论文部分内容阅读
作为直接试验的一种替代方法,利用土壤基本物理性质通过土壤转换函数预测饱和导水率简便易行,成本低廉,并且预测精度能满足实际研究的需要。本研究利用目前得到较多应用的9种基于多元回归分析建立的转换函数来构建、校正预测土壤饱和导水率的经验公式,并与人工神经网络方法相比较。结果表明,人工神经网络总体预测效果要优于基于多元回归分析建立的转换函数,并且Cosby(1984)在输入参数较少的基础上预测饱和导水率精度最高。本文以Cosby(1984)预测常熟水稻土壤饱和导水率,进一步利用GIS的空间描述能力与函数的定量分析能力,得到区域尺度饱和导水率的分布状况,为该地区区域尺度数值模拟的运行提供基础参数支持。
As an alternative method of direct test, it is simple and feasible to predict the saturated hydraulic conductivity by the soil transfer function using the basic physical properties of soil. The prediction accuracy can meet the needs of practical research. In this study, we constructed and corrected nine empirical formulas for predicting saturated hydraulic conductivity of soils by using the nine conversion functions established by multiple regression analysis, which are widely used nowadays. Compared with the artificial neural network method. The results show that the overall prediction effect of artificial neural network is better than that based on multivariate regression analysis, and Cosby (1984) predicts the highest accuracy of saturated hydraulic conductivity with less input parameters. Based on Cosby (1984) predicting the saturated hydraulic conductivity of paddy soil in Changshu, the distribution of saturated hydraulic conductivity at the regional scale is further utilized by using the spatial description ability of GIS and the quantitative analysis function to provide the numerical simulation of regional scale operation in this region Basic parameters support.