59 巻 (2006-2007) 3 号 p. 3-12
Hydrological models, which simulate rainfall infiltration and groundwater generation at forested hillslopes, have been proposed for predicting shallow-landslides. However, model performances were not adequately evaluated in comparison with observed rainwater infiltration processes. These models require soil hydraulic parameters, which describe water retention characteristics and hydraulic conductivities, as input data. While these parameters were usually determined in a laboratory by using soil samples, it was suggested that in-situ infiltration processes are not adequately described by using the laboratory-determined hydraulic parameters. In this study we examined three methods to determine hydraulic parameters; the traditional laboratory methods for measuring retention characteristics and saturated hydraulic conductivity with estimating unsaturated hydraulic conductivity by Mualem's model (Method 1), the combination of the laboratory and field methods (Method 2), and the field method which uses vertical profiles of water content and matric pressure observed in-situ during a storm event (Method 3). The accuracy of each method was evaluated by comparing observed and predicted water contents and matric pressures for 10 storm events. Results showed that Method 1 tended to overestimate water storage capacity of soil and under estimate hydraulic conductivity, resulting in a delayed wetting front propagation. On the average, the error between simulated and observed water content changes for Method 1 was 5.84 times greater than that for Method 3. Method 2 produced better prediction results than Method 1, but performed worse than Method 3. In conclusion, Method 3 was the suitable method to determine hydraulic parameters for describing rainwater infiltration processes at forested hillslopes.