Impact of land use land cover changes on flow uncertainty in Siliana watershed of northwestern Tunisia

Abstract

It is widely admitted that changes in land use land cover (LULC) influence the hydrology of the catchment. However, how these changes affect hydrological model prediction uncertainty is still a raising question. In this paper we addressed this question by investigating the impacts of the rapid change in LULC in the Siliana catchment in Tunisia on monthly flow and magnitude of flow extremes using the SWAT hydrological model while quantifying the contribution of LULC to the model parameter and prediction uncertainty. At a first step, the SWAT model parameter and prediction uncertainty were estimated using the GLUE method and acceptable parameter sets were identified. Subsequently, the SWAT model was fed with historical LULC as derived from Landsat 5 and 8 satellite images for the years 1990, 2000, 2013 and 2019, and run with the acceptable parameter sets. The results show that the increase in olive plantations (+380 %), urban area (+200 %), and irrigated lands (+309 %) from 1990 to 2019, has LULC decreased monthly flow, high flows magnitude but did not impact low flows in particular over the previous two decades. The findings also suggest that model prediction uncertainty can mask LULC effects, suggesting that model results can be misleading without explicit consideration of uncertainty when assessing the hydrological impacts of changes in LULC.