Hydroelectric power is a major component of policies to expand access to electricity and contain greenhouse gas emissions, but this power player can incur significant environmental costs. A computational approach for evaluating basin-level tradeoffs between hydropower and ecosystem services has been developed by an international research team with the goal of guiding sustainable dam siting.

The Amazon EcoVistas artificial intelligence (AI) framework identifies hydroelectric dam portfolios in the Amazon River basin, which currently hosts 158 hydropower dams, that meet energy production goals with the least environmental harm. The scheme analyzes proposed dam projects collectively, both for their energy generation as well as their impacts on the environment., by considering river flow, river connectivity, sediment transport, fish biodiversity and greenhouse gas emissions.

The Amazon EcoVistas algorithm was applied to the existing and 351 proposed dams to generate scenarios based on all possible combinations of these projects. The tool determines the Pareto-optimal frontier, or combination of hydropower projects that minimizes negative environmental effects for any given level of aggregate hydropower output.

The analysis published in Science confirms that lack of strategic coordinated planning has resulted in lost environmental benefits and recommends multi-objective optimization to identify the many dam sites that would yield particularly negative results. Simultaneous consideration of multiple criteria is critical for identifying the least detrimental projects with respect to ecosystem services, including fisheries, biodiversity, floodplain agriculture and undisrupted navigation. Basin-wide analysis, in contrast to planning a smaller scale, is also essential for preventing loss of these ecosystem service benefits.

Researchers from The Nature Conservancy (Colombia), Stanford University, Cornell University, University of Nebraska, Colorado State University, National Institute of Amazonian Research (Brazil), Michigan State University, Universidad de Ingeniería y Tecnología (Peru), Florida International University, Federal University of Juiz de Fora (Brazil), Escuela Politecnica Nacional (Ecuador), Wildlife Conservation Society Peru, Université de Montpellier (France), Universidad San Francisco de Quito (Ecuador), Federal University of Rio Grande do Sul (Brazil), Wildlife Conservation Society New York, The Nature Conservancy (Virginia), CNRS (France) and University of California Santa Barbara contributed to this study.

To contact the author of this article, email shimmelstein@globalspec.com