Satellites measure the sunlight coming in and the energy going back out to space, while thousands of floats across the oceans measure the heat piling up below the surface. The two records are independent of each other, and they agree on how fast the gap has grown.2
Most of that growth is on the sunlight side. Earth is soaking up more of the sunlight that reaches it because it reflects less than it used to, and the Sun’s own output has barely changed.1,3
Some of the change is ice and snow, since shrinking sea ice and snow cover expose darker ocean and land that take in more sunlight. More of it is clouds, with fewer of the bright low clouds that bounce sunlight back to space.1,2 One study finds the world’s belts of stormy cloud narrowing by 1.5 to 3% a decade, and calls that the main reason Earth absorbs more sunlight.4 Much of the cloud change looks like a response to warming, which would mean warming is feeding on itself.1
Cleaner air plays a part too. Sulfur pollution from power plants, factories and ships forms a thin haze and makes clouds brighter, and both reflect sunlight. As China, the US and Europe cut that pollution, the haze over them thinned, even as it grew over India.2,5
Ships are only a small part of the total, but they’re the clearest example. In 2020 a global rule cut the sulfur allowed in ship fuel from 3.5% to 0.5% to clean up air pollution, and researchers had estimated it would reduce ship-related premature deaths by about a third.8 The air along shipping lanes got cleaner, and the bright trails of cloud that satellites used to see behind ships dropped to their lowest level in decades (the pandemic slowdown in shipping helped too).6,7 How much warming that added is still debated, and one study puts it, together with two other short-term factors, at under 0.1 °C for 2023.9
Natural swings matter from year to year, since ocean cycles like El Niño move heat around and change cloud cover, which is why single years jump around. But a rise this steady over two decades is very unlikely to come from natural variation alone.10
As of September 2026, two big questions are still open. One is how much of the cloud change is a response to warming, and how much comes from cleaner air or natural cycles. The other is why climate models fall short of the rise, where one team lists ocean temperature patterns and pollution as suspects, “or something else.”3