ELY, MATTHEW R.; CHEUVRONT, SAMUEL N.; MONTAIN, SCOTT J.
There is ample evidence that distance running performance slows as weather warms (4,5,11,12). This relationship was recently quantified for marathon running (4) using the wet-bulb globe temperature (WBGT) as a weighted weather index that combines the effects of ambient temperature (Tdb), relative humidity (Twb), and solar radiation (Tg) (19). One limitation to using the WBGT is that it may not distinguish qualitatively important differences in weather. For example, a hot, dry, overcast day can result in a WBGT that is very similar to a cool and humid day with plenty of sunshine.
There exists a popular notion that cloud cover and/or low solar radiation increase the likelihood of running a fast marathon. This information can be found in anecdotal reports (2,15), authoritative reference books for runners (17), and scientific publications (9) alike. The possibility that a single weather parameter like cloud cover could be advantageous is supported by the observations that the addition of direct solar radiation is associated with increased perceived air temperature (10), heart rate, ventilation, skin temperature (1,16), sweat rate (7,16), discomfort levels (6), and oxygen consumption when ambient temperature is held constant (16). Therefore, it is plausible that the presence of cloud cover and/or low solar radiation (9,14,18) might be an additive factor contributing to the likelihood of running a fast race, which cannot be easily discerned using the WBGT.
The potential for weather to affect marathon performance differently between men and women is also a matter of some debate. When comparing marathon performances between men and women using WBGT, the sexes seem to slow similarly in response to warming weather (4). Other evidence (11) using air temperature suggests that women may perform best in slightly cooler weather compared with men. In neither study was it possible to identify "optimal" weather for marathon running, but it is clear that cool weather (Tdb 10-12°C) is the norm for fast field performances (5,11) and also extends to endurance exercise in laboratory studies (8).
One way to determine the validity of increased cloud cover/low solar radiation (2,9,15,17) and to compare "optimal" air temperatures (5,8,11) for fast marathon performances in men and women is to examine the weather conditions present during the world's fastest marathon performances. The principal purpose of this investigation was to determine whether the presence of cloud cover and/or low solar radiation is related to fast marathon performance. In addition, we sought to describe, in detail, the weather conditions present during fast marathons to further (4) compare marathon performance between the sexes.
Marathon racing performances for men and women were collected and binned into four categories: 1) first-place finishing times of the Boston (Boston, MA), New York (New York, NY), Twin Cities (Minneapolis/St. Paul, MN), Grandma's (Duluth, MN), Richmond (Richmond, VA), Hartford (Hartford, CT), and Vancouver (Vancouver, Canada) marathons of men and women for the past 37, 29, 24, 23, 6, 12, and 10 yr, respectively (4); 2) the top 10 fastest marathons ever run by men and women; 3) world record (WR) marathon performances for men and women; and 4) first-place times for the men's and women's Olympic marathons. These data are in the public domain; therefore, written informed consent was not required from individual athletes.
Hourly weather data were obtained through the Air Force Combat Climatology Center (AFCCC) for the day and time of each race. The weather data examined included dry-bulb (Tdb), wet-bulb (Twb), and black-globe (Tg) temperatures; relative humidity (RH); cloud cover; wind speed; and solar radiation (SR). Average race weather parameters and WBGT (19) were calculated for the duration of each race. Cloud cover was categorized into four conditions on the basis of the one-eighths rule used by meteorologists (clear skies = 0/8-2/8, scattered clouds = 3/8-4/8, partly cloudy = 5/8-7/8, and overcast = 8/8). To examine the impact of solar load, the range of solar radiation values (130-930 W·m−2) from the Boston, New York, Twin Cities, Grandma's, Richmond, Hartford, and Vancouver marathons were divided equally into thirds; the top third (747-930 W·m−2) was considered a high solar load, the middle third (524-746 W·m−2) was medium, and the bottom third (130-523 W·m−2) was considered low solar load. The same solar radiation ranges were used to classify the WR and top 10 races. No attempt was made to normalize solar radiation to peak values, because all marathons occurred in similar latitudes and, therefore, were exposed to similar solar angles. All AFCCC weather parameters were compared for accuracy against actual measures provided by medical directors, race officials, and race synopses when available. Solar radiation and cloud cover are not synonymous measurements, although, intuitively, they should be linked. The solar radiation values reported are based on Tg (personal communication with AFCCC), and Tg is affected by changes in direct solar radiation, indirect solar radiation from nearby clouds, and radiant heat from surrounding structures. The distinction between the two measures becomes clear in the following example: higher solar radiation measures can occur at noon when overcast than when the sky is clear at dawn.
To examine the influence of ambient temperature (Tdb) on cloud cover and solar load, the Boston, New York, Twin Cities, Grandma's, Richmond, Hartford, and Vancouver Marathons were binned by 5° increments of Tdb against cloud cover and solar load. Course records that occurred during these marathons were extracted as a subpopulation and were presented similarly. For comparison purposes the same technique was used for the top 10 and world record performances.
The primary interest of this study was to determine whether the presence of cloud cover and/or low solar load relate to fast marathon performances. To do this, odds ratios were calculated using the results of Boston, New York, Twin Cities, Grandma's, Richmond, Hartford, and Vancouver marathons to determine whether the overcast to partly cloudy conditions were present more often in record setting marathons compared with clear to scattered cloudy conditions. The odds ratio was calculated by the equation: (no. of course records set when overcast and partly cloudy/no. of races run when overcast and partly cloudy that were not course records)/(no. of course records set when clear or scattered clouds/no. of races run when clear or scattered cloud that were not course records). Similarly, an odds ratio was calculated for the likelihood of a record being set in a high compared with low solar load conditions with the equation: (no. of course records set when the solar load was low/no. of races run when solar load was low that were not course records)/(no. of course records set when solar load was high/no. of races run when solar load was high that were not course records). A 95% confidence interval around the odds ratio was calculated on the basis of Woolf's method (13). The addition of the 95% confidence interval to the odds ratio identifies the likely range of the true differences between the cloud covers and solar loads (13). To accurately describe in detail the weather conditions present during fast marathons, the mathematical means and range of binned data for the course records, top 10, world records, and Olympic races were calculated and reported to characterize weather parameters.
In an effort to better understand the effects of weather on marathon performance, winning performances for the men's and women's Olympic marathons (OM) were compared with the existing historical marathon world record (WR) for those years as a percent off of world record time [(OM − WR)/WR × 100] and are presented as means and standard errors to characterize the precision of the population performance. Hourly weather data for these races were obtained as described above. World record performances were plotted against Olympic marathon performances to visually compare race times relative to a line of identity.
The time frames selected for the Boston, New York, Twin Cities, Grandma's, Richmond, Hartford, and Vancouver marathons were consecutive years where the races had not undergone a significant course change and where detailed race results could be obtained. There were 141 race years analyzed in this data set, which included 24 course records (CR) for men (M) and 34 for women (F) (Boston 7 M, 9 F; New York 5 M, 8 F; Twin Cities 3 M, 8 F; Grandma's 2 F; Richmond 3 M, 1 F; Hartford 5 M, 5 F; and Vancouver 1 M, 1 F). The mean of the weather data for CR of men (Table 1) and women (Table 2) were very similar, though the means were slightly higher for women (Tdb = 12.8°C (M), 13.6°C (F)).
The 10 fastest marathon times run by male individuals occurred in four different cities (Berlin, London, Chicago, and Rotterdam), with 8 of the top 10 occurring in two different cities (4 in Berlin and 4 in Chicago). Three of the fastest times were run in the 2003 Berlin and 2 in the 2002 Chicago marathons. The 10 fastest marathon times by female individuals occurred in five different cities (Berlin, London, Chicago, Boston, and Beijing), with 7 of the top 10 occurring in two different cities (4 in Berlin and 3 in London). Two of the fastest times were run in the 2003 London marathon. The means of the ambient temperature data were similar between sexes (Tdb = 11.0°C (M), 12.6°C (F)) and with those for course records (Tables 1, 2).
The men's world record has been broken 17 times since June 1954 (weather data were not available before this date); however, weather conditions could only be obtained for nine of these occasions. The women's world record has been broken seven times since April 1983 (first official records). The mean and range of the weather parameters for these races are presented in Table 1 for men and in Table 2 for women. Four of the top 10 fastest times by an individual were also included in the world record-breaking races for men, and 5 of the top 10 times for women were included in the world record-breaking data set. The means of the weather data (Tdb = 11.0°C (M), 12.6°C (F)) are again similar between sexes and to the course record and top 10 data (Tables 1, 2).
Cloud cover was obtained for 134 race years of Boston, New York, Twin Cities, Grandma's, Richmond, Hartford, and Vancouver marathons. In this data set, 25 course records (11 (M), 14 (F)) were set in overcast to partly cloudy skies, and 33 (13 (M), 20 (F)) in clear to scattered cloudy skies. Sorting the data into 5°C temperature increments revealed an essentially normal distribution of races centered around the 10.1-15°C range, with 46% of races occurring in clear and scattered cloudy conditions and 54% in partly cloudy and overcast conditions (Fig. 1A). Likewise, course records were set in all cloud covers across the air temperature range (Fig. 1B). The presence of cloud cover was not associated with a fast marathon race when odds ratios were calculated using the entire data set (0.68 [95%CI = 0.28-1.6] (M); 0.51 [0.23-1.1] (F)) or when restricted to the 5-15°C ambient temperature range associated with fastest race performances (0.42 [0.15-1.2] (M); 0.43 [0.15-1.2] (F)). None of the odds ratios were different from 1.0.
FIGURE 1-Frequency d...Image Tools
Solar load values were obtained for 137 race years of the Boston, New York, Twin Cities, Grandma's, Richmond, Hartford, and Vancouver marathons. In this data set, 21 course records (9 (M), 12 (F)) were set in low solar load conditions and 17 (7 (M), 10 (F)) set in high solar load. As illustrated in Figure 2A, low, medium, and high solar radiation conditions have approximately equal distributions across the ambient temperature conditions. Likewise, course records have been set in all solar conditions independently of ambient temperature (Fig. 2B). Low solar conditions were not associated with faster race performances when odds ratios were calculated using either the complete data set (1.4 [0.5-4.1] (M); 1.3 [0.5-3.4] (F)) or when restricted to the 5-15°C range, where the majority of course records have been set (N = 14 records, 8 (M) and 5 (F); (2.3 [0.57-8.9] (M); 1.1 [0.29-4.4] (F)). None of the odds ratios were different from 1.0.
FIGURE 2-Frequency d...Image Tools
Cloud cover and solar load for the top 10 performances and world records for men and women are presented in Figure 3A and B, respectively. Cloudy conditions again were not a good predictor of race performance. In fact, 73% of these races occurred in clear and scattered cloudy conditions (Fig. 3A). Low solar loads were present for 44% of the top 10 and world record performances, whereas high solar load conditions were present for only 14% of these performances (Fig. 3B). The solar radiation values of the top 10, world record, and course records (567-614 W·m−2) are similar to solar radiation values associated with fast performances presented by Ely et al. (4).
FIGURE 3-Frequency d...Image Tools
The mean and range of the weather parameters for the men's (1956-2004) and women's (1984-2004) Olympic marathons are listed in Table 1 for men and Table 2 for women. Weather data could not be obtained for the 1960 or 1968 men's marathons. In an effort to assess the relative contributions of ambient temperature and cloud cover on race performance for this elite event, the winning times for the men's and women's Olympic marathon were plotted against the existing world record at the time the race was run (Fig. 4). The icons depicting cloud cover in Figure 4 illustrate that regardless of cloud cover, performances in the Olympic marathons have been consistently slower than the existing world records. In addition, despite overlapping solar radiation conditions between Olympic and world record races, the mean solar load has been lower during Olympic races than during world record races (Tables 1 and 2). The consistent weather feature associated with the slower marathon times during Olympic competitions has been higher Tdb. Usingthe nomogram predicting performance decrement with increasing WBGT presented by Ely et al. (4), the 8.0°C difference in WBGT between Olympic and WR conditions for men would suggest a 3.5% slowing on the basis of an average finishing time of 128.9 min. Similarly, the nomogram would suggest a 4% slowing of performance for women based on the basis of an increase of 9.2°C WBGT and an average finishing time of 140.4 min. The observed difference between Olympic and world record races averaged 2.8 ± 0.62 and 4.4 ± 1.2% (mean ± SE) for men and women, respectively.
FIGURE 4-A compariso...Image Tools
The principal finding of this study is that neither cloud cover nor low solar loads are associated with running a fast marathon. The presence of cloud cover and/or low solar conditions did not increase the odds of running fast whether odds ratios were calculated for cool weather conditions (5-15°C), which are commonly associated with fast marathon performances, or for the range of temperatures within the data set (0-30°C). Furthermore, the present data indicate that world records and all-time fastest marathons have been run in conditions incompatible with the prevailing notion in which cloud cover and/or low solar conditions offer performance advantages for marathon runners (2,9,14,15,17,18). This may be attributable, at least in part, to the presence of a low solar zenith angle during early morning hours when most races are run. Our interpretation is that cloud cover offers no advantage to performance when the environment is conducive to heat loss (Tdb < skin temperature), which is the norm for most of the races studied (Tables 1 and 2), or when the weather warms (Fig. 4), so long as conditions remain compensable (7).
The data (Figs. 1-3) indicate that most course and world records are set in the air temperature range of 10-15°C (Tables 1 and 2), which corroborates the findings of others (5,8,11). When these races are compared with high-stakes, competitive races held in warmer weather (Fig. 4), the optimal nature of cooler weather for fast running is clear. The Olympic marathon could be considered one of the most talented international fields assembled, because of qualifying standards and the prestige of winning an Olympic medal. However, despite the high level of talent in this field, races have been an average of 2.8 ± 0.62% and 4.4 ± 1.2% (mean ± SE) slower for the men and women, respectively, (Fig. 4) than the historical world record. These slower times were associated with an 8.0°C (M) and 9.2°C (F) higher WBGT. This supports previous investigations that show that small increases in air temperature cause competitive marathon performances to slow (5,11). The magnitude of performance decrement observed is consistent with the predicted quantitative effects of 3.5% for men and4.0% for women on the basis of WBGT and runner ability (4). It is illuminating to point out that "warmer" Olympic marathon weather (3-24°C) (Tables 1 and 2) is still well below what is typically considered "warm" or "hot" conditions in the laboratory (> 30°C).
In this analysis, the fastest men's and women's performances (top 10, world records) occurred in similar weather conditions (Tables 1 and 2). This unique but simple observation agrees with earlier investigations that performances by men and women are affected similarly by the weather (3,4). In addition, the odds ratios for cloud cover and solar load from this study were comparable for men and women. Thus, the temperature range in which both sexes set records (Tables 1 and 2) and the similar odds ratios for cloud cover and solar load argue against differences (11) between sexes in ability to cope with weather.
One limitation to this study is that the marathons selected for analysis (excluding Olympic summer marathons) all occur during the spring and fall seasons and at similar latitudes. Therefore, the spread of possible weather parameters is somewhat constrained, but ecologically valid. Additionally, the data set contains marathons with morning start times; therefore, temperature conditions usually increase during the race, and the absolute solar radiation is low compared with the maximal possible load for the day. This analysis of weather and marathon running also includes an overrepresentation of certain weather conditions because (1) some of the top 10 races occurred on the same day in the same race and (2) some of the top 10 were also included in the world record data set. The comparison of world records with Olympic marathon times assumes that the level of talent in the field is equivalent to what would be encountered in other major marathons. To our knowledge, the qualifying standards (currently, 2:18 (M), 2:42 (F)) for the 2000 and 2004 Olympic marathons are the most stringent in the world. In addition, some Olympic marathons have been run in the evening hours to avoid the hottest time of day, whereas many world records have been achieved in the early morning. Therefore, the data should be interpreted with caution, particularly in the estimation of performance variability. At the same time, these races still qualify as the fastest times by individuals. Although fast times in these races can be attributed to the level of talent of the field, the importance of temperature and the lack of effect of cloud cover for optimal marathon running performance remain apparent.
The presence of cloud cover and low solar load do not impact the likelihood of running a fast marathon time. Whether odds ratios were calculated for only cool weather conditions or for the range of temperatures within the data set, the presence of cloud cover and/or low solar conditions did not increase the odds of running fast. Fast marathons seem to require low ambient temperatures, as evidenced by the similar Tdb (10-15°C) for world records, course records, and top 10 all-time fastest individual performances for both men and women. Olympic marathon data further support (4) that highly competitive races slow as weather warms. Lastly, men and women seem to be similarly affected by weather.
The opinions or assertions contained herein are the private views of the author(s) and are not to be construed as official or reflecting the views of the Army or the Department of Defense. Any citations of commercial organizations and trade names in this report do not constitute an official Department of the Army endorsement of approval of the products or services of these organizations. (Funded by MRMC PRMRP 033015.)
We would like to thank SSgt Michael Ross, Air Force Combat Climatology Center, Asheville, NC for his help in obtaining very thorough weather data. In addition, Jack Fleming (Boston Athletic Association), Shane Bauer (Grandma's Marathon) and Will Nicholson (MS4 University of Minnesota Medical School) for their help in obtaining race data.
1. Adams, W. C., R. H. Fox, A. J. Fry, and I. C. MacDonald. Thermoregulation during marathon running in cool, moderate, and hot environments. J. Appl. Physiol.
2. Bodishbaugh, R. G. Boston Marathoners get red carpet treatment in the medical tent. Phys. Sports Med.
3. Cheuvront, S. N., and E. M. Haymes. Thermoregulation and marathon running: biological and environmental influences. Sports Med.
4. Ely, M. R., S. N. Cheuvront, W. O. Roberts, and S. J. Montain. Impact of weather on marathon running performance. Med. Sci. Sports Exerc.
5. Frederick, E. C. Hot times. Running Magazine
6. Gagge, A. P., and R. R. Gonzalez. Physiological bases of warm discomfort for sedentary man. Arch. Sci. Physiol.
7. Gagge, A. P., and R. R. Gonzalez. Mechanisms of heat exchange: biophysics and physiology. In: Handbook of Physiology, Section 4, Environmental Physiology
, M. J. Fregly and C. M. Blatteis (Eds.). New York, NY: Oxford University Press, pp. 45-84, 1996.
8. Galloway, S. D. R., and R. J. Maughan. Effects of ambient temperature on the capacity to perform prolonged cycle exercise in man. Med. Sci. Sports Exerc.
9. Hughson, R. L. Primary prevention of heat stroke in Canadian long-distance runs. CMA J.
10. Lee, D. H. K., and J. A. Vaughan. Temperature equivalent of solar radiation on man. Int. J. Biometeorol.
11. Martin, D. E., and J. F. Buoncristiani. The effects of temperature on marathon runners' performance. Chance
12. McCann, D. J., and W. C. Adams. Wet-bulb globe temperature index and performance in competitive distance runners. Med. Sci. Sports Exerc.
13. Metulsky, H. Intuitive Biostatistics
. New York, NY: Oxford University Press, pp. 70-89, 1995.
14. Morgan, P. P. Running into danger: heat stroke in competing runners. CMA J.
15. Newton, A. F. H. Running
. London, UK: Witherby, 1935.
16. Nielsen, B., K. Kassow, and F. E. Aschengreen. Heat balance during exercise in the sun. Eur. J. Appl. Physiol.
17. Noakes, T. D. Lore of Running
. Champaign, IL: Human Kinetics, 2001.
18. Pugh, L. G. C. E., J. L. Corbett, and R. H. Johnson. Rectal temperatures, weight losses, and sweat rates in marathon running. J. Appl. Physiol.
19. Yaglou, C. P., and D. Minard. Control of heat casualties at military training centers. Arch. Ind. Health.