黄土丘陵沟壑区县域土壤有机质空间分布特征及预测

分析、预测土壤属性空间变异及其动态是区域土地质量评价和可持续土地利用的一个重要组成部分。在陕西省横山县采集了254个耕层(0~20cm)土样，利用数字地形与遥感影像分析技术，提取了相关地形与遥感指数，分析不同土地利用类型、不同地形条件下土壤有机质空间变异及分布特征，并利用相关因子进行回归预测分析。结果表明，县域土壤有机质平均含量很低，变异性较大。不同土地利用类型土壤有机质差异显著，其中以水稻田有机质含量最高，而林地和灌木林地相对较低。不同土地利用类型土壤有机质含量次序为:水稻田>川地>梯田>坝地>荒草地>坡耕地>林地>灌木林地。不同坡度分析表明，“0~3°”这一坡度等级有机质含量显著高于其它坡度等级;不同坡向有机质含量差异不显著，但不同坡向有机质含量存在一个明显的趋势，阴坡有机质含量整体上要比阳坡高。相关分析表明土壤有机质与高程h呈现负相关关系，与坡向的余弦值COSα正相关，与复合地形指数CTI正相关;土壤有机质和修正后的土壤调节植被指数(MSAVI)以及湿度指数(WI)正相关。利用相关环境变量及遥感指数进行多元线性逐步回归分析，预测结果不甚理想，存在一个平滑效应，对于残差解释相对较低，须进一步研究以更好的解释残差。

Spatial Variability and Prediction of Soil Organic Matter at County Scale on the Loess Plateau

Analysis and forecast on the spatial distribution and dynamics of soil properties is an important element of sustainable land management. Spatial variation of soil organic matter was analyzed according to different land use types and different topography conditions, based on data from 254 points of surface soil (0~20cm) in Hengshan county on the Loess Plateau (NW China). Correlation analyses were carried out between the soil organic matter and the terrain attributes and remote sensing indices. Finally, the land use types and the terrain attributes and remote sensing indices were used to predict soil organic matter spatial distribution by multiple-linear regression analysis. Significant differences in soil organic matter among different land use types were found, the highest values in soil organic matter were measured in soils from paddy field, and lower values in the soils from woodland and shrub land. For soil organic matter, the tendency was: paddy field>irrigated farmland>terrace farmland>check-dam farmland>grassland>slope farmland>woodland>shrub land. In different slope gradients, soil organic matter in ‘0~3°’ gradients was significantly higher than other slope gradient classes. There was little difference in soil organic matter among different slope aspects, but there was a tendency that soil organic matter in northern aspects was higher. Different correlations were found between the soil organic matter and the terrain attributes and remote sensing indices. It was found that there are positive correlations between soil organic matter and the COSα, CTI, MSAVI and WI. There is a strong negative correlation between soil organic matter and elevation. Using environmental variables to predict soil organic matter, the regression model explains 34.6% of the variability of the measured soil organic matter. But the variation is rather large and there is a more smoothing effect on the predicted values for soil organic matter.