Search

This paper tests the candidacy of one qualitative response model and two quantitative models for a nationwide water erosion hazard assessment in Ethiopia. After a descriptive comparison of model characteristics the study conducts a statistical comparison to evaluate the explanatory power of the models, using an Ethiopian soil erosion data set as reference. The study, therefore, introduces a generic transformation procedure, whereby qualitative models reproduce quantitative results, while the outcomes of quantitative models are mapped on an ordered (qualitative) classification. The evaluation yields the following results. Application of the USLE model in Ethiopia is restricted by data paucity, while it ranks lowest in the statistical evaluation. However, it provides reliable results in areas where water erosion incidence is low. The Expert model, based on easily available data and expert judgements, covers a wide variability of the explanatory variables, which makes it suitable for a nationwide assessment. It is the second-best model in the statistical evaluation. Yet, its qualitative output complicates the assessment of the dynamic changes in soil productivity characteristics, while the postulated additive form of the logit model is not appropriate to assess erosion hazard. The quantitative AccDat model has the highest predictive power and is based on easily available data, but has a frail empirical basis and its application at a nationwide scale requires a careful interpretation. The varying performances in the different areas of the data domain justify the selection of a combination of models for a nationwide erosion assessment, rather than a single 'best' model. © 2011 Elsevier Ltd.

This paper introduces a simple two-layer soil water balance model developed to Bridge Event And Continuous Hydrological (BEACH) modelling. BEACH is a spatially distributed daily basis hydrological model formulated to predict the initial condition of soil moisture for event-based soil erosion and rainfall-runoff models. The latter models usually require the spatially distributed values of antecedent soil moisture content and other input parameters at the onset of an event. BEACH uses daily meteorological records, soil physical properties, basic crop characteristics and topographical data. The basic processes incorporated in the model are precipitation, infiltration, transpiration, evaporation, lateral flow, vertical flow and plant growth. The principal advantage of this model lies in its ability to provide timely information on the spatially distributed soil moisture content over a given area without the need for repeated field visits. The application of this model to the CATSOP experimental catchment showed that it has the capability to estimate soil moisture content with acceptable accuracy. The root mean squared error of the predicted soil moisture content for 6 monitored locations within the catchment ranged from 0.011 to 0.065cm^3cm^-^3. The predicted daily discharge at the outlet of the study area agreed well with the observed data. The coefficient of determination and Nash-Sutcliffe efficiency of the predicted discharge were 0.824 and 0.786, respectively. BEACH has been developed within freely available GIS and programming language, PCRaster. It is a useful teaching tool for learning about distributed water balance modelling and land use scenario analysis.