The impact of lake and reservoir parameterization on global streamflow simulation

Abstract

Lakes and reservoirs affect the timing and magnitude of simulated streamflow, and are therefore essential model components, especially in the context of flood forecasting. However, parameterization of lake and reservoir routines on a global scale is subject to considerable uncertainty due to lack of global information on lake hydrographic characteristics and reservoir operating rules. In this study we estimated the effects of lakes and reservoirs on global daily river streamflow simulations of a spatially-distributed hydrological model, LISFLOOD. Streamflow observations from 1617 catchments around the globe and six different performance metrics were used to assess model performance. State-of-the-art sensitivity and uncertainty analyses were used to examine the effects of lake and reservoir parameter uncertainty on model performance in relation to other parameters. Results indicate a considerable spread in the obtained performance among catchments with lakes and reservoirs. Model skill in terms of Nash-Sutcliff Efficiency (NSE) and Kling-Gupta Efficiency (KGE) improved respectively for 65% and 38%, stations located downstream of lakes and reservoirs. The effect of reservoirs on streamflow climatology was substantial and widespread, while the effect of lakes was spatially limited to a fewer catchments in the global domain. As indicated by sensitivity analysis, reservoir parameters often contributed substantially to model output uncertainty, although the effects varied among catchments. The effects of reservoir parameters on model performance diminished with distance downstream in favor of other parameters (e.g., groundwater-related parameters and channel Manning's roughness coefficient). This study underscores the importance of accounting for lakes and reservoirs and using appropriate parameters in large-scale hydrological simulations, with a focus in the flood forecasting context.