Glaciers ending in lakes and oceans drive more than half of global ice loss, study finds

A small fraction of the world's glaciers — those that end not on dry land but in lakes, fjords or against floating ice shelves — are responsible for more than half of all the ice that glaciers worldwide have lost since the turn of the century, according to a study published Friday in Nature Communications. The glaciers in question make up just 3.2% of the more than 216,000 glaciers the researchers examined, yet they cover more than half of the planet's total glacier area and are losing ice at a disproportionate rate, the analysis found.

The team, led by researchers at the University of Birmingham and the University of Leeds, classified 216,502 glaciers worldwide, excluding the Greenland and Antarctic ice sheets, by where each one's leading edge, or terminus, sits: on land, in a freshwater lake, or in the ocean or against a floating ice shelf. Glaciers ending in lakes, oceans or ice shelves lost a combined 134 billion metric tons of ice a year between 2000 and 2019, which the researchers say amounts to more than half of all documented glacier ice loss worldwide over that period. Lake-terminating glaciers lost mass about 57% faster, relative to their size, than otherwise similar glaciers ending on land. The same glaciers also flowed noticeably faster: ice ending in lakes moved about 10% faster, and ice ending in oceans or ice shelves about 44% faster, than land-terminating glaciers of comparable size, elevation and climate.

For context, a separate 2025 assessment known as GlaMBIE, an international collaboration published in Nature, put the overall pace of global glacier loss at roughly 273 billion metric tons a year between 2000 and 2023, with losses accelerating by more than a third in the second half of that period. The new study does not revise that global total; instead, it breaks the loss down by where glaciers end, work the authors say has not been done comprehensively at a global scale before.

Why the edge of a glacier matters

Most standard melt calculations have long treated glaciers as a single category, estimating loss mainly from surface melting driven by air temperature. The Birmingham-Leeds analysis points to a mechanism that approach can miss: when a glacier's snout sits in a lake or the ocean, relatively warm water can erode the submerged ice from below, destabilizing the front and triggering calving — the breaking-off of icebergs — on top of ordinary surface melt. Because that process responds to water temperature and depth as well as air temperature, glaciers with these "wet" termini can retreat and shed mass even during periods when air-temperature-driven melt alone would predict a slower decline. The researchers also published a global dataset classifying each glacier's terminus type, which they say is intended for other modeling groups to use.

A shifting hazard as ice retreats

The findings carry stakes beyond the bookkeeping of global ice totals. Communities in the Andes, the Himalaya, Alaska and other mountain regions increasingly depend on glacial meltwater for drinking water, irrigation and hydropower, and lake-terminating glaciers in particular are associated with a hazard that has drawn growing attention from disaster researchers: sudden glacial lake outburst floods, in which an ice or debris dam holding back a meltwater lake fails and sends a wall of water downstream with little warning. As more glaciers retreat off dry land and into newly formed lakes at their base — a transition the study's authors expect to continue as warming persists — the share of global ice loss driven by contact with water is likely to keep growing. The pattern also bears on sea-level projections, since ocean-terminating glaciers deliver meltwater and icebergs directly to the sea rather than first collecting in an inland lake.

"Glaciers ending in lakes really stand out: they're losing ice particularly quickly," said Joe Mallalieu, who carried out the research while at the University of Birmingham and is now based at the University of Leeds.

Nicholas Barrand, an associate professor of glaciology at Birmingham and a co-author of the study, said the volume of satellite and climate data amassed over the past decade made the global comparison possible. "The last decade has seen a huge increase in the global glacier and climate data available to researchers," Barrand said. He added that tracking how individual glaciers shift from one terminus type to another as they retreat — for instance, draining the lake that had been accelerating their loss, or retreating off the seafloor onto land — will matter increasingly for forecasting their future behavior. The study drew on work by six researchers in total; Phys.org, which first flagged the paper to a general audience, noted that the team combined satellite-derived ice-velocity records with mass-change data spanning nearly two decades.

What's next

The authors argue that the roughly 3% of glaciers that end in water warrant more deliberate treatment in the next generation of global glacier-loss and sea-level-rise projections, many of which have been calibrated primarily against land-terminating glaciers because those are easier to monitor from the ground or with simple satellite imagery. Lake- and ocean-terminating glaciers require additional data on water depth, temperature and calving dynamics that are harder to collect and have historically been patchier outside well-studied regions such as Alaska, Greenland's periphery and Patagonia. The paper arrives roughly a year after a UNESCO-backed international assessment estimated that the world's glaciers have lost more than 6,500 billion tons of ice since 2000, the broader backdrop against which the Birmingham-Leeds team situates its more mechanistic finding about where that loss concentrates. The research was funded by the Leverhulme Trust and a UK-Canada Arctic research partnership known as the IMAGINE project.