟水流域2016年10—11月份采集的0—20 cm的253个土壤表层数据,采用GIS与地统计学相结合的方法,研究协同克里格(COK)、地理加权回归克里格(GWRK)以及结合土地利用类型的均值、中值修正协同克里格4种方法在湟水流域的预测精度并进行比较,旨在探索复杂地形区土壤有机质制图中地统计学方法的适用性。结果 表明:(1)4种插值方法对SOM的解释能力均较优,均可用于复杂地形区土壤有机质制图。(2)均值、中值修正协同克里格方法与地理加权回归克里格模型准确度分别为0.905,0.923,0.909,中值修正协同克里格精度最高,未考虑土地利用类型的协同克里格插值精度最低,为0.883。(3)有机质的变异系数为90.90%,存在中等空间变异性; 表明其空间变异主要受土壤母质、土壤类型与土地利用管理方式、施肥等共同影响。(4)4种方法预测图表明湟水流域的SOM高值主要分布在大通县西北部、海晏县南部等地,低值主要分布在民和县东南部等地; 整体来看研究区SOM含量有利于农业发展。
Digital soil mapping is a new and effective method to express the spatial distribution of soil, which has become a research hotspot in soil science. To explore the spatial variability characteristics of soil organic matter(SOM)and mapping it in complex topographic areas can provide valuable data for land use planning, and precision agriculture and ecological environment simulation. In complex terrain area, the large terrain relief, the variety of landform types and the difficulty in obtaining soil sampling points bring some difficulties to the mapping of soil properties, and makes it difficult to use the obtained spatial distribution maps for practical use. In this paper, soil samples of 253 at the depth of 0～20 m from the Huangshui River Basin in 2016 were obtained, four kind interpolation methods including the COK, mean and median modified COK and GWRK were studied and compared for SOM contents combining GIS with Geostatistics, and the accuracies of the four methods were evaluated respectively. The results showed that:(1)the four interpolation methods had better explanatory ability for SOM, and could be used for mapping SOM in complex topographic areas;(2)the accuracies of the mean and median modified COK method and the GWRK model were 0.905, 0.923 and 0.909, respectively, in which the median modified COK method had the highest accuracy, while the COK interpolation method without considering land use types had the lowest accuracy of 0.883;(3)the variability coefficient of SOM was 90.90% with moderate spatial variability, which was mainly affected by soil parent materials, soil type, management activities and fertilization;(4)the prediction maps of four methods indicated that the high SOM contents of the study area mainly distributed in the northwest of Datong County and the south of Haiyan County, while the low SOM mainly distributed in the southeast of minhe County. On the whole, the SOM content distribution of the study area was at a medium level, which was conducive to agricultural development.
