不同土壤入渗模型参数多元非线性预测模型的精度对比分析

Accuracy Comparison of Different Soil Infiltration Model Parameters in Multivariate Analysis of Nonlinear Predictive Model

为了对不同土壤入渗模型参数多元非线性预测模型的精度进行对比分析,基于区域性黄土田间耕作土壤水分入渗实测资料,分别对Kostiakov、Kostiakov-Lewis以及Philip入渗模型的参数建立了以土壤体积含水率、干密度,黏粒、粉粒和有机质含量等土壤基本理化参数为输入变量的多元非线性传输函数,并进行了精度分析和比较。结果表明:Kostiakov入渗的二参数模型参数的平均误差在9%以下,预测精度明显高于Kostiakov-Lewis三参数入渗模型参数10.78%和Philip入渗模型参数11.72%。因此,推荐形式简单且预测精度高的二参数模型参数多元非线性预测模型为土壤入渗模型参数预测模型。

Based on the regional loess soil water infiltration farming field measured data, dry density, clay, silt and organic matter content of the soil basic physical and chemical parameters are established as input variables multivariate nonlinear transfer function for parameter Kostiakov, Kostiakov-Lewis and Philip infiltration model r with soil volumetric water content, and their accuracy is analyzed and compared. The results show that the average error of Kostiakov infiltration of two-parameter model parameters below 9%, that significantly higher than the prediction accuracy of the three-parameter Kostiakov-Lewis infiltration model parameters, 10.78% and Philip infiltration model parameters, 11.72%. Therefore, it is recommended that the simple and high prediction accura- cy of two-parameter model parameters multivariate nonlinear model be used as a predictive model to predict soil infiltration model parameters.