What have been the most extreme heatwaves in history?

In June 2021, a severe heatwave hit western North America, setting a new all-time temperature record for Canada and causing California to declare a state of emergency.

The Pacific north-west heatwave was unprecedented for the region, with local temperature records broken by as much as 4.6C. The intense heat caused hundreds of deaths, damaged railways and roads, wilted crops, impaired businesses and led to a series of wildfires. It has been estimated that the heatwave caused around $9bn in damages in the US.

But how extreme was this heatwave compared with other historical ones around the world?

Record-shattering extremes are a threat to society because communities have often adapted to a known range of temperatures. The larger the temperature shock, the more it can stress the infrastructure of a community – sometimes to breaking point.

In our new study, published in Science Advances, we show that the Pacific north-west heatwave was one of the most extreme heatwaves ever observed anywhere on the planet. However, we also find a handful of other regions that have experienced even more severe heatwaves, measured relative to their local climate.

We use daily maximum temperature to define heatwaves – the indicator used by meteorological organisations such as the UK Met Office. To compare heatwaves in different regions of the globe, absolute temperature is not ideal as a heatwave in the tropics will always be hotter than one in the Arctic, despite communities having very different adaptive capacities.

Looking at the magnitude of a heatwave compared to the local average is a step forward. However, in regions where the conditions fluctuate substantially from season to season and year to year – such as in the mid-latitudes – greater extremes are to be expected.

In a region with lower natural variability, such as the tropics, a heatwave of lesser absolute magnitude may have a bigger impact because the area may be less prepared for such unusual conditions. A region of higher variability is likely to experience more severe heatwaves, in terms of absolute magnitude, but might be able to adapt more easily due to past experiences.

For example, the European heatwave in the summer of 2003 caused at least 6,000 excess deaths in France. A heatwave of similar magnitude in 2006 caused 60% fewer excess deaths, suggesting the population became less vulnerable to heat, likely due to increased awareness of the risk and an improved warning system.

We assess heatwaves in terms of “standard deviation” – a measure of how far an event strays from the average conditions for the region. We use the ERA5 reanalysis dataset – which combines observed data with model simulations – produced by the European Centre for Medium-range Weather Forecasting (ECMWF).

Over the period we are investigating – 1950 to present – climate change has seen steady warming, with each of the last three decades be successively warmer than any preceding decade since 1850. To allow extremes earlier in the record to not be swamped by hotter events more recently, we remove the climate change trend by assessing against the average conditions of the preceding decade. This gives us a metric indicating how extreme a heatwave was relative to the climate of the time.

Using our extreme metric, the Pacific north-west heatwave is defined as 4.1 standard deviations from the modern-day average climate.

In records from 1950 to present, the June 2021 event is the most extreme heatwaves for western North America – but there are several events elsewhere that were more extreme.

The map below shows the magnitude of the greatest heatwave in each part of the world since 1950. The orange shading indicates how many standard deviations each event was from its average climate, while white shading indicates areas of disagreement between different datasets, so no results are included.

It is also worth noting that reanalysis uses models to fill in gaps in observed data. This helps overcome long-standing issues around lack of data on weather extremes in parts of the world that do not have comprehensive weather records. However, it may be that we have missed some large events in, say, parts of Africa where the observed data coverage is particularly poor.

Looking globally – and in the absence of climate change – we find there have been four heatwaves more extreme. We verify the events by assessing a second reanalysis dataset and checking against observed data from weather stations. The events we find include some that are surprising – there appears to have been little reported on the heatwave in southern Brazil in November 1985, for instance.

Overall, the most extreme heatwave we detect, ever, struck south-east Asia in April 1998. This event is linked to a strong El Niño – similar to the event of 2016 – which also led to some regional temperature records. The 1998 event caused conditions perfect for the spread of wildfires – an area of forest the size of 7 million football pitches was destroyed.

The table below shows the eight most severe events that we identified. The June 2021 heatwave in western North America is in fifth place, while some infamous events don’t make the cut.

For example, the 2003 European heatwave is not listed, as it was “only” 3.3 standard deviations from the average. The notability of this heatwave came from the overall societal impact of the event – not solely the temperature magnitude, which is the next step of the puzzle to solve. The Siberian heatwave of summer 2010 also does not feature, as it was above average for several months, rather than very extreme heatwaves for a few days.

As well as highlighting the regions with the greatest daily extreme temperatures, we can identify which regions have particularly low historical temperature extremes. By chance some regions have not experienced large extremes over the past 70 years. Therefore, with no local trigger for adaptation, they may not be prepared for heatwaves. Parts of India, Australia, and central Africa show no historical events above three standard deviations from the average.

In a warming climate, we know that the average daily temperature has increased over the past 70 years. In our study we investigate if there is evidence of extremes increasing faster than this climatological shift.

We do this by assessing extremes in terms of standard deviation from the average and then removing the underlying “signal” of warming caused by climate change. The results indicate that extremes are increasing in magnitude and frequency, but at the same rate as the average climate. That is, at least when viewed in a global context. (​​There are also other ways in which future heatwaves could change – for example, they could increase in duration or occur in more regions simultaneously – but this was outside the scope of our study.)

Running large collections – or “ensembles” – of climate model simulations are very useful for investigating extreme events. By their nature, few of these events occur in the real world, but using model ensembles many times more data is available.

Using two different models with large ensembles, we also show that extreme events are projected to increase in magnitude at the same rate as the average climate warms.

It is important to remember that the meteorology of a heatwave does not determine its impact. Many socioeconomic factors will influence the disruption a heatwave causes, including population density, housing conditions and early warning systems.

The impacts of heat are wide-reaching, affecting human health, animal health, ecosystems, food security and many other sectors. In many cases, a measure of heat stress, or the “felt” temperature, may be a better indicator of its impact. Viewing the problem in the context of this hazard highlights that the Pacific north-west heatwave was exceptional.

The June 2021 heatwave saw records broken by a staggering 4.6C, a month before the climatologically hottest part of the year. It is hard to prepare for such events, so far beyond the normal for the region.

After a record-breaking event, greater preparedness for future extremes is likely. As discussed previously, in Europe the 2006 heatwave caused fewer deaths than the 2003 event, thanks to preventive measures set up by health authorities and institutes.

With record-shattering heatwaves likely to become more frequent as the climate warms, through education and better adaptation tools – such as improved early warning systems – we can greatly reduce the societal and ecosystem impacts of these future extreme events.