Event-based quickflow simulation with OpenLISEM in a burned Mediterranean forest catchment

Abstract

Recently burnt areas typically reveal strong to extreme hydrological responses, as a consequence of loss of protective soil cover and heating-induced changes in topsoil properties. Soil water repellency (SWR) has frequently been referred to as one of the explanatory variables for fire-enhanced surface runoff generation but this has been poorly demonstrated, especially at the catchment scale. This study employs a process-based modelling approach to better understand the relevance of SWR in the hydrological response of a small, entirely burnt catchment in central Portugal, in particular by comparing hydrological events under contrasting initial conditions of dry vs wet soils. The OpenLISEM model was applied to a selection of 16 major rainfall runoff events that occurred during the first 2 post-fire years. The automatic calibration procedure resulted in good model performance, but it worsened for validation events. Furthermore, uncertainty analysis revealed an elevated sensitivity of OpenLISEM to event-specific conditions, especially for predicting the events’ total and peak flows. Also, predicted spatial patterns in runoff poorly agreed with the runoff observed in microplots. Model performance improved when events were separated by dry and wet initial moisture conditions, particularly for wet conditions, suggesting the role of variables other than initial soil moisture.