Is Record Breaking Heat the New Normal?
A look at whether we’re really setting more temperature records than in the past.
In the past two months we’ve seen polar conditions across a large part of the United States, a relentless series of storms causing huge floods in the UK, and an unprecedented heatwave striking Australia. Over the past few years, we’ve heard of more and more records being broken, for temperatures, for wind speeds, for rain and snow. Extreme weather events have become a normal part of our daily news.
Is it the case then, that new records are being set at an accelerating pace? Were records really broken less often in the past, or are we simply seeing more coverage, now that more measurements are available, and there’s a ready-packaged narrative of climate change to fit them into? I decided to have a look at some historical data, to see which is the case.
The Global Historical Climatology Network is a database of observations from approximately 90,000 weather stations around the world, some dating back to the 18th century. Each station has daily records of variables like temperature extremes, precipitation, wind speed, etc. The full dataset tips the scales at about 22 gigabytes, or about 260 million individual measurements.
Of course, with a dataset of this size, there are trade-offs. There are plenty of gaps in the historical coverage, and the variables measured vary a lot from station to station. There’s also a strong geographical bias, particularly in the earlier data, towards Europe and North America. It’s only really in the second half of the twentieth century that the coverage becomes truly “global”. That said, so long as we’re aware of its shortcomings, it’s good enough for our purposes.
For simplicity, and because it’s one of the most widely available variables, I’ve only looked at record high temperatures. It’s easy to set a record when there are few measurements, so I’ve thrown away any records set in the first ten years of any station’s operation. This stops us from seeing a rapid sequence of records set when a new station is introduced.
The number of stations has been increasing with time, so what we’re really interested in isn’t the number of records set, but the number of records relative to the number of stations in operation. This lets us compare numbers from now to the 1950s, even though the number of stations has more than doubled.
I’m also only focusing on the period from 1950 to present. In the pre-war period, the geographical bias means that fairly localised events, like the 1911 heatwave in the north-eastern US, set records at a disproportionate number of stations. 1950 is a fairly arbitrary point to set a boundary, but it roughly coincides with the global expansion of the weather station network.
In general, each station reports a new record high once every twenty to twenty-five years, meaning that in any given year, four or five percent of stations set a record. That number varies a bit, largely because the temperatures at nearby stations are correlated—if one station reports a record, it’s fairly likely that its neighbours will as well.
As for a trend, there’s certainly none visible to the naked eye. A linear regression suggests that there may be a slight increasing trend, but if there is, it’s incredibly small (about two extra records per year, on a baseline of about four hundred), and statistically insignificant even by quite relaxed standards.
So if there’s no increase in the rate at which records are being set, does this mean the climate isn’t changing as much as we’re being told? Unfortunately not. If the climate was holding steady, we’d expect to see a steady drop-off in the rate at which record temperatures were being achieved. A longer series of measurements means fewer surprises, and as time went on, the chances of seeing something new should decrease. The only reason we see new records on a regular basis is that typical temperatures are increasing, and so are the extremes.
In a warming climate, we will see new extremes of weather, but the increase will be slow. Records will continue to be broken, but probably not at any noticeably accelerated pace. Instead, gradually rising temperatures will make what would have been a record-breaking heatwave in the mid-twentieth century completely unremarkable by the mid-twenty-first. This gradual shift will mean that new temperature records are going to be a fact of life for a long time to come.
