Qin Wenjing, Fan Guisheng, and Hu Jingjuan. Estimation of Soil infiltration Process From soil Physical and Chemical Properties. Dynamic Systems and Applications 29 (2020) No. 7, 2408 – 2425

https://doi.org/10.46719/dsa20202974

ABSTRACT. In order to know the soil water infiltration process in Loess Plateau of China, a data set of Kostiakov-Lewis model parameters was established based on the large-scale infiltration test. The nonlinear prediction model was constructed with soil texture, bulk density, moisture content and organic matter content as independent variables and infiltration model parameters k, α and f₀ as dependent variables. For both training and verification samples, the absolute errors were all between -10% and 15%, the relative errors were all between -10% and10%. The correlation coefficients R for parameters k, α and f₀were 0.77, 0.86 and 0.84 respectively. Both absolute error and relative error were within acceptable range. The results showed that it is feasible to predict the infiltration parameters k, α and f0 in Kostiakov-Lewis model by using the nonlinear prediction model with soil texture, bulk density, water content and organic matter as input variables. The nopnlinear regression method is  a new stable and easy to use method to obtain the soil infiltration process with high prediction accuracy. Provides a foundation for the implementation of water-saving irrigation in the Loess Plateau.

Keywords: The Loess, Kostiakov-Lewis model parameters, soil physical and chemical properties, the nonlinear prediction model.