An attempt to cut through the haze of Air Quality Indices (AQIs), including Singapore’s Pollution Standard Index (PSI)

Recent reports have drawn attention to the poor quality of air in many cities and regions globally. Poor outdoor (ambient) air quality results in around 3 million premature deaths annually, and is particularly a problem in the eastern Mediterranean, Southeast Asia and the Western Pacific (WHO 2016).

Air Quality Indices (AQIs) are communication tools that are now relied upon in many parts of the world to disseminate information on both outdoor air quality (or more accurately the level of one or more pollutants in outdoor air) and the associated risks posed to public health. Generally, reported AQIs relate to average air quality over the previous 24 hours. The idea is that average air quality over that period provides an indication of likely conditions in the near future, and therefore a basis for people to settle on their planned activities for the day ahead.

Estimates of AQI are based on measurements of a limited number of reference (or Criteria) air pollutants, with each pollutant having their own Air Quality Standard (AQS). Most AQI are determined from the concentrations of six Criteria pollutants: sulphur dioxide (SO₂), particulate matter (PM₁₀), fine particulate matter (PM_2.5), nitrogen dioxide (NO₂), carbon monoxide (CO) and ozone (O₃).   Variety exists between countries in the AQI adopted (for a useful review of single- and multiple-pollutant indices, see Plaia & Ruggieri 2011). This can make inter-country comparisons of air quality, expressed through AQI, difficult. Figure 1) shows the effect of applying different AQI methodologies to the same set of air pollution data, in this case for Portugal.

Figure 1: AQI for Portugal on15 December 2015 calculated using the same air quality data but according to the different AQI methodologies in use in Portugal, the USA, China and Germany (from Monteiro et al. 2017).

Some geographic variation is understandable, because the importance of particular pollutants is likely to vary between locations. Levels of heavy metals, for example, may be much less of an issue away from built-up areas, main roads and industrial developments. For this reason, levels of lead (Pb) in the atmosphere are considered in the calculation of AQI in some cities in India (Monteiro et al. 2017).

Similarly people are expected to react differently to the same mix and intensity of pollutants. Indeed, links between the concentration of a pollutant (the range in levels of pollutants between “breakpoints”), particular categories of health warning (moderate, unhealthy, very unhealthy etc) and the associated range of activities advised (unusually sensitive people should reduce prolonged or heavy exertion, sensitive groups should reduce prolonged or heavy exertion etc), often involving the use of bright colours, characterising AQI are less straightforward than they might initially seem. For example, there is likely to be a great deal of variation in sensitivity to air pollutants even among people classed as sensitive. In addition, there are no safe levels of atmospheric concentrations of particulate matter (PM₁₀ and PM_2.5) or ozone (O₃); research has not been able to identify thresholds of these three Criteria pollutants below which adverse effects do not occur (WHO 2006). Furthermore, pollutants do not tend to act in isolation, yet little consideration is given in many AQIs to the effects of air pollutants as part of a mix of chemicals (Bishoi et al. 2009). Concentrations of air pollutants are a form of compositional data. The data are often incomplete, and this can compromise the value of AQIs, especially those based on the concentrations of individual pollutants. Because of this, Jarauta-Bragulat et al. (2016) suggest an alternative way of determining air quality in which AQI is established based on the geometric mean of the concentration measurements of all of the Criteria pollutants.

Cheng et al. (2007) describe the evolution of the current AQI in the US in response to increasing concerns regarding the health effects of air pollution. Originally established by the US EPA as a Pollution Standards Index (PSI) in 1976, the PSI was revised and renamed the Air Quality Index (AQI), and subsequently implemented, in 1999. The US EPA PSI ranges from 0–500, with 100 equal to the National Ambient Air Quality Standards (NAAQS), and was calculated for every pollutant that at the time had a NAAQS. The AQI also includes breakpoints for O₃ and fine particulate matter (PM_2.5), and divided the original index range 101–200 into separate categories (101–150 and 151– 200). These modifications aimed to strengthen the health information provided. Both the PSI and subsequently the AQI are relevant for a given time and location, and equate to levels of the pollutant that most exceeds its NAAQS (known as the Primary Pollutant). As a result, they may under-estimate the true health effects of air pollution.

Table 1 (upper & lower): US EPA AQI breakpoints and corresponding colour codes (from Jarauta-Bragulat et al. 2016).

China has been monitoring ambient air quality since the 1980s (Chen et al. 2016), i.e. far longer than many nations. A simplified (compared with present day) Air Pollution Index (API) has been reported on a weekly basis since 1998, with additional pollutants (including PM₁₀) added in 2000. In 2012 the API in China was expanded again, to include CO, O₃ and PM_2.5. In January 2013 China also started reporting air pollution levels according to AQI, with the latter based on new NAAQS (Table 2), making comparison between the API and the AQI difficult. As with the US EPA AQI, China’s AQI is not based on the full suite of Criteria pollutants measured, but on the Primary Pollutant, or in this case the pollutant that most exceeds its equivalent to an IAQI = 50 (values in the top row of data in Table 2). Days where pollution levels do not exceed AQI = 100 are known as “attainment days”. The number of attainment days (or “blue sky” days in Beijing) is regarded as a key indicator of urban air quality in China (Chen et al. 2016).

Table 2: Concentration limits (breakpoints) for AQI calculation in China (from Chen et al. 2016)

Public awareness of air pollution, and in particular the potential detrimental health impacts of poor air quality, has also risen dramatically in Singapore in recent years, in line with much of the rest of the world. While the National Environment Agency (NEA) in Singapore has long collected air quality data, these data have only relatively recently been made available to the general public. Singapore’s Pollution Standard Index (PSI) is based on the US EPA PSI (later AQI). The Singapore PSI runs from 0-500, and involves the measurement of concentrations of six Criteria pollutants: SO₂, PM₁₀, PM_2.5, NO₂, CO and O₃. Fine particulate matter (PM_2.5) was only considered as part of the PSI from August 2014, although according to Velasco & Rastan (2015), PM_2.5 measurements – collected at one-hour intervals – were measured from long before that date.

Determining Singapore’s PSI involves the same process as the US EPA AQI and China API/AQI in that the actual measured concentration of each of the six Criteria pollutants are used to generate a sub-index value for that pollutant, based on the breakpoints shown in Figure 2 and Table 3 (lower). As with the US and Chinese systems, the PSI is not based on the full suite of Criteria pollutants measured, but on the Primary Pollutant. Moreover, during major haze events (as determined by the NEA), the Primary Pollutant is PM_2.5.

Table 3 (upper & lower): NEA Singapore PSI breakpoints and corresponding colour codes (from NEA website. Detailed information on breakpoints available here)

Historical PSI data are available for Singapore from 1 January 2010 (reading once per day at 4pm, then 3x per day from 24 August 2012 (8 am, 12 noon & 4pm)), but only for one location on the island.  Since 20 June 2013 data have reported hourly for five locations in Singapore and are available via here. Since April 2014, PSI determinations have potentially included PM_2.5. One consequence of the current arrangement is that it is difficult to carry out analyses of long time series of past air pollution measurements in Singapore, or to compare long term variations in air quality with neighbouring countries or indeed with other nations in Asia and farther afield. This creates problems for the analysis of, for example, the impacts of past severe deviations in air quality, and therefore our ability to anticipate the effects of projected future variations.

Figure 2: Graphical representation of the relationship between AQI and concentrations of four Criteria pollutants, based on the US EPA method. Note (1) circles = break points and (2) a linear relationship is assumed between break points.

A common source of confusion in Singapore is why the PSI for the country reported by the NEA is often different from AQIs also reported for SIngapore by other organisations, such as PM haze, IQAir (a company that supplies air filter equipment!) and aqicn? Part of the reason is that some organisations (e.g. IQAir, aqicn) also make use of data from sources that may not be fully reliable (e.g. private, off-the-shelf pollution monitoring equipment that may not be properly maintained/calibrated). The main reason is, however, that the PSI is reported by the NEA on the basis of a 24-hour moving average (average air quality), whereas other organisations report AQIs based on hourly-averaged or single point data. Reporting the PSI in terms of average air quality smooths out short-lived increases in pollution levels and provides a measure that is comparable with World Health Organisation (WHO) air quality guideline values. For example, according to WHO (2006), annual and 24-hour mean values for PM_2.5 should not exceed, respectively, 10 and 25 ug m^–3 . However, those same short-lived excursions in air pollution levels can result in negative health effects for some people, especially those suffering from existing respiratory and cardio-vascular ailments.

Pollution in urban areas is known to vary greatly not only through time but also spatially. Urban pollution has a geography, with some areas (e.g. bus stops, major road junctions) associated with far higher levels of air pollutants known to be detrimental to human health than others. The current NEA network of five monitoring stations that provides estimates of PSI for the general public is unlikely to capture this spatial variability. This might become a cause of greater concern once confusion over how AQIs/PSIs are arrived at is resolved. A recent study during a haze-free period determined PM_2.5 concentrations in roadside locations in Singapore (28.88 ± 5.92 ug m-3) well above both annual and 24-hour mean WHO guideline values (Zhang et al. 2017). Prolonged exposure to PM_2.5 concentrations around 30 ug m-3 is similar, in health impact terms, to smoking passively around 15 cigarettes per day, according to research carried out in the Netherlands (van der Zee et al. 2016).

Recent developments in air pollution monitoring and mobile communications technology could provide an answer – and help shift responsibility for monitoring air quality from government and private organisatons to the general public. For example, bicycle and scooter schemes are common in many cities, Bikes and scooters could be equipped with equipment that can be used to monitor pollutants that are representative of other, more difficult to measure pollutants, with measurements relayed to members of the general public through their mobile phones. Thus equipped, members of the general public could then use the information to plan their journeys to work/school, recreational activities etc avoiding the most polluted routes. This is not as far-fetched as it seems – see for example Appmosfera. Easy access to high spatial and temporal resolution AQI information is likely to be an important part of ensuring livability in our smart cities of the future.

References

Bishoi et al. (2009) A Comparative Study of Air Quality Index Based on Factor Analysis and US-EPA Methods for an Urban Environment. Aerosol and Air Quality Research 9: 1-17

Chen, W. et al. (2016) Urban air quality evaluations under two versions of the national ambient air quality standards of China. Atmospheric Pollution Research 7: 49-57

Cheng, W.L. et al. (2007) Comparison of the Revised Air Quality Index with the PSI and AQI indices. Science of the Total Environment. 382: 191-198

Jarauta-Bragulat, E. et al. (2016) Air Quality Index revisited from a compositional point of view. Mathematical Geosciences 48: 581-593

Monteiro, A. et al. (2017) Towards an improved air quality index. Air Quality, Atmosphere & Health 10: 447-455.

Plaia, A. & Ruggieri, M. (2011) Air quality indices: a review. Reviews in Environmental Science and Bio-technology 10: 165-179

Van der Zee, S.C. et al. (2016) Air pollution in perspective: health risks of air pollution expressed in equivalent numbers of passively smoked cigarettes. Environmental Research 148: 475-483

Velasco, E. & Rastan, S. (2015) Air quality in Singapore during the 2013 smoke-haze episode over the Strait of Malacca: lessons learned. Sustainable Cities and Society 17: 122-131

WHO (2006) WHO Air quality guidelines for particulate matter, ozone, nitrogen dioxide and sulfur dioxide. Global update 2005. Summary of risk assessment. WHO Press, Geneva, Switzerland, 22 pp

WHO (2016) Ambient air pollution: A global assessment of exposure and burden of disease. WHO Press, Geneva, Switzerland, 131 pp

Zhang, Z-H. et al. (2017) Characterization of traffic-related ambient fine particulate matter (PM2.5) in an Asian city: Environmental and health implications. Atmospheric Environment 161: 132-143