基于多元回归的黄土沟壑区小流域土壤水分空间模拟

Spatial Simulation of Soil Moisture in Small Watershed of a Loessial Gully Region based on the Multiple Regression Analysis

以延河流域羊圈沟为研究对象,基于土壤水分空间自相关理论,在土壤水分环境影响因子多元回归基础上,建立了土壤水分空间化模型,并探讨了土壤水分空间分布特征。结果表明:研究区内土壤水分空间分布具有正相关性,且相关性显著,在空间上表现为聚集状态;表层土壤水分与植被覆盖度、高程、坡度、地表粗糙度、地形起伏度和地表切割程度等因素密切相关,且与地表切割程度呈正相关关系,而与植被覆盖度、高程、坡度、地表粗糙度以及地形起伏度因子呈负相关;与普通克里格(OK)、反距离加权(IDW)、径向基函数(RBF)和全局多项式(GPI)的土壤水分空间内插方法相比,多元回归+普通克里格方法(MROK)的拟合程度最高,是表层土壤水分空间化较好模型;土壤水分空间分布格局明显,主要表现为西部和东北部土壤水分含量低,东南部土壤水分含量高。该研究揭示了土壤水分环境因子相关性及空间分异规律,对区域生态文明建设具有一定的促进意义。

Based on the theory of soil moisture spatial autocorrelation, this research analyzed the multiple regression between soil moisture and environmental factors, established the spatial model of soil moisture and discussed its spatial distribution characteristics in Yangjuangou watershed. The results showed that the spatial distribution of soil moisture in the study area had a significant positive correlation and presented aggregated spatial distribution. The surface soil moisture had a positive correlation with land surface cutting degree, while it had a negative correlation with vegetation coverage, elevation, slope, land surface roughness and the relief degree of land surface. Compared to the traditional methods of spatial interpolation, such as Ordinary Kriging （OK）, Inverse Distance Weighting （IDW）, Radial Basic Functions （RBF） and Global Polynomial Interpolation （GPI）, Multiple Regression Ordinary Kriging （MROK） was the best method which took into account the environment factors and reflected the spatial distribution characteristics of soil moisture effectively. There is a geographical pattern in soil moisture distribution and the value of soil moisture in north was relative lower than that in the south of Yangjuangou. The study revealed the correlation pattern between soil water and environment factors and their spatial differentiation.