Can we zero-in on Earth’s sensitivity to CO₂?

If it were easy to pin down the exact value for our planet’s sensitivity to greenhouse gas emission, it would have been done a long time ago—and you wouldn’t be reading yet another news story about it. It’s not like we have no idea how sensitive the climate is. The range of possible values that scientists have been able to narrow it down to only spans from “climate change is very bad news” to “climate change is extremely bad news.”

But the difference between “very bad” and “extremely bad” is pretty important, so climate scientists aren’t throwing up their hands any time soon—as two new studies published this week show.

There are several basic strategies available for calculating the climate’s sensitivity. These range from studying climate changes in the distant past to building and evaluating climate models to analyzing the warming over the last century or so. Each strategy has pros and cons. A handful of studies looking at the last century made waves a few years ago for yielding oddly lowball estimates of the impact of CO₂ on warming, for example. Later studies have found problems that push those estimates upward when corrected, but one of this week’s studies demonstrates that the entire strategy is inherently problematic.

Texas A&M’s Andrew Dessler and the Max Planck Institute’s Thorsten Mauritsen and Bjorn Stevens ran a typical analysis—comparing global temperature change to the change in the greenhouse effect—but they didn’t do it with the observed temperature record. Instead, they generated a hundred different simulations of 1850 to 2005 from a single climate model and ran the analysis for each simulation.

To use an analogy, instead of just analyzing last week’s weather, they created 100 weather model simulations where last week’s weather plays out slightly differently.

And that’s where things get interesting. The typical climate sensitivity analysis didn’t turn up the same answer for each of those 100 simulations—even though the true climate sensitivity of that model was the same in every simulation. While the model has a sensitivity of about 2.9 degrees Celsius warming for a doubling of atmospheric CO₂, the estimated sensitivity in the simulations ranged from 2.1 to 3.9 degrees Celsius. That means many of those simulated histories yield misleading answers.

The reason that happens seems to be that our measurements of the balance of incoming solar radiation and outgoing heat energy depends on more than just global temperature change—it depends on regional patterns of temperature change, which can vary. So two simulations with the exact same amount of warming can actually give you different estimates of climate sensitivity.

This becomes a problem because we only have one version of actual history to analyze, so we can’t be sure whether this particular estimate method is giving us a reliable result or whether it’s the equivalent of one of the oddball model runs.

Recalculating route…

The other new study this week, from Peter Cox and Mark Williamson of the University of Exeter and Chris Huntingford of the UK Centre for Ecology and Hydrology, tries a variation on the theme of analyzing the historical temperature record. Instead of comparing greenhouse gas changes to the recorded warming trend, they decided to focus on how variable year-to-year global temperatures have been. Instead of teasing out the signal, they studied the noise.