基于环境相关法和地统计学的土壤属性空间分布预测

Prediction of the spatial distribution of soil properties based on environmental correlation and geostatistics

土壤属性是土壤质量的重要决定因素,并强烈影响土地利用和生态过程。正确理解并充分考虑土壤空间变异,对于在景观尺度上建立生态、环境过程模型是必不可少的。在黄土高原横山县采集了254个样点,应用数字地形与遥感影像分析技术,获取相关地形因子与遥感指数,分析土壤属性(土壤容重、有机质和全磷)与环境因子相互关系,并利用环境变量进行空间预测。结果表明,土壤容重、有机质与地形因子和遥感指数之间存在较好相关性,而全磷与地形因子相关性不大;多元线性逐步回归模型对于土壤容重和有机质拟合较好,而对于全磷,预测结果较差;回归-克里格预测有效地减小了残差,消除了平滑效应,与实测值较为接近。

As the most important determinants of soil quality,soil properties significantly influenced land use and ecological processes.On the landscape scale,a comprehensive understanding and consideration of soil spatial variation was essential for establishing an ecological and environmental process modeling.Spatial variation of soil properties（include bulk density（BD）,soil organic matter（SOM）and total phosphorus（TP））was analyzed and predicted according to environmental indicators based on data from 254 points of surface soil（0-20 cm）by digital terrainin and remote sensing image analysis technique in Hengshan county on the Loess Plateau.The relationship between soil properties,terrain attributes and remote sensing indices was analyzed.Finally,environment variables were used to predict spatial distribution of soil properties by multiple-linear regression analysis and geo-statistical.The results showed that BD and SOM were positively correlated with terrain attributes and remote sensing indices,but TP has little significant correlation with remote sensing indices.The multiple-linear stepwise regression model was relatively precise for the BD and SOM,but for TP,the predicted result was poor.The regression-Kriging method can effectively reduce residuals in prediction by eliminating smoothing effect,and its predicted values are quite close to the measured.