Soil erosion model predictions using parent material/soil texture-based parameters compared to using site-specific parameters

Forested areas disturbed by access roads produce large amounts of sediment. One method to predict erosion and, hence, manage forest roads is the use of physically based soil erosion models. A perceived advantage of a physically based model is that it can be parameterized at one location and applied at another location with similar soil texture or geological parent material. To test this perception, a two-part study was conducted to compare soil erosion model predictions using parent material/soil texture-based parameters to predictions using site-specific parameters. The first step was to determine site-specific erosion parameters using rainfall simulation. The second step was to compare erosion model results for a typical road network using the two parameter sets. Lake Tahoe was chosen for the study because it had parent materials similar to sites where parameterization had been performed. The coefficient of variation in runoff and sediment mass from the Lake Tahoe rainfall simulations varied from 8% to 36%. These values, although smaller than those reported from natural rainfall studies, are an indication of the inherent variability of erosion measurements. Effective hydraulic conductivity determined by rainfall simulation for a granitic parent material at Lake Tahoe was greater than that for the similar parent material. A volcanic parent material from Lake Tahoe also had higher effective hydraulic conductivity than that for another volcanic parent material. Soil interrill erodibility values for granitic parent material were similar, while values for volcanic parent material were slightly greater for the Lake Tahoe soil. Only the effective hydraulic conductivity of the granitic parent material was statistically different at the 95% confidence level from the parent material/soil texture-based values. Process-based WEPP model predictions of runoff from a typical Lake Tahoe road network were 82% and 73% less for granitic and volcanic soils, respectively, when using the site-specific Lake Tahoe values compared to similar parent material values. Corresponding differences between sediment yields were 85% and 78% less. Most of the decrease was due to fewer snowmelt runoff events. The inherent variability in soil erosion measurements results in a corresponding variability in erosion model parameters derived from those measurements. As a result, model predictions have an inherent accuracy range. In this Lake Tahoe study, the erosion measurement variability resulted in a WEPP model prediction range of approximately ±75% of the mean for both runoff and sediment mass.