Runoff and erosion in a small burnt catchment in central Portugal

Post-fire erosion is a major concern because of its potential effects on soil and water resources. Wildfire is frequently observed to increases soil erosion rates (Shakesby and Doerr, 2006). Many studies have emphasized the role of the partial or entire elimination of vegetation and ground cover (thereby exposing the mineral soil to direct impact of raindrops and overland flow) as a key factor in the increased soil losses. Also fire-induced changes in topsoil properties like total porosity (e.g. Neary et al., 2005) and water repellency (e.g. Keizer et al., 2008a) can enhance overland flow and the associated transport of soil particles. Existing process-based erosion models like e.g. WEPP (Nearing et al., 1989) and EUROSEM (Morgan et al., 1998) can successfully simulate the effect of vegetation on soil losses in individual storms but their complexity often limits their usefulness for non-academic purposes. Therefore, empirical models such as USLE (Wischmeier and Smith, 1978) continue to be widely used. For example, the Portuguese national Water Institute (INAG) applied USLE to the 300.000 ha of rural areas affected by wildfires during the summer of 2003. However, the model’s suitability for burnt areas (as opposed to the agricultural areas for which USLE was originally developed) was not supported by field observations. The present work concerns the EROSFIRE-II project (PTDC/AGR-CFL/70968/2006), which has as overall aim to assess the suitability of selected models for predicting soil erosion in recently burnt forest areas in Portugal. The proposed presentation will focus on the evaluation of the revised Morgan-Morgan-Finney model (MMF; Morgan, 2001) against slope-scale measurements of overland flow and/or erosion during the first year after fire. Earlier studies reported this model to perform satisfactorily in forests and shrublands in the humid areas of the north-eastern Iberian Peninsula (Keizer et al., 2008b; Fernández et al., 2010). The hydrological and erosion response at the catchment scale will be modeled with MEFIDIS (Nunes et al., 2005) but that is beyond the present scope