Radar satellites track Alaska's glaciers through clouds, revealing three extra melt weeks per degree

Published in Nature, February 4, 2026. Carnegie Mellon University and University of Alaska Fairbanks Geophysical Institute.

Glaciologists have a timing problem. The instruments they traditionally use to assess glacier health — optical cameras mounted on satellites and aircraft — need clear skies and good light. In Alaska, where clouds, fog, and polar darkness dominate much of the year, that means critical data often arrives late, arrives degraded, or doesn't arrive at all.

A study published in Nature on February 4 demonstrates that synthetic aperture radar, or SAR, can bypass those limitations entirely — and in doing so, it has produced the most complete picture yet of how Alaska's glaciers respond to warming.

Optical instruments versus radar: what changes

The traditional approach to glacier monitoring relies on optical imagery captured at the end of the melt season, typically late summer or early fall. Scientists look for the snowline — the boundary between the upper accumulation zone, where snow builds up, and the lower ablation zone, where ice melts away. The position of that line indicates how much ice the glacier gained or lost over the year.

The problem, as UAF Geophysical Institute scientist Mark Fahnestock puts it, is that the window for a clean optical reading is narrow and unreliable. If an early snowfall blankets the glacier before the satellite passes overhead, the snowline vanishes. Variable lighting, cloud cover, and the difficulty of distinguishing dirty firn (partially compacted granular snow) from bare ice all introduce noise.

SAR works differently. It sends microwave pulses from orbit and reads the returning echoes to build detailed images. Microwaves pass through clouds and work in darkness. The European Sentinel-1 satellite covers the same location every 12 days, and its radar can distinguish wet, melting ice from dry, frozen surfaces regardless of weather or season.

The result is continuous monitoring where before there were seasonal snapshots.

Three additional melt weeks per degree Celsius

Lead author Albin Wells, a recent Ph.D. graduate from Carnegie Mellon University, used Sentinel-1 data to track changes across nearly all Alaska glaciers larger than about half a square mile — more than 3,000 glaciers in total — from mid-2016 through 2024. Co-authors include assistant professor David Rounce of Carnegie Mellon and Fahnestock.

The team measured melt days, a metric that captures both the duration and spatial extent of melting. A single melt day can represent one 24-hour period when an entire glacier surface is actively melting, or it can reflect several days of partial melt that together equal the glacier's total surface area.

The central finding: every 1 degree Celsius increase in average summer temperature corresponds to approximately three additional weeks of melt days across Alaska's glaciers. More melt days mean longer exposure of bare ice, accelerated mass loss, and a glacier that ends each year smaller than it started.

The 2019 heat wave: a case study in rapid response

Between June 23 and July 10, 2019, Alaska experienced a heat wave that engulfed every glaciated region except the Brooks Range. Temperatures ran 20 to 30 degrees Fahrenheit above average at many locations for nearly two weeks. Anchorage hit 90 degrees Fahrenheit — in a city where typical summer highs sit in the mid-60s.

The radar data captured the glaciers' response in near-real time. Snowlines retreated nearly 350 feet in elevation during the event — a shift that, in a typical year, would not occur until roughly two months later. Heat waves also caused glaciers to lose up to 28% more of their protective snow cover than in normal years, measured at the mountain-range scale.

The speed of that response matters. Glaciers are often discussed in terms of decadal trends, but the 2019 data shows they react almost immediately to short-term heat extremes. Two weeks of unusual warmth accelerated the melt calendar by two months. As heat waves become more frequent under continued warming, these acute shocks could compound the slower, background trend of rising temperatures.

Coastal glaciers versus continental glaciers

The study also revealed consistent differences between glaciers on the coastal side of Alaska's mountain ranges and those farther inland. Coastal glaciers experience more melt days in summer but also accumulate more snow in winter. Continental glaciers melt less but also gain less. The two populations are losing ice at broadly similar rates through different mechanisms — a finding that matters for any attempt to model future glacier behavior.

"This is an important finding," Wells said, "because it corroborates prior knowledge that glaciers in Alaska on the coastal side of mountains have more melt in summer and more accumulation in winter than those on the continental side of the ranges."

The distinction has practical consequences. Meltwater from coastal glaciers feeds rivers that supply hydropower, fisheries, and municipal water systems. Continental glacier loss affects different watersheds. Understanding which glaciers are losing ice through which mechanisms is essential for downstream planning.

What radar still cannot do

SAR monitoring has clear advantages over optical methods — consistency, weather independence, year-round coverage — but it is not a complete replacement for ground-based measurement. The radar data serves as a proxy for glacier mass balance, not a direct measurement of it. Converting melt days and snowline positions into precise ice volume changes still requires calibration against field observations, which remain sparse for many of Alaska's more remote glaciers.

The Sentinel-1 record also begins only in 2016 for Alaska, meaning the radar dataset is still too short to establish long-term trends independent of the optical record. And the 12-day revisit cycle, while far better than seasonal snapshots, can still miss rapid events that unfold over days.

But as a monitoring system, the approach is now operational. What Wells built can be applied to glaciers anywhere Sentinel-1 data is available, which is most of the planet.

"What Albin has done is operationalize the tracking of surface conditions on the glaciers in a way that can be applied anywhere," Fahnestock said.

The three-weeks-per-degree relationship gives climate modelers a concrete parameter to work with. It also gives anyone paying attention a simple, sobering metric: every degree of warming buys Alaska's glaciers another three weeks of melting they cannot afford.