Physically-based modelling of the post-fire runoff response of a forest catchment in central Portugal: using field versus remote sensing based estimates of vegetation recovery

• Autor(a) principal: Van Eck, Christel M.
• Data de Publicação: 2016
• Outros Autores: Nunes, Joao P., Vieira, Diana C. S., Keesstra, Saskia, Keizer, Jan Jacob
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/29575
• topic: Post‐fire hydrology; Vegetation recovery; Remote sensing; Runoff modelling; LISEM

Forest fires are a recurrent phenomenon in Mediterranean forests, with impacts for human landscapes and communities, which must be understood before they can be managed. This study used the physically based Limburg Soil Erosion Model (LISEM) to simulate rainfall–runoff response, under soil water repellent (SWR) conditions and different stages of vegetation recovery. Five rainfall–runoff events were selected, representing wet and dry conditions, spread over two years after a wildfire which burned eucalypt and maritime pine plantations in the Colmeal experimental micro‐catchment, central Portugal. Each event was simulated using three Leaf Area Index (LAI) estimates: indirect field‐based measurements (TC–LAI), NDVI‐based estimates derived from Landsat‐5 TM and Landsat‐7 ETM+ imagery (NDVI–LAI), and the LAI of a fully restored canopy to test model sensitivity to interception parameters. LISEM was able to simulate events in relative terms but underestimated peak runoff (r2 = 0·36, mean error = −31%, and NSE = −0·15) and total runoff (r2 = 0·52, mean error = −15% and NSE = 0·09), which could be related to the presence of SWR or saturated areas, according to pre‐rainfall soil moisture conditions. The model performed better for individual hydrographs, especially under wet conditions. Modelling the full‐cover scenario showed minor sensitivity of LISEM to the observed changes in LAI. NDVI–LAI data gave a close to equal model performance with TC–LAI and therefore can be considered a suitable substitute for ground‐based measurements in post‐fire runoff predictions. However, more attention should be given to representing pre‐rainfall soil moisture conditions and especially the presence of SWR.