In semi-arid areas, high-intensity rainfall events are often held responsible for the main part of soil erosion. Long-term landscape evolution models usually use average annual rainfall as input, making the evaluation of single events impossible. Event-based soil erosion models are better suited for this purpose but cannot be used to simulate longer timescales and are usually applied to plots or small catchments. In this study, the openLISEM event-based erosion model was applied to the medium-sized (∼50 km2) Prado catchment in SE Spain. Our aim was to (i) test the model's performance for medium-sized catchments, (ii) test the ability to simulate four selected typical Mediterranean rainfall events of different magnitude and (iii) explore the relative contribution of these different storms to soil erosion using scenarios of future climate variability.
Results show that because of large differences in the hydrologic response between storms of different magnitudes, each event needed to be calibrated separately. The relation between rainfall event characteristics and the calibration factors might help in determining optimal calibration values if event characteristics are known. Calibration of the model features some drawbacks for large catchments due to spatial variability in Ksat values. Scenario calculations show that although ∼50% of soil erosion occurs as a result of high frequency, low-intensity rainfall events, large-magnitude, low-frequency events potentially contribute significantly to total soil erosion. The results illustrate the need to incorporate temporal variability in rainfall magnitude–frequency distributions in landscape evolution models. Copyright © 2011 John Wiley & Sons, Ltd.