Beatin’ the heat: cut carbon or we’re cooked

By Gareth Renowden 23/08/2013

The area of land affected by extreme heatwaves is expected to double by 2020 and quadruple by 2040, and there’s no way we can stop it happening according to a new paper by Dim Coumou and Alexander Robinson – Historic and future increase in the global land area affected by monthly heat extremes (Environmental Research Letters, open access). However, the researchers find that action to cut emissions can prevent further dramatic increases in heat extremes out to the end of the century.

The paper’s made headlines around the world — see The Guardian, Independent, and Climate Central — most focussing on the inevitability of more, and more intense, heat events in the near future. Dana Nuccitelli at The Guardian provides an excellent discussion of the science behind the new paper so, to avoid reinventing the wheel, I’m going to focus on a fascinating chart from the paper, and then ponder the implications for climate policy.

Coumou and Robinson define heat events in terms of their departure from the statistical distribution of all temperatures for any given part of the earth’s surface. If, like me, it’s a long time since that stats course at uni, you might need reminding that for things — like temperatures at any given place — that are “normally distributed”, then the frequency of given events is described by a bell curve. Here’s Wikipedia’s explanation, and graph:

Standard deviation diagram

The standard deviation — given the greek letter sigma — of a normal distribution is a measure of the variation, or spread, of events — how often they are likely to happen. By definition, 68% of events will fall in one standard deviation, 95% within 2-sigma, and 99.8% within 3-sigma. For temperatures, high temperature events fall under the right hand end of the curve, and cold under the left. Coumou and Robinson worked out how often 3-sigma heat events occurred during a period when the climate was relatively stable — from 1950 to 1980 — and then worked out how it has changed since then, and how an ensemble of models suggest it will change in the future. They found that the frequency and area of earth’s surface covered by 3-sigma events increased significantly over the last 30 years1, and then how they are likely to change in the future under two emissions scenarios — one low, one “business as usual”.

The effect of warming is to move the whole bell curve towards the right, so that very rare events — greater than 3-sigma — become much more common. Here’s what the paper says:

The projections show that in the near-term such heat extremes become much more common, irrespective of the emission scenario. By 2020, the global land area experiencing temperatures of 3-sigma or more will have doubled (covering ~10%) and by 2040 quadrupled (covering ~20%). Over the same period, more-extreme events will emerge: 5-sigma events, which are now essentially absent, will cover a small but significant fraction (~3%) of the global land surface by 2040. These near-term projections are practically independent of emission scenario. [my emphasis]

5-sigma events are very rare — “essentially absent” — in an unchanging climate, but as heat accumulating in the system stacks the dice (pace Hansen), they begin to happen – infrequently at first, but then more and more often as warming progresses. In a high emissions scenario:

By 2100, 3-sigma heat covers about 85% and 5-sigma heat about 60% of the global land area. The occurrence-probability of months warmer than 5-sigma reaches up to 100% in some tropical regions. Over extended areas in the extra-tropics (Mediterranean, Middle East, parts of western Europe, central Asia and the US) most (>70%) summer months will be beyond 3-sigma, and 5-sigma events will be common.

Figure 4 from the paper summarises the findings very nicely:


Land area is plotted on the vertical axis, and mean summer temperature across the bottom. Global temperature anomaly corresponding to those summer temps is given across the top. The black circles/triangles/crosses show the measured increase in land surface area for 1, 2 and 3-sigma events, and the coloured versions the modelled projections. We’re already seeing an increase in 3-sigma events, and it won’t be long before 5-sigma events start to show up.

So far, so technical. What does this mean in practical, policy-relevant terms?

Heatwaves are going to happen more often, they are going to become more widespread, and they are going to become more intense — unimaginably hot, for those of us who grew up in the 1950s to 1980s, the reference period for this work.

We’re already seeing this happen, and we can expect it to get much worse over the next few decades. In other words, this is not a hypothetical problem for another generation to deal with.

It is going to be very important to plan at a personal, local, regional and national level to be able to cope with extreme heat when it happens. This will be particularly important in parts of the world where extreme summer heat is not currently common. Increasing heat will worsen droughts and increase pressures on water supplies.

Developing resilience to heat extremes is something that needs to be done now, because what happens over the next 30 years is already baked in2. The climate commitment is going to bite hard, before we can do anything to prevent things getting even worse.

There is good news, however, and that is that if the global community manages to cut emissions steeply — the scenario modelled by Coumou and Alexander (RCP 2.6) effectively means transitioning to zero emissions by 2070 — then we can avoid the worst impacts.

These are the classic dimensions of planning to deal with climate change: we have to cut emissions (mitigate) and prepare for the worst (adapt). One without the other is policy madness. We can also see that the longer we delay both approaches, the costlier it will be to deal with the results.

[Dan Hicks scares himself]

  1. Confirming Jim Hansen’s findings from last year, although using different statistical techniques (see RealClimate for more discussion).
  2. Ahem. Sorry.