H53J-07
On Subsurface Flow Modeling in a Physical-based Model Dedicated to Flash Flood Forecasting

Friday, 18 December 2015: 15:10
3020 (Moscone West)
Audrey Douinot1, Hélène Roux1, Kevin Larnier1, David Labat2 and Denis Dartus1, (1)Institut de Mécanique des Fluides de Toulouse (IMFT) - CNRS, Toulouse, France, (2)Observatory Midi-Pyrenees, Toulouse, France
Abstract:
Spatial and temporal soil moisture dynamic is a major factor influencing flash flood modelling. Interactions between localized, intense rainfall patterns and initial soil moisture spatial distributions make soil saturation excess to be reached quickly (ref: Anquetin et al., 2010 and Zocatelli et al., 2010) The few observations on hillslope flows suggest significant lateral flow transfer from the soil layers to the river. Moreover, Vannier et al. (2013) and Garambois et al. (2015) point out the importance of the catchment geological properties on flash flood dynamics, and in particular on the water storage during those events.

According to these observations, this study proposes a new modelling of the soil water dynamic in the distributed and physical-based model MARINE. The objective is to improve the representation of the physical processes in an operational flash flood model.

The soil column is divided in two compartments in order to represent in the upper layer the interflow in the vadose-zone with transmissivity and suction forces depending on Van Genuchten’s formula and in the lower layer the groundwater flow and the storage in the weathered bedrock with a free water table model based on Darcy’s law. Furthermore boundary conditions of lateral flow at channel interface are adapted using the conductance concept in order to better represent the dynamic interaction between both hydrological elements.

The model is applied to several catchments of the French Mediterranean area ranging from 100 km2 to 500 km2. As expected, the new process representation impacts the simulation of the flow dynamic. The rising time of the subsurface flow is consistent with the catchment characteristic response time of a storm event. The recession curves are also improved. The sensitivity analysis suggests a strong influence of stream-aquifer flows and their representation.