Turin (Italy) hosted the 2006 Winter Games of the XX Olympic Games from February 10^th to 26^th. One of the main issues related to the organization of Olympic Games is air pollution. More in general, in urban environments, particular social events can play a negative role by increasing air pollutant levels: festivals and concerts, international meetings and sporting events ([1], [2]). The Olympic Games are among the events with the greatest impact ([3], [4], [5]). In fact, whether played or watched, athletic endeavors have the potential to produce huge environmental “footprints” in terms of their use and abuse of natural resources [6]. Ski slopes, for instance, disrupt fragile alpine ecosystems, while snowmobiles spew exhaust fumes into the air. Golf courses sprawl across the land, and consume large amounts of pesticides and water, while parking lots for stadiums and arenas produce vast paved surfaces. In addition, major sports events use energy, emit greenhouse gases, and produce voluminous trash. For example, the 2006 Super Bowl in Detroit produced 500 tons of the greenhouse gas carbon dioxide (from transportation and natural gas and electricity consumption), while the 2004 Summer Olympics in Athens produced half a million tons in two weeks (roughly comparable to what a city of 1 million people would emit over a similar period [6]). Each match during the 2006 World Cup used up to 3 million kilowatt-hours of energy (similar to the annual consumption of 700 European households) and produced an estimated 5–10 tons of trash.

In the available literature, studies usually examine the short-term impacts—that is, in the course of the Games—without considering any effects of the previous organizational phases. For instance, [7] investigated the effects of the measures implemented by the Beijing municipal government (China) to reduce urban motor vehicle emissions only during the 2008 Olympic Games. [5] examined the role of sources outside Beijing in summertime Beijing air pollution levels (that is, earlier than the 2008 Olympic Games) in order to contribute to the regional air quality management studies and new emission control strategies to ensure that the air quality goals for 2008 were met. [8] analyzed the concentrations of benzene and formaldehyde in air in two venues of the 2006 Turin Olympic Games (Turin and Pragelato), aiming to conduct an environmental health survey related to this important event. In this case, the study period was one year before the sport event.

The most attention has been paid to the impacts on human health. In fact, the goal of air quality policies is to protect public health by implementing regulatory measures or providing economic incentives that help to reduce the public’s exposure to atmospheric pollutants. For example, [9] described traffic changes in Atlanta (USA) during the 1996 Summer Olympic Games and concomitant changes in air quality and childhood asthma events. [10] considered the effects on asthma in adults during the 2008 Summer Olympic Games in Beijing. [11] analyzed data from the previous five Olympic Games obtained from athletes seeking to use inhaled β2 adrenoceptor agonists (IBA), in order to identify those athletes with documented asthma and airway hyper-responsiveness (AHR) and the evolution of their illness. In fact, breathing polluted or cold air is considered a significant etiological factor in some but not all sports. [12] described the available evidence of the effect of air quality on performance, especially as it pertains to athletes with asthma.

The present study attempts to examine this important topic in the context not only of the 2-week period of the Turin Winter Olympics but also the 3-year period when all infrastructures needed for the organization were built (2003–2005). For this reason, we compared the in-operam period (i.e. 2003–2005) with the ones before (ante-operam) and after (post-operam): 2000–2002 and 2006–2008, respectively. In particular, we analyzed the atmospheric concentrations of the primary pollutants at ground level [13]: nitrogen dioxide (NO₂), carbon monoxide (CO) and PM₁₀ (breathable particulate matter with an aerodynamic diameter < 10 μm). These molecules and compounds are included in the list of the parameters to be considered for analyzing air quality defined by the European Community in Directive 2008/50/CE. In particular, we chose these three parameters alone, as they are the only ones with uninterrupted data sets for the period examined (9 years). The study was conducted analyzing the information for Turin and Milan. We used the measurements from Milan as a control dataset for comparison with the atmospheric pollution conditions in Turin (venue of the Olympic Games). The comparison is sustainable since these two cities are both large metropolitan areas in northern Italy (Fig 1) and they are similar in terms of air pollutant sources and levels. Moreover, in order to perform a more exhaustive analysis, we compared the pattern of nitrogen dioxide, carbon monoxide and fine particulate matter during the two weeks of the Turin Winter Olympics (from 10 to 26 February 2006) against normal days unaffected by the Games.

The leading sources of nitrogen dioxide (NO₂) are traffic and, to a lesser extent, industry, shipping and households. High nitrogen dioxide levels, combined with ultrafine particles and other oxidants, have become one of the major air pollution problems in urban areas all over the world [13]. Nitrogen oxides are one of the main components of the pollution mix classically referred to as “photochemical smog”. Short-term nitrogen dioxide concentrations may vary considerably within cities and from time to time during the day and night. In addition, average concentrations depend on the distance of the measurement station from main roads [13].

Carbon monoxide (CO) is one of the most common and widely distributed air pollutants. It is a colorless, odorless, and tasteless gas that is poorly soluble in water [14]. The annual global emissions of carbon monoxide into the atmosphere have been estimated to be as high as 2600 million tons, of which about 60% are from human activities and 40% from natural processes [15]. Anthropogenic emissions of carbon monoxide originate mainly from incomplete combustion of carbonaceous materials. The largest proportion of these emissions are produced as internal combustion engine exhaust, especially from motor vehicles with gasoline engines. Other common sources include various industrial processes, power plants using coal, and waste incinerators. Petroleum-derived emissions have greatly increased during the past few decades [16]. Some widespread natural non-biological and biological sources, such as plants, oceans and oxidation of hydrocarbons, give rise to the background concentrations outside urban areas [14].

Particulate matter (PM) in urban and non-urban environments is a complex mixture with components having diverse chemical and physical characteristics. Newer research findings continue to highlight its complexity and dynamic nature: different characteristics (size, other physical features, chemical composition and source) of PM may be relevant to different health effects, and the formation of PM may be either primary or secondary, and then continues to undergo chemical and physical transformation in the atmosphere [13]. For current regulatory purposes, PM has been classified by aerodynamic diameter, as size is critical in determining the likelihood and site of deposition within the respiratory tract. In 1987, the US EPA established a standard for PM at less than 10 μm in aerodynamic diameter (PM₁₀) [17]. PM₁₀ includes those inhalable particles that are sufficiently small to penetrate to the thoracic region. In urban areas, these particles typically contain resuspended dust from roads and industrial activities, and biological material such as pollen grains and bacterial fragments. In addition, they include the earth’s crustal materials such as wind-blown dust from agricultural processes, uncovered soil, unpaved roads or mining operations. Traffic produces road dust and air turbulence that can re-entrain road dust near roadways. Near coasts, evaporation of sea spray can also produce large particles. Coarse particles may also be formed from the release of non-combustible materials in combustion processes, i.e. fly ash.

Global background concentrations of carbon monoxide range between 0.06 and 0.14 mg/m³ (0.05–0.12 ppm). During the early 1980s, there was an annual 1–2% increase in non-urban tropospheric concentrations of carbon monoxide, whereas between 1989 and 1992 concentrations began declining rapidly [14].

An overview of typical annual average concentrations of NO₂ and PM₁₀ in selected cities around the world was presented by [13]. Rural background concentrations of nitrogen dioxide in industrialized countries have been measured at around 15–30 μg/m³. In urban areas, NO₂ concentrations exceed 40 μg/m³ as an annual value (WHO’s 2000 air quality guideline, [14]) in many of the larger cities on all continents. The annual average NO₂ concentrations in major European cities were reported in 2002 to range from 14 μg/m³ in Iceland to 44 μg/m³ in France [18]. Regarding PM₁₀, in urban areas, the average concentrations were 26.3 μm/m³ (annual) and 43.2 μm/m³ (daily), while in rural areas the concentrations were 21.7 μm/m³ and 38.1 μm/m³ (annual and daily, respectively). In Europe, the typical range was 15–60 μm/m³. Focusing on Turin, the mean value is about 60 μm/m³.

Concentration class ranges and frequencies of elementary data for the three selected periods for Turin (ante-operam: 2000-2001-2002; in-operam: 2003-2004-2005; and post-operam: 2006-2007-2008) are shown in Table 1 where statistical information of concentration data are present too. Almost for all data vectors of NO₂ and CO, the valid data are higher than 95% (except for the post-operam NO₂ value at 66%). Even if PM₁₀ features slightly lower values (74–85%), these data can still be considered suitable for all the statistical analyses.

In Fig 2 are shown the histograms of frequencies obtained for each of the three data vectors (2000–2002, 2003–2005 and 2006–2008) of each pollutant. Validated values were divided into 15 increasing classes (see also Table 1). The peak is reached at 44–88 μg/m³, 0.6–1.2 mg/m³ and 44–110 μg/m³ for NO₂, CO and PM₁₀, respectively. As expected, an asymmetric distribution is found for all three pollutants, thus suggesting to correctly consider a lognormal distribution. The statistical parameters of calculated lognormal distributions are shown in Table 2.

We analyzed the atmospheric concentration data for NO₂ and CO (hourly samples) and PM₁₀ (daily samples) over a nine-year period (2000–2008) in the two most important cities of Northern Italy, Turin and Milan. We chose two atmospheric air quality monitoring stations of the same type (i.e. Urban-Traffic station): Turin-Consolata and Milan-Verziere. In this way, we could define how the infrastructural work for the Turin 2006 Olympic Games modified the air pollution trends over time (Turin), compared to a similar urban area not affected by any type of infrastructural work (Milan). Since the infrastructural construction covered a three-year period (2003-2004-2005: in-operam), we decided to compare this period to a time frame of three years before (2000-2001-2002, corresponding to the ante-operam) and three years after (2006-2007-2008, corresponding to the post-operam).

To evaluate the reliability of the lognormal distribution chosen for describing the environmental concentrations of atmospheric pollutants, we verified the “goodness-of-fit” between observed and theoretical data, applying the Chi-Square Test. This test passed for all the data vectors we used in this study, verifying the well-known assumption that pollutant concentrations in environmental media follow a lognormal statistical distribution [24]. We found χ2 (i.e. the sum of the squared differences between the empirical and the theoretical frequencies of each class of values) to be higher than r (i.e. the critical values of Chi-Square table) for each of all the 9 data vectors we used in this study, confirming with a probability of 95% that atmospheric concentrations of NO₂, CO and PM₁₀ follow a lognormal statistical distribution. Finally, the lognormal distribution turns out to be the best fitting statistical distribution for all datasets.

We found out that infrastructural work in Turin affected the NO₂ atmospheric concentrations, determining a peak in the average values during the in-operam period. This pattern did not emerge from Milan data where a decreasing trend was evident from the ante-operam to the post-operam period. This observed decreasing trend is probably due to: i) the progressive rise in the use of less polluting vehicles (the emission factors are 2100 mg/km for Gasoline Pre EURO, 80 mg/km for Gasoline EURO 3 from 2001, 48 mg/km for Gasoline EURO 4 from 2005, and 914 mg/km for Diesel Pre EURO, 803 mg/km for Diesel EURO 3 from 2001, 602 mg/km for Diesel EURO 4 from 2005, (data from ARPA Lombardia); ii) the continuous transformation of civil heating systems by oil, diesel, natural gas (the emission factors are 860 mg/kWh, 200 mg/kWh, and 90 mg/kWh, respectively, from ARPA Lombardia); and iii) the transformation of thermoelectric power stations from the single cycle steam to combined cycle gas (the emission factors are 1400 mg/kWh at 260/160 per combined cycle, from ARPA Lombardia). We can conclude that the Olympic infrastructural projects modified this decreasing trend in Turin determining a peak in average atmospheric concentrations during the working period (2003–2005). For the same reason, the shift of the frequency curves towards lower concentration classes that we observed in Milan (the most frequent class changed from 66–88 μg/m³ in the ante-operam period to 22–44 μg/m³ in the post-operam period), is not present in Turin, where the curves do not show well defined differences among the three periods (in all three periods, the most frequent concentration class is 44–66 g/m³).

CO concentration classes show a similar frequency trend between Turin and Milan. The most frequent concentration class in both the cities is 0.6–1.2 mg/m³ but Turin shows a more distinct decreasing trend related to average values and to frequencies of medium-high concentration classes (2–7 mg/m³).

The generalized reduction in CO levels found in both cities can be linked to the increasing use of catalyzed vehicles (the emission factors are 9 g/km (Pre EURO), 1.5 g/km (EURO 3, from 2001), and <0.6 g/km (Diesel), from ARPA Lombardia). The decreasing trend more evident in Turin compared to Milan can be due to the simultaneous expansion of limited traffic areas created to carry out the infrastructural work.

The effects of the infrastructural work during the in-operam period are quite evident even in the PM₁₀ concentration frequency distribution. While in Turin, the highest value in average concentration is registered in the in-operam period, in Milan we observed a clear decreasing trend throughout the whole period. This general decrease in PM₁₀ atmospheric concentrations may have been determined by new environmental policies that forced industries to adopt the best available technologies (IPPC CE Directives), reducing emissions of primary pollutants (i.e. sulfur oxides and nitrogen oxides), as well as the renewal of vehicles (from ARPA Lombardia). In Milan, the most frequent concentration class is 22–44 μg/m³ in all the three periods, unlike in Turin, where the most frequent concentration class was found to be skewed from 66–88 μg/m³ (ante-operam period) to 44–66 μg/m³ (in-operam period). Medium-high concentration classes (70–160 μg/m³) showed a clear decreasing frequency trend in Milan throughout the whole period (21.3%: ante-operam; 18.8%: in-operam; 16.8%: post-operam) that was not registered in Turin, where the peak frequency of medium-high concentration classes is registered in the in-operam period (59.2%). In the same period, in Turin a peak in the average value of concentrations (97.5 μg/m³) was registered. The general decreasing trend in PM₁₀ concentrations registered in Milan was not observed in the Turin dataset, probably due to the high impact of infrastructural work that determined a higher emission of fine dust from different sources and with different profiles (road traffic for Milan and caterpillar emissions and excavation work for Turin).

From this preliminary analysis, we can conclude that the infrastructural work for the Olympic Games influenced the air quality levels related to NO₂ and PM₁₀ concentrations in Turin. No negative effects were observed regarding the CO dataset. This evidence can be linked to the adoption of effective urban environmental policies from 2000 to 2008 (i.e. less polluting vehicles, higher efficiency civil heating systems, and more road traffic limitations). The negative effects on NO₂ and PM₁₀ levels can be linked to the consideration that during the period of infrastructural work, more constant and higher emission levels occurred in Turin (due to caterpillar tractor emissions) with an emission profile that is different from the classic road traffic emission profile. This generally leads to a reduction of extreme and isolated concentration values and to higher average atmospheric concentrations. Within this contest, the councils of cities that are going to hold the Olympic Games should adopt measures to reduce air pollution during the infrastructural work for the event, i.e. avoiding road traffic in the inner areas or limiting traffic areas. In this way, as the experience of Turin 2006 shows, at least CO levels would be limited. Moreover, traffic represents one of the most relevant sources of emissions even for other pollutants, therefore more attention should be given to this issue, not only during the days of the Olympic event but also during the previous years when all infrastructures needed for the organization would be built.