Half-century satellite record puts polar ice loss at 11.3 trillion tonnes

The Ice Sheet Mass Balance Inter-comparison Exercise team combined 42 independent estimates of Greenland and Antarctic ice-sheet mass change derived from satellite measurements of ice flow, elevation and gravity. The record draws on 27 satellite missions, reaches back to 1972 for Greenland and 1979 for Antarctica, and extends through 2023. The paper was peer-reviewed and published in Scientific Data on 16 September 2026.

For the common 1979–2023 period, the combined estimate is a loss of 11,309 ± 565 billion tonnes of ice. The authors calculate that 84% of that loss came from glacier dynamical imbalance, principally ice being discharged to the ocean faster than it was replaced, while changes in surface mass balance accounted for the remaining 16%. The study estimates the lost ice added 31.4 millimetres to global mean sea level.

The main contribution is the length and consolidation of the observational record rather than a new physical mechanism. Earlier IMBIE assessments focused on shorter periods; the new work integrates historical Landsat measurements with later altimetry and gravimetry to reconstruct nearly five decades of change and quantify uncertainty across the different methods.

The record shows losses accelerating from the 1990s into the 2010s. It also shows a temporary slowdown from 2020 to 2023, associated with unusually high snowfall over East Antarctica and milder Greenland summers. The authors treat that recent slowdown as short-term variability within the longer loss trend rather than evidence of a reversal.

The assessment is a synthesis of observational estimates and regional climate models, so it depends on corrections for processes such as glacial isostatic adjustment, firn compaction and periods with sparse early satellite coverage. The authors explicitly increase uncertainty where measurements are poor or require interpolation. The dataset also excludes changes such as ice-shelf thinning or retreat when they do not directly change grounded ice mass contributing to sea level.

That makes the dataset most useful as a constraint on sea-level models and as a common observational baseline. The paper does not itself forecast future sea level. Its value is in reducing disagreement about the historical mass-balance record and in showing that dynamic glacier discharge, which is strongly influenced by ocean conditions, dominates the combined loss signal over the period studied.