摘要
为了准确推求包气带土壤的持水性能和水分运动参数,以黄土高原区田间耕作层土壤水分特征曲线的试验资料为数据样本,建立了以易于通过试验手段获取的土壤基本理化参数为输入变量,以土壤水分特征曲线Gardner模型参数为输出变量的BP预测模型。所建立的预测模型对两参数预测值的平均相对误差在6%以下,在可接受的范围。研究结果表明:选取土壤体积质量、有机质量、黏粒量、粉粒量以及无机盐量等基本理化参数作为预报模型的输入因子是合理的,所建立的土壤水分特征曲线Gardner模型参数BP预报模型可靠。研究结果可为黄土高原区包气带土壤持水性能和水分运动参数的准确获取提供借鉴。
In order to acquire water holding capacity and water movement parameters of unsaturated soil, based on the data of soil moisture characteristics curve of the loess, the prediction models are established by using BP Neural Network, in which physical and chemical parameters are chose as input variables and the Gardner model parameters are chose as output factors. The average relative errors of the two parameters are both below 6%. The prediction accuracy of the two models can both meet the application requirements. The results show that it is feasible to choose soil texture, bulk density, organic matter and inorganic salt contents of soil as input variables to predict the SWCC Gardner model parameters and it is reasonable to use BP Neural Network to establish predictive models for SWCC Gardner model parameters. The research can provide reference for obtaining the water retention and water movements of vadose soil in the Loess Plateau region.
作者
赵红光
樊贵盛
于浕
舒凯民
ZHAO Hong-guang FAN Gui-sheng YU Jin SHU Kai-min(Water Conservancy Science and Engineering College, Taiyuan University of Technology, Taiyuan 030024, Chin)
出处
《节水灌溉》
北大核心
2017年第10期22-25,30,共5页
Water Saving Irrigation
基金
国家自然科学基金项目(40671081)
关键词
土壤水分特征曲线
Gardner模型
土壤理化参数
BP模型
精度检验
soil moisture characteristics curve
Gardner model
BP neural network
physical and chemical parameters of soil
check ofprediction accuracy
