Modelling runoff and soil erosion in logged forests: Scope and application of some existing models

Predictive erosion models are useful tools for evaluating the impact of land-use practices on soil and water properties, and as often used by environmental protection authorities, for setting guidelines and standards for regulation purposes. This study examines the application of three erosion models of varying complexity and design for predicting runoff and soil erosion from logged forest compartments in south eastern Australia. These are: the Universal Soil Loss Equation (USLE), the Water Erosion Prediction Project (WEPP), and TOPOG, a physically based hydrologic modelling package. Data on rates of soil loss and redistribution collected during a series of large-scale rainfall simulator experiments were used as model input parameters and validation. The models were evaluated in terms of general ease of use, input data requirements and accuracy of process understanding and prediction. Results suggest that in this application the USLE overestimated soil loss, and have the limitation that it does not predict sediment yield or sediment redistribution for specific storm events. When used at the hillslope scale, WEPP and TOPOG have predicted runoff and soil loss reasonably well, particularly on disturbed surfaces such as skid trails. On less disturbed surfaces such as the general harvesting area, both models performed less accurately, generally under-predicting soil loss and sediment yield, notably on sites with low observed values. The complexity and data requirements of WEPP and TOPOG limit their usability as a general-purpose, erosion hazard predicting tool. In terms of process understanding, none of the existing models accurately depict the nature and extent of sediment redistribution quantified in the rainfall simulator experiments. In order to advance the application and accuracy of modelling tools in forestry environments, this redistribution process should be considered integral to the refinement and redevelopment of future models.