The Physicists Who Proved Climate Change Before It Was Political

In 1967, a researcher at a US government lab published a short paper with a number in it. Doubling the carbon dioxide in the atmosphere, it said, would raise the average surface temperature by about 2 degrees Celsius.

The computer that produced that number was slower than the chip in a modern car remote. The paper ran to a few pages. Almost nobody outside a small circle of atmospheric scientists noticed it at the time.

Nearly sixty years later, that early estimate has held up better than most predictions in any field. The link between carbon dioxide and a warming planet was settled physics long before it became a political argument.

The Greenhouse Idea Was Already Old

The basic mechanism is not subtle, and it is not new. Some gases in the air let sunlight pass through to warm the ground, then absorb part of the heat the ground radiates back up and send some of it downward again. More of those gases, more heat kept near the surface.

A Swedish chemist named Svante Arrhenius did the arithmetic by hand in 1896 and concluded that more carbon dioxide would warm the planet measurably. He had the direction right. He was wrong about the timescale, guessing it would take thousands of years rather than a couple of centuries, because he could not foresee how fast fossil fuel use would climb.

The physics sat there, broadly accepted and largely ignored, for another seventy years. Knowing that carbon dioxide traps heat is not the same as knowing how much the actual planet, with its oceans and clouds and winds, would respond.

What Manabe Actually Did

Syukuro Manabe, a Japanese scientist who had moved to the United States, took the problem apart into something a 1960s computer could handle. He modeled a single vertical column of atmosphere and tracked how energy moved up and down through it.

His key move was to let water respond to temperature. Warm air holds more water vapor, and water vapor is itself a heat-trapping gas, so a small nudge from carbon dioxide gets amplified as the air moistens. Manabe's model captured that feedback, and it pushed the warming up by more than half.

The result, published with his colleague Richard Wetherald, was the roughly 2-degree figure for doubled carbon dioxide. It was the first time anyone had a physically honest estimate rather than a guess.

Scientists build models like this for the same reason engineers build wind-tunnel replicas: you cannot run an experiment on the only planet you have, so you build a smaller version whose rules you trust and see what it does.

Did You Know?

In 1979, a US National Academy of Sciences panel led by the meteorologist Jule Charney reviewed the young field and concluded that doubling carbon dioxide would likely warm the planet somewhere between 1.5 and 4.5 degrees Celsius. More than forty years of satellites, ocean sensors, and vastly better models have barely moved that range. The 1979 answer is still, roughly, the answer.

The Second Problem: Proving It Was Us

A prediction of future warming is one thing. Once the warming actually started, a harder question appeared: how do you show that a particular hot decade is caused by carbon dioxide and not just the climate's normal noise?

This is where Klaus Hasselmann came in. In the 1970s he showed that the slow climate system effectively integrates random daily weather into longer swings, the way a large lake barely ripples in a gust but slowly rises with the season. That meant natural variability had a predictable statistical shape, and a human signal could in principle be separated from it.

He then built the method for doing it, often called fingerprinting. Different causes of warming leave different patterns. Greenhouse warming heats the lower atmosphere while cooling the layer above it, warms nights faster than days, and warms winters faster than summers. The sun getting brighter would warm every layer at once, days and nights alike. Volcanic dust would cool things sharply for a year or two, then fade.

The technique works like handwriting analysis. A rising thermometer alone tells you little, the same way a signature tells you little if you have never seen the person's hand before. But hold the observed pattern of changes, across altitude, season, and time of day, against the set of expected patterns, and only one matches. The measured fingerprint of the last century lines up with carbon dioxide and with nothing else on the list.

Hasselmann also showed why early attempts had failed. Detecting a weak signal buried in a noisy record is a signal-processing problem, and he adapted tools from that field to weight the parts of the climate record where the human signal stands out most and downplay the parts dominated by natural swings.

Attribution science is why official assessments moved over the years from cautious phrasing to flat statements that human activity is the dominant cause of modern warming.

Manabe's employer for much of this work was the Geophysical Fluid Dynamics Laboratory, part of the US weather agency. The same modeling lineage that produced his climate projections also underpins modern hurricane forecasts and the multi-day weather predictions most people now take for granted.

What the Old Models Got Right

The predictions are now old enough to grade. Early climate models, run decades ago, called several things correctly: the overall rate of surface warming, the cooling of the stratosphere, the fact that the Arctic would warm faster than the tropics, and that nights would warm faster than days.

They were not perfect. Regional details, the behavior of specific cloud types, and the pace of ice loss have all needed correction, and some changes, like the speed of recent ocean heating, have run at the high end of what was expected.

The consequences those models implied are now visible without instruments. Glaciers that fed rivers for centuries are retreating, and high mountain lakes swollen with meltwater have become a standing hazard for the valleys below them. Shifts in where and how hard it rains are straining water supplies in the regions that were already closest to the edge.

Knowlegic Perspective

It is easy to assume the science of climate change is recent, because the public argument about it is recent. The two are decades apart. The core result was in hand in the 1960s, checked and rechecked through the 1970s, and treated as established by the people who studied it long before it reached a ballot or a boardroom.

That gap between scientific confidence and public acceptance is not unique to climate. It shows up wherever a finding carries an uncomfortable cost. The work of Manabe and Hasselmann is a reminder that "we did not know" is rarely the honest description. More often, we knew, and the knowing simply had not yet become inconvenient enough to fight about.

Their models were built with the humility of people who understood they were making a rough sketch of an enormous system. The sketch turned out to be accurate in the ways that mattered most.

The link between carbon dioxide and a warming planet was settled physics long before it became a political argument, and the people who settled it did so with tools that would now fit in a corner of a phone.

Sources & References

• The Nobel Prize in Physics 2021: Popular Science Background, The Royal Swedish Academy of Sciences (2021)
• Prof John Mitchell: How a 1967 study greatly influenced climate change science, Carbon Brief (2017)
• Explainer: How scientists estimate climate sensitivity, Carbon Brief (2018)
• Syukuro Manabe makes the first modern climate model, American Physical Society (2025)
• 5 forecasts early climate models got right, The Conversation (2025)
• The Discovery of Global Warming: The Carbon Dioxide Greenhouse Effect, Spencer Weart, American Institute of Physics (2023)