Comrades marathon 2022 date: how bad will the heat be?

The Comrades marathon is a staple of South Africa’s sporting and cultural life, and the event’s cancellation in 2020 and 2021 was a great loss to the country, robbing tens of thousands of people of a life-affirming and sometimes life-changing experience. However, there is light at the end of the tunnel; there is every indication that the worst of the COVID crisis is behind us, and that life will be gradually returning to normalcy. South Africa is in the early stages of a fourth wave of the virus, but one that will likely be of considerably lesser severity than the third wave now well behind us.*

As a precaution to further reduce the risk of COVID affecting the 2022 edition of the race, the race date has been shifted from its usual mid-winter spot and very nearly into Spring, occurring on the 28th of August. When this announcement was made, the CMA (Comrades Marathon Association) pre-empted concerns about heat by pointing out that the average temperature in August is just half a degree warmer than in June, and the average humidity just 1% higher at 71%.

This pronouncement crinkled some brows, for a couple of reasons. One is prosaic; the average conditions for August are not necessarily a good representation of the conditions at the very end of August; temperatures are rising quite rapidly over the course of that month and so the average values for the month are likely to understate the severity of conditions. The second reason is that the statement itself is not actually particularly informative. Intuitively, we are actually quite poor at interpreting temperatures and humidities in terms of how they will affect our comfort and performance. Intuitively we tend to over-interpret the importance of temperature and largely neglect humidity, but even considering both doesn’t quite cover the complexities of thermal maintenance of the human metabolism.

The human body produces heat at all times and, in accordance with the first law of thermodynamics, that heat must be discharged at the same rate in order for the body’s temperature to remain stable. The harder you are working, the more heat must be discharged. Effectively, pretty much all energy produced by your metabolism, plus energy added to the body through other means (mainly sunlight) must be removed from the body as heat, which is discharged through a range of different means. The first is conduction, which is driven by Fourier’s law which dictates that heat flows from high temperature material to lower temperature material, at a rate proportional to the magnitude of that temperature difference. We are also familiar with sweating as a means of discharging heat; this serves two functions. The first is that it increases the transmissibility of heat from the skin and into the air, because water conducts heat very efficiently. The second mechanism is that the evaporation of sweat removes heat from the body.

Another, less well-known mechanism for heat discharge is breathing. We inhale air at whatever temperature and humidity it may be at, and we exhale it at body temperature and saturated with water (ie 100% humidity). That warming of the air and evaporation of water respresents a significant removal of energy from the body.

The relative prevalence of these mechanisms varies depending on conditions and activity levels. In cold weather, conduction, along with normal breathing, is sufficient (or more than sufficient) to discharge your metabolic waste heat. A scarf covering the mouth is highly effective at keeping you warm for the reason that it captures some of the heat exhaled in one breath, and uses it to pre-heat the next breath.

At higher temperatures, the body increasingly produces sweat to increase the rate of evaporation and thereby discharge heat more rapidly. When running or otherwise working hard, your breathing rate also increases, commensurately increasing the rate at which heat is discharged through that mechanism.

Determining the exact rates of heat transfer through the various mechanisms is a more involved endeavour than appropriate right now (I give lectures on it to Chemical and Biomedical engineering students at Wits, trust me here), but there is a simplified approach that we can take to compare the body’s ability to discharge heat under different conditions.

For the mechanism of conduction, that process is incredibly simple, because the rate of conduction is limited in proportion to the difference between your body’s core temperature and the atmospheric temperature. Body temperature is approximately 35°C, and so rate of conduction is proportional to 35°C - T. ie Q(conduction) ∝△T.

This allows us to highlight one of the ways that our intuitive responses to temperature can be misleading. If the temperature is 5°C, then △T=30°C, and at 20°C, △T=15°C. That 15 degree difference in atmospheric temperature halves your ability to discharge heat through conduction. Another 15 degree increase, however, would reduce △T to 0, which means that your body’s ability to lose heat through conduction drops to 0. The closer you are to body temperature, the more significant each degree difference becomes.

A better way of thinking about temperature in terms of its livability would be to report on △T rather than temperature itself, but that would still not be all that informative, because the evaporative effects are far more significant under severe conditions, making humidity a key variable.

At this point it’s useful to clear up the definition of humidity. Firstly, 100% humidity does not mean that your furniture floats away and you need scuba gear to chase after it. Relative humidity is defined as the amount of water vapour in the air divided by the maximum amount of water vapour that could be carried by the air at that temperature. **

That maximum amount is, in turn, very strongly dependent on temperature and so, 50% humidity at 35°C represents very considerably more water than 50% humidity at 10°C, for instance. The evaporation of water requires a huge amount of energy, making this mechanism a very powerful means of removing heat. It is possible for humans to survive at temperatures in excess of 50°C, as long as the air is dry enough to allow for evaporation to occur. By contrast, conditions of 35°C and 100% humidity are lethal. That is not hyperbole; under those conditions you are not able to discharge heat from your body and, as a result, you die.

This really hammers home the fact that temperature is, by itself, a poor indicator of your ability to discharge heat. An alternative measure that is sometimes reported is something that is termed ‘wet-bulb temperature.’ Physically, this measurement is the temperature that is reached by a thermometer which is wrapped in a thin film of gauze that is soaked in water, with unrestricted air-flow. Evaporation constantly cools such a thermometer, but only up to a hard limit imposed by the amount of water that can be evaporated into a unit of air before it becomes saturated at 100% humidity.

The difference between your body temperature and the wet-bulb temperature would serve as a good indicator of capacity for heat discharge; there is a fairly tight proportionality. However, it is still possible to do a little bit better.

Evaporative cooling, whether by way of sweating or breathing, works through the mechanism of air at a particular temperature and humidity, leaving at approximately body temperature and 100%. By way of the first law of thermodynamics, the energy taken up by this process must be equal to the difference in energy content of the air as it departs (including water vapour), and the energy content of the air as it arrives plus the energy content of the water that gets evaporated. It follows that, with all else being equal, the maximum rate at which heat can be discharged is proportional to this difference in energy contents, hereafter referred to as △H.

We can use a fairly straightforward set of relations to evaluate this. Engineers often use something called a psychrometric chart for evaluating the various conditions of air and water vapour, but we will use some equations instead, because psychrometric charts are monstrously confusing to the uninitiated (and the initiated-but-rusty). Instead, we can use the Arden-Buck equation to give us the maximum water content of an air stream based on its temperature, and then applying a collected set of Specific Enthalpies for the substances involved (namely dry air, liquid water, water vapour) along with the enthalpy of vapourization for water allows us to calculate △H for any particular set of conditions. I have uploaded a simple excel calculator for this calculation here, which can be used for more or less any set of conditions.

Now, the world meteorological organization reports mean temperatures in Durban for June and August as 16.6°C and 17.7°C respectively, with mean humidities of 72% and 75% respectively. If we simply plug these values in, we get a △H that is 4.4% lower in August than in June. These values don’t match those quoted by the CMA, whose statement unfortunately did not point to the source of their data. If we were to take a fair guess using their quoted difference figures, we could for instance assume temperatures of 17°C and 17.5°C, and humidities of 70% and 71%. That would result in a difference in △H’s of 1.9%.

Of course, a 4.4% reduction in power output would be highly significant to performance, and 1.9% is not entirely trivial either. However, this result is not necessarily particularly instructive, because at those average conditions, heat discharge by evaporation is not necessarily a limiting factor on performance. Temperatures around 17°C are not onerous at all, even though they exceed what has been determined to be the optimum range for running marathons, which is temperature between 7°C and 15°C, and humidity between 30% and 50%. Both sets of mean conditions lie outside the optimal range, and that for August lies slightly further outside, suggesting worsened performance.

However, it is only at fairly stringent conditions that heat discharge is a limiting factor on performance and, further, it is also only at fairly difficult conditions that the two evaporative mechanisms totally predominate heat loss over conductive losses.

In a race like the Comrades, heat is likely to only be limiting to power output for about a third of the race duration, so we could loosely estimate a reduction in power output of between 0.63% and 1.46%. This doesn’t factor in the psychological factor of discomfort and the risk of heat exhaustion, nor does it include the somewhat higher sun exposure contributing heat to the system in August. Overall, it’s safe to say that running in late August has to be considered equivalent to adding an additional kilometer to the race.

In 2019, 14.7% of Comrades finishers (~2400 runners) reached the finish line within the last 15 minutes before the 12-hour cutoff time. Applying a simple blanket 1.46% reduction in speed and (wrongly) assuming homogenous distribution within that window, we would have excluded some 1682 of 2019’s finishers or a staggering 10% of the race’s finishers.

Of course, sport doesn’t quite work that way — the heavy clustering of finishers in the 15 minutes prior to each medal’s cutoff time shows that people adjust their effort levels according to the goals that are within reach. And of course, weather doesn’t work that predictably either — Durban produces cold days in August and sweltering days in June, every so often, so the results will always be at the mercy of daily variance.

Nonetheless, a Comrades run in late August will tend to have slower times and fewer finishers than one run in June. That’s still far better than yet another race cancelled due to COVID, but it does come at a cost.

* (Footnotes not directly related to the matter of weather at the Comrades, should possibly be a separate article): The combination of vaccination progress, greater prevalence of natural immunity due to prior infection and considerably warmer weather all indicate that there will be fewer infections and a lower fatality rate among those infected. When the fourth wave will begin is a matter of some uncertainty; projections for that date are based on the lag time between previous waves but there is no certainty that the same behaviour will be exhibited now. Wave behaviour is presumed to arise because of a phenomenon of ‘transient non-susceptibility’ which differs from conventional SIR (Susceptible-Infected-Recovered) models in that people can be removed from the pool of susceptible individuals without necessarily having acquired an active infection themselves, at a rate proportional to the number of actual active infections. This may be some combination of behavioural and immunological phenomena. The behavioural factors include active measures to curb transmission, self-isolation as a result of direct contacts, behavioural changes arising from individual concerns about high infection rates, as well as more durable behavioural changes such as higher levels of caution exhibited by people who have lost someone close to them. Immunological factors contributing to transient non-susceptibility are a far more speculative area. There are two plausible but largely unproven mechanisms. The first is that of undetected asymptomatic infections, which will confer some amount of resistance to the virus but not to the same degree as a full-blown symptomatic infection. This lesser degree of resistance will wane far more rapidly, resulting in temporary reductions in population transmission which decline sharply in a matter of months, allowing transmission to pick back up again. This phenomenon would neatly model the observed wave behaviour and is consistent with the findings of population-wide serological studies and so, while it is unproven, can be presumed with some confidence to be a contributing factor. Another plausible mechanism is low-grade variolation due to regular exposure to replication non-viable COVID virions, or to sub-clinical innoculating doses that do not result in an infection. For either or both of these mechanisms, it cannot necessarily be assumed that the periodicity of wave behaviour will be the same for the Delta variant as for previously-circulating variants, for the reason that the Delta variant appears to be far more replication-competent than those previous variants. Consequently, the proportionality between active infections and transient non-susceptibility may differ, affecting the time period between waves. It may be higher or lower, or it may be unchanged — it is a matter of uncertainty.

It may also be the case that the greater replication competence of Delta reduces the proportionality betwee non-infectious contact with the virus and infectious contact, which would yield larger initial waves but greater drop-off between waves. This, too, is a matter of uncertainty.

** More correctly, it is the partial pressure of water vapour in the air divided by the vapour pressure of water at the temperature of the air.