Estimation of the uncertainty of hydrologic predictions in a karstic Mediterranean watershed

The Koiliaris River basin is a semi-arid Mediterranean karstic watershed where water needs during the summer are exclusively covered by the karstic springs flow. Uncertainty assessment of the hydrologic projections for karstic watersheds may reveal possible water deficits that cannot otherwise be taken into account. The Soil Water Assessment Tool (SWAT) along with a karstic model (Karst-SWAT) is used to assess the composite spring and surface flow. The parameter uncertainty of both the surface and karstic flow models is estimated by combining the SUFI2 interface and the @RISK by PALISADE software. Eleven combinations of five Regional Climate Models (RCMs) and three Representative Concentration Pathways (RCPs) provide input to the hydrologic models. Representative rainfall time series for certain scenarios are stochastically modeled with the LARS weather generator. Monte Carlo simulations are used to investigate the effect of input internal variability on the flow output. The uncertainty of karstic flow due to the parameter uncertainty of the SWAT and Karst-SWAT models is 10.0% (Coefficient of Variation), which is comparable to the estimated uncertainty due to climate change scenarios (10.1%) until 2059. The combined uncertainty for the total flow at the basin exit due to both models' parameter uncertainty is 6.6%, comparable to the uncertainty due to the internal variability (5.6%). The total uncertainty of karstic flow, combining model parameter uncertainty and the internal variability of the climate scenarios is 11.0%. The total uncertainty estimate is used in conjunction with the lowest karstic flow projection to assess the most adverse scenario for the future mean annual karstic flow. This is the first study which estimates the combined uncertainty of surface and karstic flow prediction due to model parameter uncertainty and internal variability. Our study provides a rigorous methodology for uncertainty estimation and analysis which is transferable to other karstic regions of the world. (c) 2020 Elsevier B.V. All rights reserved.