On any warm afternoon at a tropical reservoir, you can sometimes see it with the naked eye: clusters of bubbles breaking the surface in irregular bursts, each one carrying a small charge of methane from the muck below. Multiply those bursts across roughly 58,000 large dams worldwide, and the numbers add up fast. A 2021 modeling study published in the Journal of Geophysical Research: Biogeosciences estimated that reservoirs release about 10.1 teragrams of methane per year from their water surfaces alone. Nearly 90 percent of that, around 8.9 teragrams, escapes not as a gentle seep but as violent bubble events from bottom sediments, a process scientists call ebullition. Those figures represent one model-based estimate rather than a consensus value; other approaches to the global budget may yield different totals, and the authors themselves note substantial regional uncertainty.
That distinction matters. Methane is more than 80 times as potent as carbon dioxide over a 20-year window, and as dozens of countries race to build new hydropower capacity to replace coal and gas, the methane escaping from behind their dams risks quietly eroding the carbon savings those projects are supposed to deliver.
The bubble problem, mapped
For years, researchers measuring reservoir greenhouse gases focused on diffusion, the slow molecular transfer of dissolved methane across calm water. That approach consistently underestimated total emissions because ebullition is episodic and spatially patchy. A sampling crew that visits a reservoir on a calm Tuesday morning might record almost nothing; the same site could release a large methane pulse on a warm Thursday afternoon when sediment temperatures spike.
The 2021 study, led by Harrison et al., built a gridded daily dataset at 0.25-degree resolution and was among the first to separate the two pathways globally. Oak Ridge National Laboratory turned that work into a publicly downloadable product, the ORNL Global Reservoirs Methane dataset, which maps where and when bubbles dominate. The patterns are striking: tropical reservoirs, where warm water and abundant organic matter accelerate decomposition in sediments, are the biggest emitters. Shallow zones are especially active because lower water pressure allows gas pockets to expand and break free more easily, creating localized “chimneys” of rising bubbles that can shift position from one day to the next.
A separate global synthesis published in BioScience in 2016 (Deemer et al.), drawing on hundreds of field measurements from reservoirs on every inhabited continent, reached a similar conclusion: methane contributes the bulk of climate forcing from reservoir surfaces, and ebullition is the dominant pathway. The authors warned that studies relying on daytime-only sampling, fair-weather visits, or a single season almost certainly undercount total emissions. Because that meta-analysis is now nearly a decade old, its underlying field data predate many advances in continuous bubble-flux monitoring. More recent syntheses, including the Harrison et al. (2021) modeling work, have built on its foundation, but an updated global meta-analysis incorporating post-2016 field campaigns has not yet been published as of spring 2026.
What happens below the dam
The problem does not stop at the waterline. When methane-rich water from deep in a reservoir passes through turbines or spills over a dam, the sudden pressure drop releases dissolved gas into the air downstream, a process called degassing. A modeling study published in Global Biogeochemical Cycles found that including this pathway pushes total reservoir-system emissions above surface-only figures, sometimes substantially. Deep dams that draw water from methane-saturated layers near the bottom are the worst offenders. (The specific study is frequently cited in summary form by the American Geophysical Union, but the underlying paper’s authorship and DOI have not been independently confirmed for this article, so readers should verify the primary source before relying on precise downstream figures.)
No single agreed-upon multiplier exists for degassing. The increase depends on dam design, turbine intake depth, discharge volume, and how long water has been sitting in the reservoir. A high-altitude run-of-river project with a small, fast-flushing impoundment will behave very differently from a massive tropical storage reservoir with a years-long residence time. As of early 2026, no integrated dataset combines the ORNL daily surface emissions product with downstream degassing estimates in a single consistent framework, leaving system-wide methane budgets partly stitched together from separate lines of evidence.
Scale in context
To understand what 10.1 teragrams means, it helps to place it against the global methane budget. Total methane emissions from all sources, natural and human, run to roughly 580 teragrams per year, according to the Global Carbon Project’s most recent methane budget. Reservoirs, then, account for a small but not trivial slice, comparable in magnitude to emissions from rice paddies or biomass burning in some estimates. And unlike a coal plant, whose emissions are measured at the stack and reported in national inventories, reservoir methane has largely flown under the regulatory radar.
Most national greenhouse gas inventories still treat hydropower as low- or zero-carbon based on fossil fuel displacement, without fully incorporating reservoir methane into life-cycle assessments. That gap has drawn increasing attention since the Global Methane Pledge, launched in 2021, committed more than 150 countries to collectively cut methane emissions 30 percent below 2020 levels by 2030. If reservoir methane is not counted, it cannot be cut.
Measurement push in the U.S.
Federal agencies have started to take the problem seriously on home soil. The Environmental Protection Agency has conducted field measurements at reservoir sites, testing how variables such as algal blooms, sediment temperature, and hydrostatic pressure influence bubble formation. The EPA treats ebullition as the bubble release of methane from sediments and has been working to identify the environmental triggers behind emission surges, including rapid warming events and sudden increases in organic matter inputs from upstream. However, specific project names, site lists, and published results from these EPA campaigns have not been compiled into a single publicly available dataset as of May 2026, making it difficult to evaluate the scope and findings of the work independently.
The Department of Energy has funded multi-reservoir field campaigns designed to capture emissions at high temporal resolution across different seasons and operating conditions. The explicit goal is to measure multiple pathways, including bubbling, diffusion, and degassing, rather than relying on snapshots. As with the EPA efforts, detailed project-level publications and standardized data releases from these DOE campaigns remain limited in the public record. Translating agency measurements into enforceable operating guidelines or revised carbon accounting for hydropower remains an open policy question. As of spring 2026, no U.S. federal regulation requires dam operators to monitor or report reservoir methane.
Can operators do anything about it?
In theory, yes. Researchers at Oak Ridge have noted that reservoir drawdowns and seasonal water-level management could alter greenhouse gas emissions, since shallower conditions appear to favor more bubbling. Adjusting when and how water levels drop might reduce ebullition in some cases. Other ideas floated in the literature include aerating reservoir bottoms to discourage the oxygen-free conditions that produce methane, mixing water columns to prevent stratification, or routing outflows to avoid the most methane-rich layers.
In practice, none of these strategies has been demonstrated at scale in a long-term, peer-reviewed field trial. And some carry trade-offs: drawing down a reservoir to reduce surface ebullition could increase degassing below the dam if methane-saturated deep water is released in the process. Aeration systems require energy, which chips away at a dam’s net power output. Any operational change also has to be weighed against the reservoir’s other purposes, including flood control, irrigation, drinking water supply, and fish habitat.
Direct statements from reservoir operators about emission mitigation strategies are largely absent from the published literature. The gap between what scientists have measured and what dam managers are prepared to act on remains wide.
Where the science still falls short
The 10.1-teragram global estimate, while the most spatially detailed available, relies on modeled extrapolations rather than direct sensor readings at every reservoir. Large regions of sub-Saharan Africa, mainland Southeast Asia, and the Amazon basin have few or no ground-truth measurements, which means local emission rates in those areas carry wider error bars. These happen to be the same regions where hydropower construction is expanding fastest, making the data gap especially consequential.
The BioScience meta-analysis, published in 2016, aggregated studies of varying quality and duration. Many of the underlying field campaigns were not originally designed to capture ebullition, so the synthesis likely still underestimates the most intense, short-lived bursts. Even so, the convergence between that older work and the more recent global modeling strengthens confidence that ebullition dominates reservoir methane emissions.
Why reservoir methane complicates the hydropower calculus
For policymakers weighing the climate implications of new dam construction or comparing hydropower to solar and wind, the most reliable approach is to trace quantitative claims back to underlying datasets, examine how ebullition and downstream pathways were treated, and recognize that current global estimates, while increasingly robust, still carry real uncertainty for individual projects. The bubbles, meanwhile, keep rising.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
