摘要：Selective water release from the deeper pools of reservoirs alters the temperature of downstream rivers. Thermal destabilization of downstream rivers can be detrimental to the riverine ecosystem by disturbing the growth stages of various aquatic species. To predict this impact of planned hydropower dams worldwide, we present a framework called “FUture Temperatures Using River hISTory (FUTURIST).” The framework used historical records of in‐situ river temperatures for existing dams in the U.S. and remote sensing observations for those in other regions to train an artificial neural network (ANN) model that predicts temperature change between upstream and downstream rivers. Validation of FUTURIST‐modeled impacts for dams worldwide showed promising results with a root mean squared error of 2.5°C (0.9°C) and categorical accuracy of 63% (88%) during the summer (winter) season. The trained ANN model afforded prediction of the likely thermal impacts of 216 planned dams. Results suggest that during the summer season, 73% of future dams will potentially cool downstream rivers by up to 6.6°C. Winter season operations were predicted to consistently warm downstream rivers by temperatures of up to 2°C. Reservoirs that experience strong stratification have the most potential to impact downstream pre‐dam thermal regimes. For copious existing or planned dams worldwide that are yet to be mapped of their thermal impacts, FUTURIST provides an efficient path forward to carry out a global thermal assessment and design sustainable hydropower expansion plans so that the upcoming dams can be operated in a more eco‐sensitive manner than the existing ones.
