Research on field-scale spatial variability of soil infiltration characteristics

Abstract: Soil water infiltration characteristics is one of the important factors for estimating the infiltration parameters at field scales. Based on the double-ring infiltration experiments conducted in typical fields of the first terrace (Sandy loam) and the third terrace (Clay loam) in Yangling District, Shaanxi Province, China, the spatial variability of soil infiltration characteristics under different soil textures were studied at field scales. The objectives of this paper are to apply the normalization method for modified Kostiakov function, to analyze the fractal characteristics of normalization factors under single scale and multi scales, and finally, to develop the Pedo-transfer function for estimating the normalization factors. The results show that the normalization method is feasible for modified Kostiakov function, which has high accuracy in predicting the infiltration process of every soil infiltration experiment spot. The correlation coefficients between the predicted values of the cumulative infiltration and the measured values of all the experimental spots in the first terrace and third terrace are 0.9759 and 0.9576, the errors of the root mean square are 2.795cm and 1.266cm, and the mean absolute values of relative error are 10.58% and 9.26%, respectively. Under the single scale, the semi-variance functions of the normalization factors in first terrace and third terrace are shown to be the spherical model, the spatial variance is mainly caused by the structural elements, and the fractal dimensions are 1.796 and 1.840 respectively. The results show that the uniformity of spatial variance of the soil infiltration characteristic in the third terrace is higher than that of the first terrace. During the multi-scale analysis, the spatial variation of the normalization factors in the first terrace are greatly influenced by silt particle, sand particle, and soil bulk density, and the normalization factors in the third terrace are greatly influenced with clay particle, silt particle, sand particle, and soil initial moisture content(0.01 level of significance). The Pedo-transfer function for estimating the normalization factors is accurate and reliable, and the measured and the estimated values of the normalization factors of the verification samples in the first terrace and the third terrace are in good agreement with each other, the mean absolute value of relative error are 13.15% and 9.95%, respectively. The research results of this paper could be used to convert point-based soil infiltration measurements to regional scale.