Antarctica gained 695 billion tonnes of ice
It was the largest gain in the satellite-gravity record. It was caused by warming. And it has already stopped.
Between July 2021 and April 2023 the Antarctic ice sheet put on about 695 billion tonnes of mass — the largest 22-month increase in the entire satellite-gravity record, and almost all of it in East Antarctica. A study published in Nature on 19 August 2026, led by researchers at the Institute of Oceanology of the Chinese Academy of Sciences, set out to explain how that happened during a period of record global heat.
The answer arrived from about eight thousand kilometres away.
A wave that starts in the tropics
The team combined gravity-satellite observations with ice-core accumulation records, water-vapour tracking and atmospheric circulation simulations. What they found was a chain: sustained warming of the tropical warm pool — the reservoir of very warm surface water in the western Pacific and eastern Indian Ocean — changed where deep convection happened. That convection launched a Rossby wave train, a standing pattern of atmospheric ridges and troughs that propagates poleward. Over East Antarctica the wave set up a persistent north–south dipole in the circulation, and the dipole reorganised moisture transport onto the continent. The result was exceptional snowfall, sustained for the better part of two years.
This is not a mysterious reprieve. It is a well-posed piece of atmospheric dynamics in which the driver is ocean warming.
What the number does and does not mean
Over the two decades to 2023 the ice sheet lost mass at an average of roughly 140 billion tonnes a year. The 2021–23 episode handed back about five years of that — briefly, and in the wrong currency. Snow accumulating on the high, cold East Antarctic plateau is a surface deposit that responds to weather. The losses that dominate the long-term trend are dynamic: warm water reaching the undersides of West Antarctic ice shelves, thinning them, releasing the brake on the glaciers behind. Those grounding lines retreat; they do not advance again because it snowed inland.
There is a second reason to keep the sign straight. A warmer atmosphere holds more water vapour, and for a long time models have projected increased Antarctic snowfall as a partial offset to ice loss this century. The 2021–23 event is a vivid, short-lived instance of exactly that mechanism — which makes it evidence for the projection, not against the trend the projection is offsetting.
The episode has already ended. What it leaves behind is a useful measurement: an estimate of how much tropical ocean heat can move into polar snowfall, and how quickly. That number goes back into the models, where it will mostly be used to make the error bars on sea-level projections narrower rather than smaller.