Modeling sediment delivery from a highly erodible mountain catchment

Abstract:

Draix observatory is located in the French Alps on a highly erodible substrate of shale. Most of the observatory is in a badland area characterized by steep gullies and high erosion rates (up to 1cm/year). Within the observatory, the study focuses on the Moulin, which is an 8ha catchment located at an elevation of 850-925m, with 54% of badland area.

Available data includes DEM, meteorological data, high-frequency records of discharge and suspended sediment concentration during the floods, cumulative values of bedload transport for each flood, high-frequency records of bedload transport for a few events from a Birkbeck sampler.

Modeling sediment delivery in such a catchment is challenging because 1) most available models have been designed for low-relief regions and do not account for steep slope processes such as debris flow and landslides; 2) hydrology (especially flashfloods) in mountainous regions is not well understood; 3) soil properties are very heterogeneous ; 4) multiple time scales are involved: seasonal sediment production on the slopes, storage in the bed and exportation requires to work on yearly times scales, while summer floods and most sediment delivery events occur over a few minutes only.

We evaluate the ability of the SHETRAN model to reproduce sediment delivery patterns from the catchment. First, we calibrate the hydrological model using one year of meteorological and hydrological data. We then apply the sediment transport module over several flood events, using in-situ measurements of bed and slope grain-size distributions. Finally we investigate how sediment available on the slopes moves through the catchment over a year.

Event-scale volumes of sediment simulated by the model are comparable to observed values within an order of 2. Sediment delivery rates are very sensitive to the slope grain-size distribution. Depending on sediment availability on the slopes and on soil erodibility, the catchment is running either in a supply-limited or transport-limited mode. Finally, when feeding the model with frost-induced loose sediment on the slopes at the end of winter, as suggested by field observations, the model is able to reproduce a seasonal pattern of slope erosion, temporary storage in the upper catchment then bed scouring.