3rd HyMeX workshop 1-4 June 2009 Heraklion (Gournes), Crete-Greece
Assessment of initial soil moisture conditions to improve event-based rainfall-runoff models in a small headwater catchmentYves Tramblay (HSM/IRD); Christophe Bouvier, Claude Martin, Jean-François Didon-Lescot, Dragana Todorovik, Jean-Marc Domergue
Flash floods are the most destructive natural hazard that occurs in the Mediterranean region. Rainfall-runoff models can be very useful for flash flood forecasting and prediction.
Event-based models are very popular for operational purposes, but their efficiency s is still limited because of the uncertainties related to the initial moisture conditions prior to a flood event. This study aims to improve existing event-based rainfall-runoff models by using several soil moisture indicators: local TDR measurement of soil moisture, outputs of a SVAT model (SIM), antecedent precipitation and base flow. A distributed version of the SCS rainfall-runoff model is used to model the flood events in a small headwater catchment in the Cevennes region (France).
The model involves three parameters: the S runoff parameter, considered here as the water deficit in the soils of the catchment at the beginning of each event, and Vo and Ko, which are routing parameters and remain the same for all the events. The model was calibrated from a 21-flood sample, and led to a mean Nash value of 0.76.
The local TDR measurements, in particular in the deepest layers of soil (80-140cm), were found to be the best predictors of the calibrated initial soil moisture (r2 =0.85). TDR measurements averaged over the whole soil profile, SIM, and base flow were found to be less efficient predictors, since r2 coefficients decreased to respectively 0.79, 0.73 and 0.76. The good correlations observed between the TDR predictors and the S calibrated values indicate that monitoring soil moisture could greatly improve simplified event-based models.