Melt rates are significantly higher - sometimes up to a factor of 100 - than existing theory would predict.
Glaciers could be melting underwater at a faster rate than previously thought, new research suggests.
Scientists have developed a method to directly measure the submarine melt rate of a tidewater glacier.
Their results suggest current theoretical models may be massively underestimating glacial melt.
Previously direct melting measurements have been made on ice shelves in Antarctica by boring through to the ice-ocean interface beneath.
But where there are vertical-face glaciers that end at the ocean, those techniques are not possible.
University of Oregon oceanographer Dave Sutherland said: "We don't have that platform to be able to access the ice in this way. Tidewater glaciers are always calving and moving very rapidly, and you don't want to take a boat up there too closely."
In a National Science Foundation-funded project, a team of scientists led by Mr Sutherland studied the subsurface melting of the LeConte Glacier, which flows into LeConte Bay south of Juneau, Alaska.
The findings, which scientists say could lead to improved forecasting of climate-driven sea level rise, are published in the journal Science.
In the past, research on the underwater melting of glaciers has mainly relied on theoretical modelling, measuring conditions near the glaciers and then applying theory to predict melt rates. But the theory has rarely been tested, the researchers say.
The team deployed a multibeam sonar to scan the glacier's ocean-ice interface from a fishing vessel six times in August 2016 and five times in May 2017.
Co-author Rebecca Jackson, an oceanographer at Rutgers University-New Brunswick, said: "We measured both the ocean properties in front of the glacier and the melt rates, and we found that they are not related in the way we expected.
"These two sets of measurements show that melt rates are significantly, sometimes up to a factor of 100, higher than existing theory would predict."
There are two main categories of glacial melt - discharge-driven and ambient melt. Subglacial discharge occurs when large volumes, or plumes, of buoyant meltwater are released below the glacier.
When the plume combines with surrounding water to pick up speed and volume, the current steadily eats away from the glacier face. Most previous studies have focused on these discharge plumes.
However, they typically affect only a narrow area of the glacier face, while ambient melt covers the rest of it.
The researchers say predictions have estimated ambient melt to be 10 to 100 times less than the discharge melt, and therefore it is often disregarded as insignificant.
The team found submarine melt rates were high across the glacier's face over both of the seasons surveyed, and that it increases from spring to summer.
While the team recognises the study only focused on one marine-terminating glacier, they believe the new approach should be useful to any researchers.
"Future sea level rise is primarily determined by how much ice is stored in these ice sheets," Mr Sutherland said.
"We are focusing on the ocean-ice interfaces, because that's where the extra melt and ice is coming from that controls how fast ice is lost. To improve the modelling, we have to know more about where melting occurs and the feedbacks involved."
Nina Massey is the PA science correspondent.