| Soil water infiltration is a non-equilibrium process in many environmental conditions,it determines the surface runoff,velocity and distribution of irrigation water and rainfall into soil water,and then affects the quality and effect of irrigation(irrigation uniformity,irrigation efficiency and water storage effect),and is the main basis for determining reasonable irrigation irrigation technology parameters in different conditions.Therefore,the study of soil water infiltration parameters is a key technology to improve the irrigation technology,and to improve the utilization rate of irrigation water.Based on the regional scale of the large-scale cultivation of soil field infiltration test,the paper attempts to establish a simple and feasible method for determining the parameters of soil infiltration parameters by using the soil parameters transfer theory,and the quantitative relationship between soil infiltration model parameters and the parameters of soil parameters.The multilinear,nonlinear,BP neural network model of soil transfer function between the soil infiltration parameters and the physical and chemical properties of soil were established on the regional scale,realized the purpose of predicting soil infiltration parameters.he research results can provide strong technical support for the implementation of the optimization of irrigation technology parameters,and to a certain extent,enrich the development of the theory of soil transfer function.Research results show that:(1)Soil water infiltration is affected by many factors,for example soil structure,texture,moisture content,organic matter content,salt content and so on.Physicochemical parameters of soil texture,structure,water content,organic matter content have Quantitative relationship of linear and nonlinear with soil infiltration parameters.Soil dry bulk density,soil initial moisture content clay,content of soil texture and soil salinity are inversely proportional to the infiltration capacity;soil organic matter content is proportional to the infiltration capacity.(2)It is feasible to predict the soil infiltration parameters based on the conventional soil physical and chemical parameters through using soil transfer function.The significant test of Kostiakov two parameter infiltration model,the Kostiakov-Lewis three parameter infiltration model and the Philip infiltration model are obvious.The significant test value of each independent variable and the regression equation were all larger than the corresponding F0.95,which showed that the regression coefficient and the regression equation of the transfer function are obvious.With the increasing of the number of variables,themultiple correlation coefficient model is close to the R=1,the correlation is getting better and better,ultimately the total correlation coefficient R of the calculated value and the measured value for all multivariate linear regression model function is 0.9 ~ 0.95,R2 are greater than 0.81.It shows that the multivariate linear transfer function obtained from the experimental data is better,and it is feasible to predict the soil infiltration capacity and infiltration parameters with the soil physical and chemical characteristic parameters.(3)The reasonable input variables of Kostiakov two parameter,Kostiakov-Lewis three parameter and Philip two parameter infiltration model include soil sand content,clay content,uneven coefficient,curvature coefficient,volumetric water content,weight moisture content,dry bulk density,organic matter content and so on.(4)Kostiakov two parameter prediction model have lower relative error,higher accuracy,better parameter correlation and high fitting degree through linear,nonlinear and BP model.(5)The relative error of Kostiakov-Lewis three parameter prediction model are in reasonable range.Kostiakov-Lewis three parameter prediction model have low relative error,significant correlation,good fitting degree,and can realize the prediction of soil infiltration parameters through linear,nonlinear and BP model.(6)The relative error of Philip infiltration parameters prediction model is in a reasonable range.The prediction parameters are feasible,but the relative errorvalue is too large,and the accuracy is low through linear,nonlinear and BP model.(7)In Kostiakov two parameter prediction model,Kostiakov-Lewis three parameter prediction model and Philip infiltration model,the best prediction result of Kostiakov-Lewis model,he forecast effect of Kostiakov-Lewis model is the best,which can predict the process of soil water infiltration,especially for long duration water infiltration,which has a higher and more stable prediction accuracy than other models.The accuracy of the BP model is higher than that of the multivariate nonlinear model,but the stability of the model is not equal to the multivariate nonlinear model. |
PDF Full Text Request