← All ice fields

Methodology

How this was made

Each ice field on this site has four things to look at: a satellite picture for every summer since 1986, a chart of how much ice the satellites saw, a map of how warm each summer was, and an estimate of how much ice has been lost. This page explains, in plain words, where each one comes from and how far to trust it.

The satellite pictures

The pictures come from Landsat, a series of satellites run by NASA and the U.S. Geological Survey that has photographed the whole Earth again and again since the 1980s. There is one picture for each summer.

Why summer. Each picture is taken at the end of the melt season, when last winter’s snow has gone and the glacier ice itself is easiest to see. In the north that is July to September. In Southern Patagonia it is January to April, and on the Antarctic Peninsula it is late November to the start of March. For those two southern sites, a year means the summer that ends in that year: Patagonia “2026” is January to April 2026.

How clouds are removed. A single satellite photo is often half covered by cloud. So every photo taken over a site in one summer is laid on top of the others, and the clear parts of each are kept. The result is one mostly cloud-free picture of that summer.

Pictures, not measurements. The colour of the sea is evened out so that a hazy year does not look like a different ocean. This makes the pictures easier to read, but it also means they are pictures to look at, not measurements. No number on this site is taken from them.

Why some sites start later than 1986. In the early years, the satellites photographed some remote places rarely or not at all. Where a summer has too little to show, there is no picture for it. Several sites in the far north of Canada, for example, begin in the 1990s.

Around each picture is a darker background image of the surrounding land and sea. It is there only so the ice field sits in its landscape. It is not from any one year, and nothing is measured on it. Each map is drawn so that areas appear at their true size, which ordinary web maps do not manage near the poles.

The ice area chart

The chart shows how many square kilometres of ice the satellites could see each summer. Snow and ice are bright in ordinary light but dark in infrared light, which almost nothing else in a mountain landscape is. A computer uses that difference to mark every patch of ice in each summer’s Landsat data, and adds the patches up. Only ice inside the mapped outline of the site’s glaciers is counted, so sea ice and distant snowfields are left out.

These are our own estimates, made for this site. They are not official figures, and there are a few things they cannot do:

For these reasons the chart shows each summer as it was seen and draws no trend line through them. A line that stays flat can also be a real result: published studies find that the Karakoram’s glaciers have held steady or grown slightly since the 1990s.

Sites with no area chart. Greenland’s four sites and the NE Antarctic Peninsula show pictures but no area. In Greenland we do not yet have a reliable outline of the ice to count inside, and without one the figure would mostly follow the sea ice and snow around it. Over Antarctica, the satellite data is not positioned precisely enough to measure with.

The temperature layer

The Show temperature button colours the map by how warm that summer was, compared with the average for the same place and the same weeks of the year. Red means warmer than average and blue means colder, on a scale from 3 °C colder to 3 °C warmer. The average is taken over the thirty summers from 1991 to 2020. The weeks are the same ones the satellite picture covers.

The data is the temperature of the air about two metres above the ground. It comes from ERA5, a reconstruction of the world’s past weather made by combining millions of measurements from weather stations, balloons, ships and satellites. ERA5 describes the weather in squares about 28 km across, so the layer shows the weather of the region, not of a single glacier.

Why air temperature. It is the best simple guide to how much a glacier melts: warm air heats the ice directly, and a warm summer usually brings more of the other things that melt ice too. The temperature of the ice surface itself would show very little, because melting ice stays at 0 °C however warm the summer is.

The layer is there for context. It does not change any picture or any figure on the site. The figure for the most recent summer can still shift slightly, because the newest ERA5 data is revised a few months after it first appears.

The ice volume lost

Each site page opens with how much glacier ice was lost between 2000 and 2024, and how many Olympic swimming pools the meltwater would fill (one pool holds 2,500 cubic metres). Area and volume answer different questions. A glacier can get thinner for decades while its edge barely moves, so the area can stay almost the same while a great deal of ice is lost.

Volume cannot be read from a picture, so we do not measure it ourselves. The figures come from the World Glacier Monitoring Service, which publishes an estimate for every glacier in the world. It combines two kinds of evidence: satellite measurements of how much the surface of each glacier has dropped over the years, and about 500 glaciers that scientists visit and measure on the ground. Single years in that record are partly estimated, so the site shows five-year periods, never one year.

These are estimates with a wide margin of error. At some sites, such as the Karakoram, the margin is about as large as the loss itself, and the smallest five-year figures cannot be told apart from no change. The volume is never drawn on the same chart as the area, because nobody has measured how the two are linked at these sites.

Data sources

Everything on this site is built from data that is free and open to the public. These are the sources, what each one is, and what it is used for here.

Landsat satellite images

U.S. Geological Survey (USGS) and NASA · usgs.gov/landsat-missions

Landsat is the longest continuous record of the Earth’s surface seen from space. Its satellites (Landsat 4, 5, 7, 8 and 9 are used here) pass over the same spot every 8 to 16 days and see detail down to 30 metres, in ordinary light and in infrared. Because the same kind of image exists for every year since the 1980s, it is the one source that lets a summer in 1986 be compared fairly with a summer in 2026.

Used for: every satellite picture, and every ice area figure.

Landsat Collection 2 imagery courtesy of the U.S. Geological Survey.

ERA5 weather record

Copernicus Climate Change Service, produced by the European Centre for Medium-Range Weather Forecasts (ECMWF) · Climate Data Store

ERA5 is a reconstruction of the weather everywhere on Earth for every month since 1940. Most glaciers have no weather station anywhere near them, and ERA5 is the standard way climate scientists fill that gap. It gives a consistent temperature for all of the sites, back to the first picture.

Used for: the temperature layer (monthly average air temperature, two metres above the ground).

Contains modified Copernicus Climate Change Service information. Neither the European Commission nor ECMWF is responsible for any use of it.

Glacier mass-change estimates

World Glacier Monitoring Service (WGMS) · doi:10.5904/wgms-amce-2026-02-10

The WGMS has collected glacier measurements from scientists around the world for decades, under the United Nations and international science bodies. This dataset is its estimate of how much ice each of the world’s glaciers gained or lost in every year. It is the most complete worldwide record of its kind, which is why the site uses it instead of trying to work out volume from pictures.

Used for: the ice volume lost from 2000 to 2024, and the Olympic pool comparison.

WGMS (2026): Annual mass-change estimates for the world’s glaciers, version 2026-02-10. Method: Dussaillant et al. (2025), Earth System Science Data 17, 1977–2006, doi:10.5194/essd-17-1977-2025.

Glacier outlines

GLIMS (Global Land Ice Measurements from Space) and the Randolph Glacier Inventory (RGI) · glims.org

These are maps of where the world’s glaciers are, drawn by glaciologists from satellite images and checked by hand. They tell the computer which ice belongs to an ice field and which is sea ice, a frozen lake or snow on a neighbouring mountain.

Used for: the boundary inside which ice area is counted (GLIMS outlines), and the list of glaciers whose volume is added up for each site (RGI 6.0).

Background imagery

Esri World Imagery · esri.com

A ready-made picture of the whole Earth, stitched together from many satellite and aerial photos taken in different years.

Used for: the land and sea shown around each site’s pictures. It is scenery only, and its full credit is on every map.

Coastlines

Natural Earth, public domain · naturalearthdata.com

Used for: the world map on the front page.

The twenty ice fields

Ice Legacy

Ice Legacy is the expedition project crossing the world’s twenty largest ice fields. Its list of missions decides which places this site shows.