Colin Gleason (University of Massachusetts, United States); Casey Brown (University of Massachusetts, United States); Jida Wang (University of Illinois Urbana-Champaign, United States); Rangel Daroya (University of Massachusetts, United States); Fiona Bennitt (University of Massachusetts, United States); Elisa Friedmann (University of Massachusetts, United States); Dongmei Feng (University of Cincinnati, United States)

Dams control a substantial portion of global river systems. Numerous efforts have been made to create comprehensive data on their location and hydropower capacity. However, to date, these dam databases remain incomplete, particularly in regions where validation from remote sensing approaches and ground-based observations is limited. This study presents a deep learning framework to detect dams using direct observations from the recently launched Surface Water and Ocean Topography (SWOT) satellite mission. By analyzing node-level river profiles of water surface elevation, surface area, and width, we trained a one-dimensional convolutional neural network to classify dammed reaches based on Global Dam Watch, GeoDAR, and Global River Obstructions Database. The best dam detection model from our initial tests achieved an average accuracy of 0.83, with precision, recall, and F1-score of 0.82, 0.85, and 0.83, respectively. From this model, we can identify undocumented dams missing from existing databases, and also differentiate hydropower-producing dams from non-hydropower dams. Then, we will estimate the potential hydropower generation of rivers globally based on SWOT-estimated discharge and hydraulic head taken from height observations on either side of dams, which is expected to reveal the untapped capacity in underreported basins and the currently tapped but unreported capacity. This study will offer a reproducible method for monitoring dam infrastructure and evaluating hydropower resources without relying on national reporting or in-situ data. Our findings will highlight the potential of SWOT to advance global dam mapping and hydropower assessment.