中文摘要：

土壤侵蚀产沙是一个极其复杂的物理过程,其物理机理十分复杂,往往难以用数学方式来描述.该文针对区域土壤侵蚀机理不甚清晰、基础资料收集困难等问题,探索性地在小流域土壤侵蚀预报中引入支持向量机理论与方法,并以皇甫川强烈侵蚀区小流域为例构建小流域土壤侵蚀模数预报的支持向量机预报模型.研究表明,该方法的模拟和预测精度都明显高于最小二乘法;支持向量机预报模型依据多年的反映区域侵蚀产沙演进规律的侵蚀产沙监测资料,建立小流域土壤侵蚀模数的支持向量机预报模型,是对流域侵蚀产沙复杂运行机制的综合体现.

英文摘要：

Soil erosion is a very complex physical process and it is difficult to be characterized by mathematic equations. In allusion to the problems of indistinct mechanism of soil erosion and difficultly collecting of basic data, Support Vector Regression （SVR） was introduced to predict the soil erosion of small watershed exploringly. Taking the most intensive soil erosion watershed in Huangfuchuan Catchments as an example, the predicting model with SVR method for soil erosion modulus was formed. It indicates that the SVR method is more excellent than the general method. Based on the soil erosion data reflecting the soil erosion rules, forming the predicting model with SVR method for soil erosion is an integrated embodiment for the complex process of soil erosion.