Abstract

General Circulation Models (GCMs), Greenhouse Gases (GHG) concentration scenarios, and downscaling methods are typically used to generate future non-stationary climate time series and represent local climate conditions, particularly when assessing the hydrologic impacts of climate change at basin scales. Unfortunately, due to its complexity and the computational cost, uncertainty analysis of coupled GCM-GHG is typically simplified or even neglected. Moreover, existing uncertainty analyses commonly focus on the impact of the GCM and GHG scenario selection on precipitation and flow discharge simulated by hydrological models. To the best of our knowledge, the uncertainty in reservoir performance due to the GCM-GHG selection has not been studied yet. This study evaluates this uncertainty for a reservoir system located in the Limarí River basin, Chile. For this purpose we selected 149 precipitation and temperature realizations which combine different GCMs and Representative Concentration Pathways (RCP). These climate time series were then downscaled using a quantile mapping bias correction method, and used as inputs in a hydrologic/reservoir operation model. Finally, we characterized the performance of the reservoir system under future flow discharges using reliability, resilience, and vulnerability indices. Results show strong correlations between these indices and precipitation changes associated with each of the GCM-GHC realization. These relationships were then used to group the GCM-GHG in terms of their expected impact over the performance of the reservoir system. Overall, the study demonstrates the relevance of the GCM-GHG selection when studying the future operation of hydraulic infrastructure under climate change, and the necessity of a proper selection to avoid bias in the performance estimation.