Climate sensitivity

• 1. Introduction

  ›

  Climate sensitivity is the number climate scientists argue about most, and its value could determine whether the Paris Agreement survives or collapses. It answers one precise question: if the concentration of carbon dioxide in the atmosphere doubles, how much warmer does Earth become? That sounds simple. It is not. The answer depends on a chain of secondary effects called feedbacks, and those feedbacks are exactly what scientists struggle to pin down. Whether the number lands closer to 2 degrees Celsius or 4 degrees Celsius is not an academic distinction. One study suggests that halving the uncertainty in just the short-term version of this measure could save trillions of dollars in avoided economic damage. Another study found that if the long-term value exceeds 3.4 degrees Celsius, emissions cannot be reduced fast enough to keep warming below 2 degrees at all. The first person to put a number on this question was Svante Arrhenius in the 19th century. Getting from his estimate to the modern one required more than a century of argument, observation, and computing power that Arrhenius could not have imagined.

• 2. The Feedback Problem

  ›

  When carbon dioxide traps heat, the direct warming it produces is, in principle, easy to calculate using the Stefan-Boltzmann law. A doubling of CO2 from the pre-industrial level of 280 parts per million delivers a radiative forcing of about 3.7 watts per square meter, and without any feedbacks that would produce roughly 1 degree Celsius of warming. Scientists agree on that figure; it is not contested. The uncertainty comes from what happens next. Warming melts ice, and ice reflects sunlight. Less ice means more sunlight absorbed by the darker ocean and land beneath, which means more warming still. This is the ice-albedo feedback. Warming also increases evaporation, which puts more water vapour into the atmosphere, and water vapour is itself a greenhouse gas, adding another layer of warming on top. These are self-reinforcing feedbacks. They push the final temperature higher. Balancing feedbacks work in the opposite direction, increasing the rate at which a warmer planet radiates heat to space. Climate sensitivity depends on which set of feedbacks wins, and by how much. The uncertainty in climate sensitivity estimates traces almost entirely to these feedback interactions, especially the behaviour of clouds.

• 3. Two Measures, One Question

  ›

  Transient climate response, or TCR, measures temperature change averaged over a 20-year window centered on the moment when CO2 doubles, assuming a 1 percent annual increase in concentration. It is the near-term reading. The Intergovernmental Panel on Climate Change estimates TCR likely falls between 1 and 2.5 degrees Celsius. Equilibrium climate sensitivity, or ECS, is something different. It measures where temperature eventually settles after the planet has fully adjusted to doubled CO2, including feedbacks that take centuries or millennia to play out. The deep ocean, for instance, takes many centuries to reach a new steady state after warming begins, and while it is still absorbing heat it acts as a brake on surface temperatures. Once the ocean equilibrates, that brake lifts, and temperatures climb further. That is why ECS is always higher than TCR. The IPCC Sixth Assessment Report placed ECS, with high confidence, in the range of 2.5 to 4 degrees Celsius, with a best estimate of 3 degrees. A further extension of the concept, called Earth system sensitivity, layers in feedbacks that operate over millennia, such as the melting of continental ice sheets and changes in vegetation, and may be roughly twice as large as ECS. The carbon cycle, however, is deliberately excluded from the Earth system sensitivity definition.

• 4. Reading the Past

  ›

  Paleoclimatologists can check modern estimates against the deep record of Earth's history. The Last Glacial Maximum, about 21,000 years ago, provides especially useful data because the atmospheric CO2 concentration and the radiative forcing from that cold period are relatively well documented. The Paleocene-Eocene Thermal Maximum, about 55.5 million years ago, is another reference point. During that roughly 20,000-year episode, massive amounts of carbon entered the atmosphere and average global temperatures rose by approximately 6 degrees Celsius. A 2007 estimate using data spanning the most recent 420 million years produced a sensitivity consistent with current climate models. Studies of the last 800,000 years found that climate sensitivity was greater during glacial periods than during interglacial periods. Older periods are harder to reconstruct; the oldest continuous ice core, from which past CO2 concentrations can be read directly, is less than one million years old. Climates further back therefore carry more uncertainty. One well-studied warm period, the Pliocene, ran from 5.3 to 2.6 million years ago, and the cooler Pleistocene from 2.6 million to 11,700 years ago; researchers have examined both as possible windows into how the climate behaves under conditions that differ from today.

• 5. Arrhenius to the Charney Report

  ›

  Svante Arrhenius made the first quantitative estimate of warming from doubled CO2. His initial paper placed the figure at 5 to 6 degrees Celsius. He later revised it downward to 4 degrees. To account for the role of water vapour, Arrhenius assumed that relative humidity would stay constant as the planet warmed, an assumption that has held up. The first calculation to use detailed absorption spectra and a computer for numerical integration came from Syukuro Manabe and Richard Wetherald in 1967. They computed an equilibrium climate sensitivity of 2.3 degrees Celsius per doubling of CO2, then rounded it to 2 degrees in their abstract. Their work has been described as, quote, arguably the greatest climate-science paper of all time. In 1979, a committee convened by the United States National Academy of Sciences and chaired by Jule Charney placed equilibrium climate sensitivity at 3 degrees Celsius, plus or minus 1.5 degrees. As Manabe recounted in 2004, Charney took the Manabe estimate of 2 degrees and the Hansen estimate of 4 degrees, added 0.5 degrees to one end and subtracted 0.5 from the other, and arrived at the 1.5 to 4.5 degree range that then appeared in every major greenhouse assessment that followed. Stefan Rahmstorf noted in 2008 that when the Charney report was published, that range rested on very shaky ground.

• 6. Decades of IPCC Assessment

  ›

  The IPCC First Assessment Report in 1990 placed equilibrium climate sensitivity between 1.5 and 4.5 degrees Celsius, with a best guess of 2.5 degrees. Its ocean models were simplified. The 1992 supplementary report used full ocean circulation models and found no compelling reason to change the 1990 estimate. The Second Assessment Report said the same. The Third Assessment Report in 2001 retained the identical likely range. Much of the persistent uncertainty across those reports was attributed to insufficient understanding of cloud processes. The Fourth Assessment Report in 2007 broke the pattern, stating that confidence had increased substantially. That report concluded ECS is very likely above 1.5 degrees and likely between 2 and 4.5 degrees, with a most likely value of about 3 degrees. The Fifth Assessment Report in 2013 reverted to the older 1.5 to 4.5 range because some estimates using Industrial Age data came in low; it also stated ECS is extremely unlikely to be below 1 degree and very unlikely to exceed 6 degrees. In preparation for the Sixth Assessment Report in 2021, a new generation of models spanning 27 research groups produced ECS estimates ranging from 1.8 to 5.6 degrees, with 10 of the 27 models exceeding 4.5 degrees. The median ECS estimate shifted from 3.2 to 3.7 degrees. The main driver of that increase was improved cloud modelling: warmer temperatures, the new models suggested, cause sharper reductions in low cloud cover, allowing more sunlight to reach the surface.

• 7. The Hot Model Problem

  ›

  A fifth of the models contributing to the Sixth Assessment Report ran notably hotter than the rest. These so-called hot models projected that worst-case warming could exceed 5 degrees Celsius above pre-industrial levels by 2100, a level characterised as having a catastrophic impact on human society. The problem is that those same models reproduced known historical climate poorly. They struggled to match observed 20th-century warming and failed to replicate the cooling pattern of the last ice age. Empirical observations combined with physics-based modelling place the very likely range for warming at 2.3 to 4.7 degrees, well below what the hot models predicted. For those reasons, the IPCC gave the hot models reduced weight in its 2022 analysis. The gap between the hot models and observation-constrained estimates points back to the same unresolved difficulty: clouds. Improved cloud physics drove the estimates upward, and deficiencies in cloud simulation may then have pushed some models too far. The Coupled Model Intercomparison Project, running since the 1990s, exists partly to pool these models and identify where they diverge, and the persistent spread around the cloud feedback question is why the 3 degree best estimate from the Sixth Assessment Report still carries an uncertainty band of more than a full degree in either direction.