| China is one of the countries where are seriously drought in the world. The agricultural irrigation water is more than 60% of the total water consumption,while the effective utilization coefficient of irrigation water is only 0.5 that is far lower than the world advanced level.The low utilization rate of agricultural irrigation water is mainly influenced by many factors, such as the natural attributes of the land, the structure and the management of the land. But the most fundamental reason is not according to the reasonable irrigation technical parameters for irrigation. Soil infiltration parameters determine the rate and distribution of irrigation water to soil water, and then affect the irrigation quality and effect. Therefore, the soil infiltration parameters are the basis for determining the reasonable irrigation technical parameters under various irrigation technologies. The research of soil infiltration parameters is the key to improve irrigation water use efficiency and improve the surface irrigation technology.Taking into account the field testing method to retrieve the soil infiltration parameters are time-consuming, laborious, complex issues, and so on. Themethod of soil transfer function was used in this paper. It was based on the large sample that was performed through the indoor experiment and field soil permeability test in Shanxi eight cities and 20 counties by the research team in last 20 years. The 1Stopt software and MATLAB software were used to respectively establish the multivariate nonlinear transfer functions between the parameters of Kostiakov three parameter cumulative infiltration model and soil physicochemical parameters about non frozen non saline alkali soil, frozen soil and saline alkali soil. The multivariate nonlinear transfer functions of non frozen non saline alkali soil were established between the parameters of Kostiakov two parameter cumulative infiltration model and soil physicochemical parameters.After the significant tests and overall function test of the transfer function, and the relative error between the measured value and the calculated value, the multivariate nonlinear prediction of the Kostiakov cumulative infiltration quantity models in different stage was realized. The comparison was made between the multivariate nonlinear prediction and the linear prediction and BP prediction. Research results showed that;(1) It was feasible to predict the soil infiltration parameters by using the method of multivariate nonlinear transfer function. The nonlinear prediction of soil infiltration parameters of non frozen non saline alkali soil, frozen soil and saline alkali soil was feasible.The variables and overall models of each model were all significant, and the relative errors between the predicted values and measured values were in acceptable range which was less than 10%.(2) The nonlinear prediction models of soil infiltration parameters in different stages had different input variables or influencing factors. For non frozen saline alkali soil, the main influencing factors of infiltration parameters included soil moisture rate, structure, texture and organic matter; for frozen soil,the main influencing factors of infiltration parameters included soil moisture rate, structure, texture, organic matter and soil temperature; for saline alkali soil,the main influencing factors of infiltration parameters included soil moisture rate, structure, texture, organic matter and salinity.(3) The comparison results of non frozen saline alkali soil between Kostiakov two cumulative infiltration quantity models and Kostiakov three cumulative infiltration quantity models found that, the prediction accuracy of the two models could meet the need of earthquake prediction, but the accuracy of the two parameter infiltration model was higher than that of the three parameter infiltration model. Therefore, the Kostiakov two parameter cumulative infiltration model was suitable for short duration prediction of soil infiltration parameters, and Kostiakov three parameter cumulative infiltration model was suitable for long duration prediction of soil infiltration parameters.(4) The comparison and analysis between multivariate nonlinear prediction model and linear prediction model and BP model found that the form and algorithm of multivariate nonlinear model were simpler than the BP model,and the prediction accuracy of multivariate nonlinear model was higher and closer to the reality than the linear model. Therefore, the multiple nonlinearmodel was proposed to predict the soil infiltration parameters.The main innovation of this paper:(1) Based on the combination of 1Stopt software and MATLAB software,the process of realizing multivariate nonlinear fitting was explored, and a new method of multivariate nonlinear fitting was realized.(2) For frozen soil and saline alkali soil, the multivariate nonlinear predictions of Kostiakov three parameter cumulative infiltration model’s parameters with different soil parameters as input variables were realized, and their errors were acceptable.(3) For the first time, the structural deformation of loose soil during irrigation was considered and applied to the prediction of different parameters,that leaded to the prediction results of the infiltration parameters more close to the actual infiltration process.In view of the difficulty in obtaining the soil infiltration parameters,multivariate nonlinear prediction models of non frozen non saline alkali soil,frozen soil and saline alkali soil were established by using soil transfer function method. It realized the prediction of soil moisture infiltration parameters, and provided a new method for obtaining the infiltration parameters at different stages. However, due to the complexity of the problem and the diversity of variables, the level of the physical and chemical parameters of the research progress there is potential to be dug and to be further in-depth study. The purpose of prediction of soil infiltration parameters is to make the application ofinfiltration parameter prediction model more widely, and it has important practical significance to improve the surface irrigation technology and irrigation water use efficiency, and ease the contradiction between supply and demand of water resources. |
