The inuence of meteorology and air quality on parkrun athletic performance

Despite increased awareness of climate change and urban air pollution, little research has been performed to examine the inuence of meteorology and air quality on athletic performance of the general public and recreational exercisers. Anecdotal evidence of increased temperatures and wind speeds as well as higher relative humidity conditions resulting in reduced athletic performance has been presented in the past, whilst urban air pollution can have negative short-and long-term impacts on health. Furthermore, pollutants such as Ozone, Nitrogen Dioxide and Particulate Matter can cause respiratory and cardiovascular distress, which can be heightened during physical activity. Previous research has examined these impacts on marathon runners, or have been performed in laboratory settings. Instead, this paper focuses on the potential impacts on the general public. With the rise of parkrun events (timed 5 km runs) across the UK and worldwide concerns regarding public health in relation to both air quality and activity levels, the potential inuence of air quality and meteorology on what is viewed as a ‘healthy’ activity has been investigated. A weekly dataset of parkrun participants at fteen events, located in London UK, from 2011–2016 alongside local meteorological and air quality data has been analysed. The biggest inuencer on athletic performance is meteorology, particularly temperature and wind speed. Regression results between parkrun nishing times and temperature predominantly show positive relationships, supporting previous laboratory tests. Increased relative humidity also causes slower nishing times but in several cases is not statistically signicant. Higher wind speeds also result in slower times and in contrast to temperature and relative humidity, male participants are more inuenced than female by this variable. Although air quality does inuence athletic performance to an extent, the heterogeneity of pollutants within London and between parkrun events and monitoring sites makes this dicult to prove decisively. It has been determined that temperature and relative humidity can have the largest detrimental impact on parkrun performance, with Ozone also having an impact. The inuence of other variables cannot be discounted however and it is recommended that modelling is performed to further determine the extent to which ‘at event’ meteorology and air quality has on performance. In the future, there results can be used to determine safe operating and exercise conditions for parkrun and other public athletics events.


Materials And Methods
The parkrun events examined in this study are all located within Greater London. This location was chosen because of the relatively high spatial coverage of air quality monitoring stations and parkrun events. Furthermore, London often breaches European air quality limits with poor UAQ contributing to an estimated 9,400 premature deaths, which costs between £1.4 and £3.7 billion per year [78][79][80][81]. Finishing times for participants of fteen parkrun events (Fig. 1) from 2011-2016 were provided by the parkrun research board. These were selected due to their close proximity to Department for Environment, Food and Rural Affairs (DEFRA) monitoring stations (< 15 km) to utilise as accurate 'at event' readings as possible due to the high spatial variability of air quality [82][83][84][85]. The parkrun dataset contains details of the parkrun location, event date, individual run times of each participant on that corresponding date, their gender and age group. The parkrun nishing time data was anonymised prior to research access being given in accordance with the completed and agreed ethics procedures (ERN_17-1583).
For each parkrun event, the weekly mean nishing time was calculated and then used for further analyses. This was for the complete participant list before being broken down into male and female times. It is important to note that due to the increasing success of parkrun events, average nishing times continue to increase due to growing participation levels. Therefore, decomposition of the run times was performed and the remainder value was used for analysis against the explanatory variables of temperature, relative humidity, wind speed, O 3 , NO 2 and PM 2.5 (Fig. 2).
The removal of long term trend and seasonality is determined to be required due to the variation in parkrun numbers over time as a result of parkrun gaining popularity and changes in participants over the course of the year. The decompose function in the R package 'forecast' was used to determine the seasonal, long term and random components within the data via an additive model. This is used because the seasonality variation remains relatively constant despite an increase in participation.
Meteorological data was obtained from the British Atmospheric Data Centre (BADC) using the Met O ce Integrated Data Archive System (MIDAS). Seven stations ( Fig. 1) were used due to their proximity to the parkrun events being examined. Observations were downloaded for 09:00 on Saturdays to match the starting time of parkrun events and ensure that the values used in analyses were as accurate as possible to those the parkrun participants were exposed to. Air temperature, relative humidity and wind speed variables were downloaded. The worldmet package within R was utilised to import meteorological data for analyses [86]. This was quality checked against the MIDAS datasets and it was shown that temperature values were the same but relative humidity in some cases varied by up to 2%, although this is likely due to the formatting algorithms used in processing the data [87].
Air quality data for Greater London was retrieved from the DEFRA Automatic Urban and Rural Network (AURN) SITES, between 08:00 and 10:00 local time at background monitoring sites. This includes hourly readings for NO 2 , O 3 , PM 2.5 and PM 10 . PM 10 was subsequently removed from analyses due to its high correlation to PM 2.5 , while the latter was retained due to the greater association of smaller particles with deleterious health effects. Locations of the monitoring sites can be seen in Fig. 1 and were selected due to them being urban background sites, i.e. not in direct proximity to roadsides and vehicular pollution, measuring all or most of the above pollutants and their proximity to parkrun events. The mean 08-10:00 air quality values were found and used for analysis to capture the air quality participants were potentially exposed to before and during the events.
Each parkrun site was paired with the closest DEFRA AURN and Met O ce locations (Table 1). Although some are not optimally placed, they are indicative of the local air quality. Due to not all measurement sites recording all of the desired explanatory variables, some events have been analysed against a reduced times series as dates containing missing data have been removed from analysis. Likewise, with discrepancies in the meteorological data. Prior to analysis, parkrun nishing times over ninety minutes were discarded, as these were technical issues indicated by parkrun [88]. Table 1 The analysed parkrun events and their corresponding air quality and meteorological monitoring location along with distances between the sites. Correlation analyses between the decomposed nishing times and the explanatory variables were performed for the whole data set as well as gender subsets, as used by Helou et al. [25]. Each of the parkrun events was also examined separately to determine whether certain locations were more in uenced by the measured variables. Linear regression analyses, the common technique used in aforementioned marathon studies [44], was used to compute the R 2 value, showing the total percentage of variance in nishing times explained by the control variables.
Analysis to determine the in uence of UAQ and meteorology on the average weekly parkrun nishing time was achieved by multiple linear regression analysis that considered the combined in uence of NO 2 , O 3 and PM 2.5 on nishing times. For meteorology, temperature, relative humidity and wind speed were used as the independent variables. This analysis method reintroduces a form of natural seasonality that is initially striped from the time series. This is done to remove the 'slowing' in uence of New Year's resolution runners and general loss of physical tness over the Christmas period, rather than leaving the long term trend and seasonality in from the beginning of analyses. It also allows for a more representative insight into real world processes and in uences. Post-test analysis was also performed using the following diagnostic tests; Quantile-Quantile, Scale-Location, Fitted vs Residuals, Cooks-Distance and ACF plots and histograms of residuals (Fig. 3).
It needs to be noted that this research follows a time series rather than space-time series analysis. Although there could be variation between parkrun nishing times and the local air quality and meteorology, there are other factors that would also need to be considered such as differences between event surfaces and elevation pro les that could lead to false conclusions. Controlling these factors over a spatial analysis would prove challenging and probably a paper in its own right. Examination of individual parkrun events shows that ve locations have signi cant relationships between nishing times and temperature (Table 2). Of these, however, Bromley parkrun has a negative relationship with nishing times, suggesting that quicker performances occur under warmer conditions (p = 0.05). Examination of age groups showed some interesting results. Increased temperatures and wind speeds were detrimental to nishing times of the age groups shown in Table 3. Temperature shows signi cant positive relationships with the middle-aged to older age groups, with no apparent in uence on the children, youth and young adult competitors in the 25-29 and younger age groups. Correlation and linear regression analysis for this explanatory variable shows a number of signi cant relationships. With the exception of the 25-29 age group and the Richmond event (overall and male subset), these all show that increased RH causes slower nishing times (p = < 0.08).

Wind Speed
Signi cant results were only found at seven of the fteen events as well as the overall and male and female subsets (p = < 0.08, Fig. 4). R 2 values ranged from 1-12% and a student's T-test revealed a signi cant difference between the mean run time at high (> 6 ms − 1 ) and low (< 6 ms − 1 ) wind speeds (p = < 0.01) for the overall and male datasets. At a number of events, wind speed increases saw correspondingly higher, thus slower, parkrun nishing times. No particular age group showed a greater in uence of wind speed on their nishing times compared to the others (Table 3).
In most cases, male competitors showed a signi cant relationship with wind speed that was not matched by the corresponding female analysis. For example, at Wimbledon parkrun male nishing times were shown to be negatively impacted by increased wind speeds through correlation (coe cient − 0.16), regression (coe cient − 2.25) and multiple linear regression (coe cient − 2.75) analysis (p = 0.01).

Combined in uences
Multiple linear regression was performed using the in uence of temperature, relative humidity and wind speed (Eq. 2). These three variables explained 10% of the variance in average parkrun nishing times (p = < 0.01), with increased values resulting in slower nishing times. Results of the multiple linear regression are shown in Table 4, with 16% of the variance in nishing times at Bushy Park attributed to the three variables.  Despite some runners preferring adverse conditions, running is a weather interference sport where certain conditions, such as the meteorology, will in uence performance [89]. This is particularly so for higher temperatures that can alter the bodies thermoregulatory systems and increase fatigue and power output [23,36,37]. Regression results between parkrun nishing times and temperature predominantly show positive relationships. Although individual parkrun analysis shows a third to have signi cant results, temperature appears to be the largest in uencer on running times, supporting the laboratory tests performed by No and Kwak [90] with real-world results from London. Time increases of seconds compared to the larger and more substantial performance reductions shown by Helou et al. [25] is to be expected considering the differences in event length and duration. This difference between parkrun and marathon studies is most likely due to the reduced distance and period of time required to complete parkrun events, and therefore the reduced environmental exposure experienced by participants.
Gender analyses suggest that female run times are more susceptible to increased temperatures than male. This contrasts the work of Vihma [33] who showed male athletes to be more susceptible to high temperatures during the Stockholm marathon. This is theorised to be due to males generally have a smaller ratio of surface area to body mass compared to females, making them less e cient at dissipating heat build-up during exercise and prompting earlier decreases in performance due to temperature regulation [34,35,91]. However, other studies have shown female participants to be in uenced more than male with this being attributed to females having a higher core temperature that is a disadvantage when exercising in warmer conditions [92]. Finally, several events showed no impact of temperature on performance, as shown by Maffetone et al. [44] and Havenith et al. [93], and this is supported by others who have shown temperature to have little to no effect on female performance whilst male athletes have an optimal range [42,43]. Overall, this research suggests that both genders are, in general, impacted by meteorology, as would be expected based upon previous research [25,32,34,35].
Age group analysis suggests that temperature and wind speed is detrimental to several age groups, predominantly in the thirty years of age and older groups.
This partially contrasts research that suggests that younger demographics are most negatively impacted by temperature through increased heat gain and reduced dissipation [45,46,94,95]. It should be noted, however, that the impact on older competitors does correlate with research that ageing reduces muscle mass, metabolism and thus thermoregulatory adjustments [45,46,94,95]. This would explain the decrease in performance under higher temperatures for older age groups and also agrees with previous research into the in uence of urban heat islands and pollution on vulnerable population demographics (i.e. young and old) [76,[96][97][98][99][100].
With regards to temperature in uences across the London parkrun events, studies on the Greater London urban heat island (UHI) effect show that Kew Gardens is around 1.5 o C warmer than Bracknell, whilst Heathrow is about 0.5 o C warmer [101]. Due to these variations across the study site, the strong temperature relationships at some events could also be partially attributed to the cities UHI effect, although parks and urban green spaces can reduce the effect of the UHI [102], potentially reducing the in uence of the UHI on performance slightly.

Relative Humidity
Results suggest that in most cases elevated relative humidity causes slower nishing times. These decreases in performance in elevated humidity is due to a reduced ability to disperse excess body heat generated during exercise, leading to earlier and increased fatigue in participants [34,35,47]. Interestingly, the 70-75 and 80-85 age groups showed the largest decreases in performance when relative humidity rose from 40-55% to over 85%, suggesting that they are also less e cient at dispersing excess heat. These results also support those reported by Helou et al. [25] who determined humidity to be a signi cant in uencer on performance after temperature. This suggests meteorology is the main external control on athletic performance [25,[32][33][34][35].

Wind Speed
Around half of the events saw signi cant decreases in performance as wind speed increased and there is a signi cant difference between nishing times under high (> 6 ms − 1 ) and low (< 6 ms − 1 ) wind speeds. This corresponds with the work of Davies [48] who showed that headwinds will result in a greater drag force and slower running times. However, the majority of parkrun course are lapped with participants encountering multiple wind directions over the course of the ve km, potentially leading to the insigni cant results shown.
The gender analysis showed a number of occasions where female participants were not signi cantly in uenced by wind speed. This could be due to male competitors potentially being larger than their female counterparts and therefore having a larger silhouette to move through the increased wind resistance.
Previous studies, utilising the performance of cyclists in time trials, have demonstrated that the cooling effect from wind can improve performance by up to 4.4% [103,104]. Perhaps more importantly in the UK, wind chill can have a negative impact on performance in cooler conditions, reducing core body temperature and increasing the amount of anaerobic glycolysis in active muscles, leading to increased fatigue [105]. Furthermore, greater tness levels does not necessarily result in improved cold weather performance, this is more often dictated by body shape, size and sex [106].

Combined In uences
The combined in uence of temperature, relative humidity (RH) and wind speed has been noted earlier in this research [25,32,33], with the results mostly mirroring those and suggesting that they can signi cantly result in slower running times. This is supported by the work of Pezzoli et al. [107] who believed these three variables to be the greatest in uencers on running performance.

Ozone
Only the male results showed a close to signi cant relationship (p = < 0.09) between Ozone and nishing times. Despite a lack of statistically signi cant results between nishing times and O 3 , there are consistent positive relationships shown between the two variables at most events and subsets. This likely suggests that the irritant quality of the pollutant on the respiratory system could potentially in uence athletic performance and to some extent supports the work of Helou et al. [25] on marathon performances [17]. Furthermore, past research has shown that O 3 can decrease lung function therefore performance in laboratory tests [53][54][55][56][57][58][59][60], although this may not be the case under real-world conditions when the potentially exposure of runners could vary considerably.

Nitrogen Dioxide
There were no signi cant or close to signi cant relationships shown between NO 2 and nishing times at the London parkrun events. However, the majority of results show a negative trend which could suggest that quicker nishing times are recorded under higher NO 2 conditions, which is unlikely due to it also being an irritant [17]. However, due to the processes linking Nitrogen Oxides, VOCs and sunlight (often in elevated temperatures) that create O 3 , it could be suggested that high NO 2 levels are found in cooler temperatures, conditions that are likely to improve running times [25,40]. Additionally, the inverse of this can also be partially suggested for the decrease in performance shown in elevated O 3 levels, which are likely to occur in increased temperatures, although the irritant quality of the pollutant is still likely to be detrimental in performance, as suggested by Helou et al. [25]. This temperature-O 3 relationship cannot be investigated in multiple linear regression due to the multicollinearity of the predictor variables. A nal possible explanation for the improved nishing times shown in elevated NO 2 levels is that Nitrate and related species are vasodilators, reducing arterial pressure and increasing blood ow, although examination of the in uence of this on athletic performance has not been performed [108,109,110].

PM 2.5
With the exception of negative relationships at Bushy, Bromley and Lloyd parkruns, PM2.5 has no signi cant results when compared to nishing times. The three aforementioned signi cant negative relationships are result that, although not disagreeing with previous studies, goes against common thinking of particulate matter being an irritant and highly detrimental to human health [17,97]. However, particulate matter impacts are often seen as a long-term health hazard, possibly explaining the lack of a detrimental impact on short-term performance, if not the few 'bene cial' relationships [97]. Overall, the overall trend and majority of the results suggests that there is no real relationship between short-term athletic performance and PM 2.5 concentrations.
Due to spatial variability in air quality, how representative the air quality data is also needs to be considered. O 3 is generally a regional pollutant with less variation over London and between monitoring locations and parkrun events. Therefore, the signi cant relationships between O 3 and parkrun nishing times can be considered accurate. However, NO 2 and PM 2.5 levels are more likely to be in uenced by local sources and may result in discrepancies between monitoring and parkrun locations, potentially contributing to insigni cant relationships. Although monitoring locations are not ideally located in some instances, they are indicative of air quality levels and this research is not aimed to create a predictive model, more generate insights. To overcome this, in-situ monitoring at parkrun events could be performed, albeit a potentially costly option. Modelled air quality data could also be used to reduce the likelihood of variability in air quality levels between monitoring sites and parkrun events.
Some further considerations that could be utilised in the future for additional studies include tracking performance changes over time of individual parkrun participants, although this could not be done as parkrun ID numbers were not provided due to data protection. This meant that individual runner's performances could not be followed over the study period to determine whether air quality and meteorology impacted their performance. Consequently, this research provides a useful overall view on the impact of these variables, but not the resolution to determine the impact on individuals. Being able to follow individuals, along with their physiological data such as lean body mass ratio, heat dissipation of tissues and relative maximum oxygen consumptions that are all in uential for performance could also provide additional insight in to our results [92].

Conclusions
The increasing popularity of parkrun events has allowed this research to examine the in uence of local air quality and meteorology on short-term athletic performance at a weekly time scale. Previous research has focused on laboratory-based studies or yearly marathon events. Through fteen Greater London parkrun events, DEFRA AURN and meteorological monitoring analysis has shown a number of relationships between variables and running performance. This includes additional subsets of parkrun data to examine gender differences.
Although the variance in run times explained by these variables are small, the results correlate well with previous laboratory and real-world marathon studies, particularly for temperature, relative humidity and O 3 [25,32]. This also highlights the importance and impact of body temperature regulation and power output shown by past research [23,[34][35][36][37]. Additional analysis in to the impact of wind speed, wind chill, precipitation and radiation has also been performed and has highlighted the impact, or lack thereof, of those variables on performance.
Overall, this research suggests that meteorology has the greatest in uencers on short-term athletic performance, along with O 3 , which is potentially linked to increased temperatures. NO 2 may improve performance, but this is likely to be linked to decreased temperatures whilst PM 2.5 does not appear to have any signi cant impact (at least in the short-term). Furthermore, this research has started to address the gap surrounding short duration athletic performance in the UK, along with utilising a wider spectrum of participants rather than elite runners.
Despite these potentially hindering effects of UAQ and meteorology, it is important to stress that the health bene ts of participating in parkrun events outweighs the short-term exposure to poor UAQ [18]. This is supported by research showing that regular exercise protects against premature deaths attributed to UAQ [22]. Finally, it is important that parkrun and other event organisers, along with policy makers and health care providers are aware of the extent to which air quality and meteorology can impact participants, particularly under future predictions of climate change and urban air quality [110].     Example post-test analysis of residuals, their distribution and ACF plot for the in uence of Ozone on nishing times at Bushy parkrun.

Figure 4
Results of linear regression analysis for the overall (row A), female (row B) and male (row C) parkrun subsets with the three meteorological variables examined.