Why Are We So Confident Heatwaves Have Been Influenced By Human Activities?

19 Jul 2018

It's been a rough summer, with heatwaves throughout most of the northern hemisphere. Maybe the most striking example is that Arctic temperatures have gone over 90°F, leading to forest fires near the Arctic circle. Along with this, there have been articles in the press linking the heatwaves to climate change, and the heatwaves are even convincing more people of the reality of climate change. I'm normally wary of attributing individual events or even seasons to climate change, but people seem to be pretty confident that climate change is making heatwaves more frequent and more severe. There are also lots of attribution studies linking the increased frequency and severity of heatwaves to global warming, and even claiming that individual heatwaves are the result of human influence (Some examples: Stott et al. (2004); Rahmstorf and Coumou (2011); Fischer and Knutti (2015)).

Another way of seeing our confidence that recent heatwaves are the result of anthropogenic global warming is this map of extreme events from Carbon Brief, which also shows whether studies have linked each event to human influence in some way. The criterion for whether events were influenced by humans is a bit sketchy, but if you just filter for heatwaves you can see that human activities played some role in the vast majority.

So why are we so confident that recent heatwaves have been influenced by human activities? This might seem like a silly question -- it's called global warming -- but if you filter for other weather extremes on the Carbon Brief you generally find that a smaller fraction have been attributed to human influence. Maybe a better way of putting the question is why is our confidence that heatwaves have been affected by humans higher than for other extreme events?

I think there's some subtlety to the answer, with three main reasons. The first two are that (1) we generally have good records of temperature and (2) that climate models do a reasonable job of reproducing these temperature distributions. Long, reliable temperature records means that we can trust statistical analyses, and in particular that we have a good sense of the internal variability, while (2) means that we can have confidence in studies which use climate models to do attribution. Other extreme events like hurricanes and droughts are both much rarer and much harder to simulate: global climate models can just barely resolve hurricanes.

The third reason comes from a point made in a 2015 Nature Geoscience Perspective by Ted Shepherd: increasing CO2 concentrations is a thermodynamic perturbation to the climate system. This means that every aspect of climate change that we are confident in comes from thermodynamics, whereas impacts that come from changes in the circulation of the atmosphere are less robust, because the circulation only feels the perturbation indirectly.

While atmospheric circulation certainly affects temperatures, thermodynamics are still the main control. In contrast, precipitation is more closely tied to the atmospheric circulation, though warmer temperatures do allow the atmosphere to hold more water. Figure 2 of the perspective illustrates this difference by showing that there's much more confidence, in the sense of model agreement, in projected temperature changes than in projected changes in precipitation.

Shepherd is mostly concerned with changes in the mean, but the same ideas apply to extreme events. When we talk about heatwaves we usually mean extended periods of time when temperatures are above some fixed threshold, so if the whole distribution of temperatures shifts, we would expect to cross this threshold more often (though changes in the atmospheric circulation, sea-surface temperature variability and land-use do modulate the frequency and intensity of heatwaves). Extreme precipitation events usually happen in storms, like the winter storms we get in the northeast, which are part of the atmospheric circulation and whose changes under global warming are complex. For instance, the storm-tracks in mid-latitudes -- where most extratropical storms are found -- are expected to shift polewards, so changes in the strength of storms are mixed with changes in their location.

Going back to hurricanes, we expect these to get stronger because a warmer atmosphere can hold more water, but as far as I can tell the jury is still out on how the frequency and distribution (e.g., the number of Category 2 storms versus Category 4 storms) of hurricanes, which are controlled by a complex set of factors, will change. Since we also haven't actually experienced that many strong hurricanes, especially in the satellite era, and we can't simulate them well in global models, it's hard to say whether a particular hurricane has been influenced by climate change.

Droughts are complex phenomena involving interactions between the atmosphere and land surface processes over many years, which are difficult to study in nature and to represent in climate models. The drought record is also pretty sparse, though we do have paleorecords of drought and we know that there were periods of "mega-droughts" in the American southwest during the Medieval period. We seem to be confident that droughts in the American southwest will get worse, but the signal is less clear in the Sahel. Based on the Carbon Brief map there seems to be some disagreement on whether the recent California drought was influenced by climate change.

A simpler problem is actually forest fires, since these just require an few weeks of warm, dry conditions. Human influence was found for both fire events on the Carbon Brief map, and presumably the Arctic fires will also be linked to human-induced climate change.

If instead of increasing CO2 concentrations we were applying some kind of dynamic perturbation to the climate system, then it would be easier for us to link particularly strong storms to human influence. We might also be able to directly influence hurricanes by putting dust in the atmosphere at the right times, and land-use changes could be changing the frequencies of droughts, though my understanding is we can't say anything for sure. But because we're increasing CO2 concentrations, more frequent and more intense heatwaves are one of the clearest signs of a changing climate.