Haykel Sellami

Parameter and rating curve uncertainty propagation analysis of the SWAT model for two small Mediterranean catchments

Abstract The SWAT model was tested to simulate the streamflow of two small Mediterranean catchments (the Vène and the Pallas) in southern France. Model calibration and prediction uncertainty were assessed simultaneously by using three different techniques (SUFI-2, GLUE and ParaSol). Initially, a sensitivity analysis was conducted using the LH-OAT method. Subsequent sensitive parameter calibration and SWAT prediction uncertainty were analysed by considering, firstly, deterministic discharge data (assuming no uncertainty in discharge data) and secondly, uncertainty in discharge data through the development of a methodology that accounts explicitly for error in the rating curve (the stage−discharge relationship). To efficiently compare the different uncertainty methods and the effect of the uncertainty of the rating curve on model prediction uncertainty, common criteria were set for the likelihood function, the threshold value and the number of simulations. The results show that model prediction uncertainty is not only case-study specific, but also depends on the selected uncertainty analysis technique. It was also found that the 95% model prediction uncertainty interval is wider and more successful at encompassing the observations when uncertainty in the discharge data is considered explicitly. The latter source of uncertainty adds additional uncertainty to the total model prediction uncertainty. Editor D. Koutsoyiannis; Associate editor D. Gerten Citation Sellami, H., La Jeunesse, I., Benabdallah, S., and Vanclooster, M., 2013. Parameter and rating curve uncertainty propagation analysis of the SWAT model for two small Mediterranean watersheds. Hydrological Sciences Journal, 58 (8), 1635−1657.

Quantifying hydrological responses of small Mediterranean catchments under climate change projections

Catchment flow regimes alteration is likely to be a prominent consequence of climate change projections in the Mediterranean. Here we explore the potential effects of climatic change on the flow regime of the Thau and the Chiba catchments which are located in Southern France and Northeastern Tunisia, respectively. The Soil and Water Assessment Tool (SWAT) hydrological model is forced with projections from an ensemble of 4 climate model (CM) to assess changes and uncertainty in relevant hydrological indicators related to water balance, magnitude, frequency and timing of the flow between a reference (1971-2000) and future (2041-2071) periods. Results indicate that both catchments are likely to experience a decrease in precipitation and increase in temperature in the future. Consequently, runoff and soil water content are projected to decrease whereas potential evapotranspiration is likely to increase in both catchments. Yet uncertain, the projected magnitudes of these changes are higher in the wet period than in the dry period. Analyses of extreme flow show similar trend in both catchments, projecting a decrease in both high flow and low flow magnitudes for various time durations. Further, significant increase in low flow frequency as a proxy for hydrological droughts is projected for both catchments but with higher uncertainty in the wet period than in the dry period. Although no changes in the average timing of maximum and minimum flow events for different flow durations are projected, substantial uncertainty remains in the hydrological projections. While the results in both catchments show consistent trend of change for most of the hydrologic indicators, the overall degree of alteration on the flow regime of the Chiba catchment is projected to be higher than that of the Thau catchment. The projected magnitudes of alteration as well as their associated uncertainty vary depending on the catchment characteristics and flow seasonality.

Simulation of future groundwater recharge using a climate model ensemble and SAR-image based soil parameter distributions — A case study in an intensively-used Mediterranean catchment

We used observed climate data, an ensemble of four GCM-RCM combinations (global and regional climate models) and the water balance model mGROWA to estimate present and future groundwater recharge for the intensively-used Thau lagoon catchment in southern France. In addition to a highly resolved soil map, soil moisture distributions obtained from SAR-images (Synthetic Aperture Radar) were used to derive the spatial distribution of soil parameters covering the full simulation domain. Doing so helped us to assess the impact of different soil parameter sources on the modelled groundwater recharge levels. Groundwater recharge was simulated in monthly time steps using the ensemble approach and analysed in its spatial and temporal variability. The soil parameters originating from both sources led to very similar groundwater recharge rates, proving that soil parameters derived from SAR images may replace traditionally used soil maps in regions where soil maps are sparse or missing. Additionally, we showed that the variance in different GCM-RCMs influences the projected magnitude of future groundwater recharge change significantly more than the variance in the soil parameter distributions derived from the two different sources. For the period between 1950 and 2100, climate change impacts based on the climate model ensemble indicated that overall groundwater recharge will possibly show a low to moderate decrease in the Thau catchment. However, as no clear trend resulted from the ensemble simulations, reliable recommendations for adapting the regional groundwater management to changed available groundwater volumes could not be derived.

Climate models and hydrologic parameter uncertainties in climate change impacts on monthly runoff and daily flow duration curve of a Mediterranean catchment

ABSTRACT Climate models and hydrological parameter uncertainties were quantified and compared while assessing climate change impacts on monthly runoff and daily flow duration curve (FDC) in a Mediterranean catchment. Simulations of the Soil and Water Assessment Tool (SWAT) model using an ensemble of behavioural parameter sets derived from the Generalized Likelihood Uncertainty Estimation (GLUE) method were approximated by feed-forward artificial neural networks (FF-NN). Then, outputs of climate models were used as inputs to the FF-NN models. Subsequently, projected changes in runoff and FDC were calculated and their associated uncertainty was partitioned into climate model and hydrological parameter uncertainties. Runoff and daily discharge of the Chiba catchment were expected to decrease in response to drier and warmer climatic conditions in the 2050s. For both hydrological indicators, uncertainty magnitude increased when moving from dry to wet periods. The decomposition of uncertainty demonstrated that climate model uncertainty dominated hydrological parameter uncertainty in wet periods, whereas in dry periods hydrological parametric uncertainty became more important. Editor M.C. Acreman; Associate editor S. Kanae

Assessment of the SWAT model prediction uncertainty using the GLUE approach A case study of the Chiba catchment (Tunisia)

Hydrological models predictions are always affected with uncertainty that have to be addressed in order to make appropriate use of these models in water resources studies and management. In this paper the efficiency of the SWAT model for discharge prediction in partially gauged semi-arid catchment is evaluated and the model prediction uncertainty is assessed using the GLUE approach. Based on the results, SWAT can be used to predict the discharge in semi-arid catchment but its efficiency is marked with high variability due to the inter-annual variability of rainfall. The uncertainty analysis of the model prediction suggests that model parameters uncertainty alone cannot compensate for all uncertainty sources and that in order to provide more realistic uncertainty estimation all uncertainty sources have to be considered.