No one checks the weather forecast more obsessively than a marathon runner with a race ahead. We all dream (like OutsideMartin Fritz Huber put it so poetically in a recent column) “that perfect meteorological cocktail – temperatures in the dry 50s, a tailwind that magically follows you like a forest sprite”.
But what are the precise ingredients of this ideal cocktail? There has been a lot of research over the years to try to determine the best temperature, but even the best scientists in the world have not come to a consensus. During Eliud Kipchoge’s first attempt in a marathon of less than two hours in 2017, for example, the starting temperature of 52 degrees Fahrenheit was apparently too high by some calculations, but just according to the scientists organizing the race. And what about the role of humidity, wind and even solar radiation?
A new study in Medicine and science in sport and exercise, from a group led by Andreas Flouris from the University of Thessaly in Greece, attempts to address all of these questions at once by applying machine learning to a massive database of nearly a century of race results . The resulting information offers a surprisingly handy guide to how much you can expect to slow down in a given set of inclement weather conditions.
The researchers collected the results of the major marathon competitions, 10,000 meters, 5,000 meters and 3,000 meters steeplechase, as well as the 50 km and 20 km races. This included the Olympics, World Championships, Diamond League Track Meetings, World Athletics Gold Label Road Races and other such events dating back to 1936. Accurate weather records were collected for each competition. and the results of elite athletes (top three) and well-trained runners (25th, 50th, 100th and 300th places) were compared to the event record at the time of competition. There are obvious limitations to this approach: times will be affected by tactical races and other factors like altitude (eg at the 1968 Mexico Olympics). But on a large data set, it gives you an idea of the impact of the weather in a given year on the weather.
The four main meteorological elements were air temperature, relative humidity, wind speed, and solar radiation (adjusted for cloud cover). These can be considered independently or in composite indices such as the Wet Globe Temperature (WBGT), which is a weighted average that takes into account all four parameters.
There’s a whole bunch of mathematical analysis going into the study, using different approaches (including a machine learning method called a decision tree regressor algorithm) to sift through all the data and look for meaningful patterns. I’m not going to go through everything (the document is free to read online if you want to dig deeper), but there are a few highlights that I want to take away.
The simplest question is how often do athletes run in rain or shine. Using existing World Athletics classifications based on the WBGT, the researchers found that 27% of the races studied were conducted in cool conditions, 47% in neutral conditions, 18% in moderate heat, 7% in high heat. and 1% in extreme heat. We can probably expect the number of competing in unpleasant heat to increase in the years to come, but for now that means you have about a 50-50 chance of getting served that perfect weather cocktail for a race. given.
The machine learning algorithm also offered an estimate of how important each weather parameter is to performance. Unsurprisingly, air temperature was the most important factor, achieving a “feature importance score” of 40%. Then come relative humidity (26%), solar radiation (18%) and wind speed (16%). According to a study I wrote last year, this suggests cloud cover is as important as a lack of wind for running fast. This obviously also depends on other factors: cloud cover is more important in hot weather than in cool weather, while a headwind will slow you down regardless of the temperature.
Regarding the meteorological sweet spot, the general conclusion was that a WBGT between 45.5 and 59 degrees Fahrenheit is best. This is interesting, because World Athletics guidelines consider “neutral” conditions to be a WBGT between 50 and 64.4 degrees. For running fast, the best temperatures seem to be a bit cooler than previously thought. If you venture outside of this area, expect to slow down by about 0.3-0.4% per degree of WBGT. Of course, few of us have easy access to WBGT measurements. If you just look at the air temperature, the sweet spot is between 50 and 63.5 degrees, which is a bit warmer than you expected.
There are other nuances, however, if you look at the individual events. Shorter runs seem to be less affected by heat: the maximum temperature over 5,000 meters is 59 degrees; for 10,000 meters, it is 50 degrees; and for the marathon it’s 45.5 degrees, which gets cold. (Walking is another story: it’s a less efficient movement than running, which means more of the energy you burn is lost as heat, which is probably why the majority of heat illness episodes in elite track competitions happen to walkers.)
Finally, the practical side. Here’s a table prepared by the authors that shows, for marathon runners at different gaits, how much you should expect to slow down based on air temperature, WBGT, or heat index.
These calculations are based on very elite runners, who are different from you and me. They work fast, which generates a ton of heat and can promote cooler temperatures. Conversely, they don’t carry a lot of insulation, which can predispose them to prefer warmer temperatures. That said, you can roughly extrapolate from this graph: if the weather conditions suggest that a two-hour marathoner will slow down by two minutes, you can estimate that a four-hour marathoner might slow down somewhere in order. four minutes. – but keep in mind that your mileage may vary.
For more Sweat Science, join me on Twitter and Facebook, sign up for the email newsletter and check out my bookEndure: mind, body, and the strangely elastic limits of human performance.