Abstract
Reservoir sedimentation is an escalating threat to global water security. While most existing assessments focus on large, well-managed reservoirs, they overlook the vast and vulnerable network of smaller reservoirs, which hampers our understanding of global sedimentation risks. Here we present a high-resolution global assessment of reservoir sedimentation in over 550,000 reservoirs, ~95% of which are <1 km2, based on a physics-guided machine learning model and a comprehensive reservoir inventory. Sedimentation rates have been underestimated in over 75% of reservoir-bearing regions when large reservoirs alone are considered. This oversight conceals a loss in global average reservoir water storage of 7.3% ± 2.8% per decade, with nearly one in five reservoirs now facing high-risk status. We identify 16 global high-sedimentation hotspots across the Western Americas and Afro-Eurasian Sediment Belts, threatening water supplies to over 2 billion people and impacting ~26% of global irrigated land. Without intervention, we project that, by 2060, more than half of all reservoirs could become functionally inoperable, including 58.6% of small and 38.1% of large reservoirs. These findings highlight an urgent need for targeted, nature-based sediment management, particularly in dryland regions, where small reservoirs are critical for domestic water access and food production.
My town’s reservoir was built in the 1950s, several decades before the seismic risks of the USA west coast were fully understood. There is a real threat of the dam collapsing if there is an earthquake, so they can no longer store water (its filled 1/8th). It’s unclear what the options for replacing the dam are or how it would be paid for. All the shoreline infrastructure like campgrounds, beaches and boat ramps are comically high above the water level.
There are so many infrastructure investments that could become stranded assets in the future. This dam also had power generation, water control, flood protection and all kinds of other uses, many of these systems also need maintenance and reinvestment. At what point does it stop making sense to invest?

