Keywords

Model comparison; Sub-daily simulation; Surface runoff; River discharge; Mediterranean watershed.

Location

Session A1: Environmental Fluid Mechanics - Theoretical, Modelling and Experimental Approaches

Start Date

12-7-2016 9:30 AM

End Date

12-7-2016 9:50 AM

Abstract

Due to climate change, the frequency of intense rainfall events and consequent flash floods are expected to increase in the next decades across the Mediterranean coastal basins. To date, few spatially-explicit models are able to simulate flash floods with accurate details. The MARINE model is one of them: it is a process-oriented fully distributed model operating dynamically at the rainfall event time-scale. It includes infiltration and saturation excess processes along with subsurface, overland and channel flows. It does not describe ground-water processes as the model's purpose is to simulate individual flood events during which ground-water processes are considered negligible. The SWAT model is a conceptual time-continuous semi-distributed model that dynamically simulates above- and below-ground processes assuming several simplifications in equations. It has been recently upgraded to sub-daily time-step calculations. However, its sub-daily module has only been tested in small catchments (~1 km²). Considering a ~800 km² Mediterranean river coastal basin (southwestern France) as a case-study, the objective of this study was to assess the ability of the SWAT model to simulate the discharge during flash floods by comparing its performances to the performances of the MARINE model. We first calibrated and validated the two models based on the same input dataset (topography, land- use, soil classes, and hourly rainfall data). We then compared the performances of the two models during 6 major flood events (2009-2013). Nash-Sutcliffe efficiencies vary from -0.79 to 0.54 for the SWAT model and from 0.32 to 0.80 for the MARINE model. Results show that the SWAT model is able to reproduce peak discharges with a satisfactory accuracy with respect to the MARINE model. Next step will be to implement the SWAT model to simulate the hydrology of the Gulf of Lion along with the suspended sediment load input into the Mediterranean Sea.

Share

COinS
 
Jul 12th, 9:30 AM Jul 12th, 9:50 AM

Modelling Flash Floods at Sub-daily Time-step: Comparison of the Performances of the Conceptual SWAT Model and the Process-oriented MARINE Model

Session A1: Environmental Fluid Mechanics - Theoretical, Modelling and Experimental Approaches

Due to climate change, the frequency of intense rainfall events and consequent flash floods are expected to increase in the next decades across the Mediterranean coastal basins. To date, few spatially-explicit models are able to simulate flash floods with accurate details. The MARINE model is one of them: it is a process-oriented fully distributed model operating dynamically at the rainfall event time-scale. It includes infiltration and saturation excess processes along with subsurface, overland and channel flows. It does not describe ground-water processes as the model's purpose is to simulate individual flood events during which ground-water processes are considered negligible. The SWAT model is a conceptual time-continuous semi-distributed model that dynamically simulates above- and below-ground processes assuming several simplifications in equations. It has been recently upgraded to sub-daily time-step calculations. However, its sub-daily module has only been tested in small catchments (~1 km²). Considering a ~800 km² Mediterranean river coastal basin (southwestern France) as a case-study, the objective of this study was to assess the ability of the SWAT model to simulate the discharge during flash floods by comparing its performances to the performances of the MARINE model. We first calibrated and validated the two models based on the same input dataset (topography, land- use, soil classes, and hourly rainfall data). We then compared the performances of the two models during 6 major flood events (2009-2013). Nash-Sutcliffe efficiencies vary from -0.79 to 0.54 for the SWAT model and from 0.32 to 0.80 for the MARINE model. Results show that the SWAT model is able to reproduce peak discharges with a satisfactory accuracy with respect to the MARINE model. Next step will be to implement the SWAT model to simulate the hydrology of the Gulf of Lion along with the suspended sediment load input into the Mediterranean Sea.