Study On Detection And Spatial Variability Of Soil Hydraulic Parameter

The soil-water movement is one of the most complex important part of the terrestrial water cycle, and also is an important part of the material recycling in soil, at the same time soil-water is the carrier of the heat and solute transport in soil. Studying the laws and determining parameters of soil-water movement are the basis of calculating the influx water through soil profile, designing the engineering of irrigation and drainage, and also getting and mastering the soil-water hydraulic parameters have very important significance in understanding and controlling soil-water and land surfance water pollution. The calculating models of soil-water hydraulic parameters are the basis of studying and describing soil-water transport, and the simplification of models, measuring paremeters fastly and accurately become very critical. Therefore it will have great practical significance that improves all kinds of current transport models of soil-water, perfects medols of measuring parameters and puts forwars a kind of calculating models of soil-water hydraulic that measuring paremeters not only fastly but also accurately.Based on the analysis of theory and infiltration experiments in the field, such as different initial head Philip–Dunne infiltration, different diameters single-ring infiltration, and soil-air permeability. The paper studied brown forest soil infiltration characteristic, amendment and improvement of infiltration models, and also it forecast the soil-water hydraulic parameters used Philip formula, the theory of statistical analysis, regression model, Back-Propagation network, Radial basis function, Projection pursuit regression. In the studied area, it analyzes spatial variability of soil-water infiltration parameters with the method of traditional statistics,Geostatistics. The results are as follows:1,In the method of detecting the saturated hydraulic conductivity k_susing single-ring infiltration, forecasting the steady-state flow rate Q_sbased on Philip formula and further calculating k_scan get the infiltration parameters in the condition of longer time than the practical infiltration time. At the same time, the Q_sforecasted and k_scalculated that based on Philip formula using part of the experimental data, in the non-steady-state, are also smaller than actual measured values, and the standard deviation, coefficient variation of k_sare also smaller, they not only can save time and water but also can reflect the water-soil infiltration characteristics theoretically. 2,In the method of detecting k_susing single-ring infiltration, the getting k_s, measured using many steady-state heads, has better stability and weaker variability than measured in the condition of two steady-state heads. using many steady-state flow can lower measured error caused by soil non-uniformity and mutation in the field , In the two steady-state heads flow selecting more distance between two flow heads can increase measured accuracy.3,In the method of detecting k_susing single-ring infiltration, the getting steady deep water, in the method of water supplement rate constantly, has the weaker deviation than that of the steady water supplement in the method of constant deep water. Therefore, to get the saturated hydraulic conductivity k_s in the method of water supplement rate constantly can reduce the number of sampling points.4,To forecast the saturated hydraulic conductivity using the soil-air permeability rate can be achieved a certain degree of accuracy, the method of detecting k_s not only can save time and water but also can reduce the disturbance to soil.5,In the method of detecting k_sandΨusing Philip-Dunne infiltration, the positive linear relationship is found between the saturated hydraulic conductivity k_sand the cumulative infiltration time, especially with the infiltration time of the middle water-level( 1 /tmed), and also there is a positive power function relationship between wetting front suctionΨand the ratio of t_maxand t_med( t_max/ t_med). Therefore, it is not necessary to make complete experiments'datas in every experiments'point to get the soil-wate infiltration parameter in a large field, the model established based on parts of experiments'datas can be used to forecast the other points'saturated hydraulic conductivity. This method of detecting k_scan save time and water in infiltration experiments.6,Philip formula can be used to analysis the dynamic process of Philip-Dunne three- -dimensional infiltration, although the parameter'significance of the Philip formula has been changed when it is used to analysis dynamic process of Philip-Dunne three-dimensional infiltration, there is relationship between the parameter A and the saturated hydraulic conductivity k_s, to calculate k_sbased on parameter A achieved in analysis dynamic process of Philip-Dunne infiltration can simplify the computation in a certain degree.7,It has been found that there are close relationship between the skewness, kurtosis of infiltration time and the saturated hydraulic conductivity k_s, wetting front suctionΨ, at the same time, the saturated hydraulic conductivity and wetting front suction that were forecasted based on the series of cumulative infiltration time can reach a high prediction accuracy in the methods of regression model, Back-Propagation network, Radial basis function and Projection pursuit regression. And in contradistinction, the k_s andΨforecasted in the method of Projection pursuit regression can reach the highest prediction accuracy.8,In the method of detecting k_s andΨusing Philip-Dunne infiltration, the infiltration parameter k_s andΨhad moderate spatial variability under three different initial head conditions, and the spatial variability, as well as the accuracy, ofΨis higher than k_s under every same experiments conditions. As the initial head increase the spatial variability of k_s andΨare weaker while the accuracy of them are higher. Therefore, the infiltration parameter k_s andΨcan reach the highest prediction accuracy 50cm initial head.