基于多辅助变量的区域土壤可蚀性因子制图及其不确定性分解研究

Accurate prediction of regional soil erodibility distribution and assessment of temporal and spatial uncertainties are important to controlling error and improving accuracy of soil loss calculation. This study tries 1) to map regional soil erodibility by stochastic co-simulation based on multiple auxiliary variables, such as soil types, remote sensing images, and digital elevation models, 2) to assess total temporal and spatial uncertainties of regional soil erodibility distribution, and 3) to partition the errors into different sources, such as sampling scales, auxiliary variables, and models, by the variance, covariance and cross-covariance decomposition. The accurately mapping of regional soil erodibility and uncertainty budget have theoretical significances on soil loss prediction and ecological model, and also have practical significances on soil erosion survey, regionalization and plan.

准确建立区域土壤可蚀性因子的空间分布图，并分析其存在的误差或不确定性，对控制水土流失量预测误差、提高精度具有重要作用。本研究采用基于土壤图、遥感影像和地形图等多辅助变量的协同随机模拟方法制作区域土壤可蚀性因子空间分布图，量化制图的空间不确定性，并采用方差、协方差、交互协方差分解的方法对空间尺度、辅助变量、模型等导致的空间不确定性进行分解，同时还探讨了历史数据的时相误差。实现了基于历史数据的区域土壤可蚀性因子准确制图和空间不确定性定量及分解，对水土流失预测、生态环境建模、水土保持普查等，具有一定的理论和现实意义。

准确建立区域土壤可蚀性因子的空间分布图，并分析其存在的误差或不确定性，对控制水土流失量预测误差、提高精度具有重要作用。本研究利用分析得到降雨侵蚀力因子、坡度因子、坡长因子、生物措施因子、工程措施因子、耕作措施因子以及土地利用类型因子，采用多辅助变量的协同随机模拟方法制作区域土壤可蚀性因子空间分布图。