耕作后土壤水分入渗参数BP预报模型

The Infiltration Parameters of BP Prediction Model After Tillage

根据黄土高原区耕作后土壤水分入渗资料所建立的基于Kostiakov三参数经验模型公式的BP神经网络预报模型,对土壤水分入渗参数进行预测。预测结果表明:利用土壤基本理化参数耕作层(0~20 cm)土壤含水率及其密度、砂粒含量、粉粒含量、有机质含量预报入渗模型参数具有可行性。入渗模型参数入渗指数α、入渗系数k的平均相对误差约为1%;稳渗率f_0的平均相对误差约为8%,在可接受范围内;90 min累计入渗量I_(90)的平均相对误差小于3%。模型整体预测精度较高,实现了利用土壤基本理化参数预报土壤水分入渗参数,可为黄土高原区耕作后土壤的灌水技术参数的确定提供理论与技术支撑。

According to the BP model which was proposed with neural network that based on Kostiakov three parameters empirical formula and built on account of soil-water's infiltration data of the loess plateau come with farming,it predicted the infiltration data of soil moisture.The predicted result shows that plow layer( 0 ~ 20 cm) which is using basic soil physical and chemical parameters,its magnetism of soil water content,bulk density,sand content,silt content,as well as the infiltration model parameters of the organic content forecast has certain feasibility. Infiltration model parameters including the infiltration index a and infiltration coefficient k,the average relative error of them is around 1%; the steady infiltration rate's f_0 average relative error is about 8%,that is well with the acceptable limits; And the accumulative infiltration amount I_(90) in 90 min and its average relative error is less than 3%. The overall prediction accuracy is higher and achieves to predict the infiltration parameters of soil moisture by using basic soil physical and chemical parameters. Important theory and technical support are provided for the research results which confirms irrigation technique parameters after farming in loess plateau.