EXPERIMENTAL EVALUATION OF INFILTRATION MODELS FOR DIFFERENT LAND USE AND SOIL MANAGEMENT SYSTEMS

Soil structural and water transmission properties, as influenced by land use and soil management, affect the coefficients of infiltration predictive models. Ten infiltration models were analyzed to assess these coefficients. The models tested include Green and Ampt (1911), Kostiakov (1932), Horton (1940), Mezencev (1948), Philip (1957), Holtan (1961) —two and three-parameters, Stroosnijder (1976), Swartzendruber (1987), and Kutilek and Krejca (1987). Parameters were evaluated for time-dependence and precision and with regard to the effects of land use/soil management and landscape positions. The field water infiltration data used in these models were based on double ring infiltrometer tests conducted for 3 h at six different land use/soil management treatments at the North Appalachian Experimental Watersheds (NAEW) near Coshocton, Ohio. The treatments were no-till without manure (NTWM), no-till with manure (NTM), no-till corn-soybean rotation (NTCSR), conventional tillage (CT), meadow (M) and forest. Measurements were made at three landscape positions (e.g., shoulder or upper slope (US); back or middle slope, (MS); and foot slope (FS)). The algebraic parameters of the infiltration models and nonlinear least squares regression were fitted using measured infiltration time [I (t)] data. Among process-based infiltration models, the Swartzendruber model performed best and matched the measured I (t) data with lower sum of squares (SS) and higher model efficiency (EF) and Wilmot's index of agreement (W). Overall the three-parameter Horton model gave the best representation of the I (t) relationship with the lowest SS and the highest EF and W for most of the land use treatments, including forest. The treatments had significant influence on the parameters related to initial infiltration rate or sorptivity and final steady state infiltration rate or hydraulic conductivity near saturation, whereas landscape positions had significant influence on the sorptivity parameter only. The fitted parameters (i.e., sorptivity and hydraulic conductivity near saturation) were time dependent and were higher for NTM than for other treatments.