The answer to this depends a lot on something scientists call the “climate sensitivity” – a measure of how much the climate warms in response to greenhouse gases.
Until now, scientists have been grappling with how to reconcile the fact that different ways to estimate the climate sensitivity have, so far, come up with quite different answers.
This uncertainty has never been a reason to question whether climate change will be serious or to delay action to tackle emissions, though it has often be misused by climate skeptics this way.
But a question mark over the value of climate sensitivity has meant that projections of future temperature rise are less precise than scientists would like. It also makes it harder to gauge our chances of staying below a given temperature limit, such as 2C above pre-industrial levels.
A new paper published in Nature Climate Change says there is, in fact, no disagreement between the different methods after all. In reality, they measure different things and once you correct for the fact that the historical temperature record underestimates past warming, the gap closes.
The widest range of climate sensitivity supported by recent observations is 1.0-4.0C, with a best estimate at around 2.0C. At face value, this suggests that models, with a range of 2.0–5.6C, are altogether too sensitive.
A sensitive question
One way is to compare observations of how much the land, ocean, ice and atmosphere have warmed over the industrial period to how greenhouse gases and other factors that influence temperature, known as forcings, have changed in the same time.
But estimates of climate sensitivity from these so-called “energy budget models” tend to be lower than those simulated by climate models, raising the question of whether models are overestimating the scale of the climate response to greenhouse gases.
Climate models work differently, by calculating the warming expected from greenhouse gases and all processes in the climate system known to amplify or dampen the speed of warming, known as feedbacks.
In a News and Views article accompanying the new research, Prof Kyle Armour from the University of Washington, says: “The widest range of climate sensitivity supported by recent observations is 1.0-4.0C, with a best estimate at around 2.0C. At face value, this suggests that models, with a range of 2.0–5.6C, are altogether too sensitive.”
“Apples to oranges”
The new study argues that issues with the way global temperatures have been measured historically mean that they underestimate the total warming over the industrial period.
First, there are places in the world where temperature data doesn’t exist. Importantly, the most poorly sampled areas are also the fastest warming. In the HadCRUT4 dataset, it is well known that gaps exist over remote parts of the Arctic, for example.
By reducing the geographical coverage in the models to match the observational record – a process known as “masking” – the scientists worked out that imperfect geographical coverage effectively pushed observation-based estimates of warming down by about 15 per cent.
Secondly, the techniques scientists use for making temperature measurements have advanced over time. Hawkins says: “Historical weather observations include air temperatures measured over land areas and sea surface temperatures in ocean regions, but only where there were existing thermometers (& ships).”
But “blending” these different – and often sparse – measurements to calculate a global average temperature has unwittingly caused problems, Armour writes: “[S]hip-based measurements are actually of the ocean’s surface layer, which has been warming at a slightly slower rate than the air just above.”
Since climate models simulate only air temperatures, this is not a like-for-like comparison.
Adapting the models to account for the difference between seawater and near-surface air temperatures in the models allowed the scientists to see what difference the instrumental differences make. Hawkins explains: “As we cannot travel back in time to take additional historical observations, the only way to perform a like-with-like comparison is by treating the models differently.”
Doing so, suggested that the instrumental changes within the HadCRUT4 dataset underestimated global temperature change over the industrial period by a further 9 per cent, to a total of 24 per cent when combined with the effect of incomplete observations.
The difference this is not trivial, says Hawkins: “According to the CMIP5 simulations, more than 0.2°C of global air temperature change has been ‘hidden’ due to our incomplete observations and use of historical sea surface temperatures.”
Based on their calculations, the scientists’ best estimate for how much the observational record underestimates transient climate sensitivity is 1.7C, with a range of 1.0-3.3C.
Scaling up previous estimates based on the imperfect observational record brings them in line with the best estimate from models of 1.8C, with a range of 1.2-2.4C. Or Armour says: “Climate sensitivity estimates have now been reconciled.”
Prof Piers Forster from the University of Leeds is cautious about overinterpreting the results too soon. He tells Carbon Brief: “I think it just emphasises that [climate sensitivity] estimates based on historical observations remain quite large. There may not be a discrepancy between the various types of estimates. Or we may find physical reasons why they differ, as other papers suggest. In my mind, the jury is still out on this one.”
In other words, questions about climate sensitivity are complicated and won’t be solved by any single bit of research.
The impact that gaps in the temperature record have on climate sensitivity is likely to diminish in future as observational coverage improves, the study notes.
But the problem will still remain for the historical period unless scientists can rescue additional, currently undigitised, weather observations, says Hawkins.
The study raises another interesting point: which global mean temperature is relevant for informing policy? The imperfect record based on observations, or the model record with gaps filled in? On this point, Hawkins says: “If it is decided that climate targets refer to the latter, then the warming is actually 24 per cent (9-40 per cent) larger than reported by HadCRUT4. And that is a big difference, especially when considering lower global temperature targets.”