Uncertainty analysis due to Digital Elevation Models (DEMs) in distributed catchment modelling is applied to Var catchment in Southern France. The Catchment is represented on a 384 384 array of squares having dimensions of 300 300 m2 and the ground level varies from Sea level to more than 3000 m. Catchment topography is reconstructed between grid points by extrapolating the slopes of the topography at the adjacent grid points to represent the real topographic feature that are smaller than DEM resolution. A grid point is shifted either to the left or to the right up to a maximum of half a grid cell in both and directions. Uniform weighting factors and are applied in both directions to consider the spatial variability of DEM points and random possible DEMs are generated by varying the weighting factors as and . Each of the generated random DEMs is then used as input in rainfall-runoff modelling. The analysis shows that rainfall-runoff modelling is not sensitive to DEM. It also shows that the simulation produced wrong result that can not be and of course should not be corrected by calibration. Also, rainfall-runoff modelling is done by including subsurface flow from topsoil, which is a physical behavior of most mountainous catchment. The result shows that the peak of the hydrograph increases and the falling limb becomes higher (increase in magnitude). Finally, the sensitivity of ground water flow modelling to aquifer parameters (hydraulic conductivity, leakage coefficient and specific yield) is analyzed for average potential head and variations (standard deviations) of the potential head. The result shows that average potential head is sensitive to specific yield and leakage coefficient and the standard deviation is sensitive hydraulic conductivity and specific yield.
Back to the list of MSc abstracts