«Sef van den Elshout October 2014 AQI & AQ Communication D2.3/v07 PartII Work Package 2, Task 2.2 Deliverable D2.3 Version number V07 Ver Date Author ...»
- The short-term exposure to air pollution is no longer a major problem in most of the developed countries, for the majority of the population. However people with certain medical conditions do experience problems due to short-term exposure. One can question the usefulness of a communication strategy, using an AQI+website aimed at the general public that is mainly relevant to only a small part of the population. One might consider delivering different messages to different target groups, in different ways. A more general AQI (policy/standards based) for the general public + health relevant warnings (text messaging, apps) to those who
really need to be informed/warned. In this way one can also overcome/minimise the paradox in the previous bullet.
- The timely relevance of the message is sometimes also an issue. Behavioural advice is useful when one is warned timely in advance else it is impossible to adapt one’s behaviour. This implies that these kind of AQI-s should be derived from high quality predictive models, preferably one or two days ahead of an event (e.g. forecasts of Day +1, Day +2 or at minimum a nowcast (Day +0)). This should be accompanied by monitoring data with a short averaging time (preferably hourly). This is the only way that timely and relevant information can be delivered. Relying on 24 hour average data, or reporting yesterday’s data (still quite common for AQI-s) is rather useless if providing health and behavioural advice are the main objectives.
- Consistency of the health messages is an issue. Epidemiological knowledge continues to evolve and hence the thinking about safe levels of pollutants evolves as well. Apart from the fact that knowledge evolves it is also, apparently, interpreted in different ways. This can easily be seen by comparing the health advice given at different concentration levels for different AQI schemes. Since the AQI approaches are different it is hard to make this comparison. Hence we compare the concentrations for the first index band where the advice to ‘reduce activity’ is given and for the concentration where the worst AQI category starts. See table 3.
Table 3 shows that the appreciation of the risk of a certain concentration differs considerably between the US, the UK and China. Apart from the lower end PM2.5 concentrations, the US and the Chinese indices agree fairly well. For the interested lay person comparing these indexing schemes this lack of consistency is awkward. It might contribute to a general sense that the provided information is flawed (see also box 1 where, amongst others, the whole principal of indexing is questioned) or in the case of China, that the government is ‘cheating’ while doing the calculation. This problem is discussed in chapter 4 of the Part I document.
The UK recently revised and updated their AQI (COMEAP, 2011) and the document underlying the revision provides a good review of recent knowledge about short-term exposure to air
pollution and health. The US tightened the lower bands of the PM2.5 iAQI early 2013, making them consistent with the lower PM2.5 bands in the UK.1 1.1.2. Sub-types of health and behavioural advice AQI-s
The health and behavioural advice AQI-s can be further subdivided into two2 sub-types:
a) A single pollutant determines the index (the highest iAQI); interaction between pollutants is disregarded. This type is more common and the UK and US indices are prominent examples. Especially the US AQI is influential as it is wholly or partly copied in several other countries. See Annex A1.1 for a description.
b) Indices that take pollutant interaction into account are for example the South African DAPPS (Cairncross et al., 2007) and the Canadian AQI (Stieb et al. 2008). See annex A1.1 for a description. Hong Kong revised their index and is now (since 2014) using an AQHI (air quality and health index) following the Canadian example (Wong et al. 2013). See also section 2.2.
Type 1.a: The AQI-s that don't take interaction into account follow the same principle as almost all AQI-s, for each pollutant a sub-index or individual AQI (iAQI) is calculated.
The pollutant with the highest iAQI at a given time determines the overall index or AQI. The information that is communicated to the public is usually the AQI + the name of the pollutant determining the AQI.
Type 1.b: The indices that take pollutant interaction into account are theoretically better than those that don’t.
Pollutant interactions do occur. However, given all the uncertainty surrounding the exact relation between air quality and health, devising such a complex AQI can be seen as being overly ambitious, suggesting a level of accuracy that in reality can never be met. This is aggravated by the fact that air quality tends to vary within cities or areas, whereas monitoring only takes place at a few locations and often the AQI is provided as one, or a few numbers per city/area. So the seeming precision gained in the exposure calculation is often not matched by the spatial detail. Unless a particular air quality situation exists that makes it necessary to take pollutant interaction into account, I don't recommend this approach. The calculations are complicated which is a disadvantage from the viewpoint of other communication objectives (accountability, transparency). Secondly, if the Canadian approach is taken, dose response relations have to be derived locally: e.g. the Canadian urban population.
The results cannot be easily transferred to places/countries where public health is completely http://www.epa.gov/pm/2012/decfsstandards.pdf (accessed 24-5-2013).
Some authors (e.g. Ruggieri and Plaia 2012) are not happy with the simplification of one pollutant determining the index. Without taking full pollutant interaction into account they propose to add a second characteristic to indices that shows if only one or more pollutants are (almost equally) high. In my view this leads to overly complicated indices, whilst the purpose of the index is to simplify the information.
different and dose response functions are likely to be different.3 For a good discussion of this conceptually interesting approach see Stieb et al (2008).
1.2. Type 2: policy/standards based AQI-s In the previous section a number of issues with health and behavioural advice AQI-s were mentioned. They are an important reason why policy/standards based AQI-s are popular in many countries. These indices avoid the multiple complicated issues mentioned in the previous section and mainly serve as a tool to reduce complex monitoring information into a simple relative measure of the amount of pollution. This can be used for communication purposes such as policy monitoring, for being accountable and for providing access to environmental information and raising awareness.
Also the paradox that air quality is safe from a short-term exposure perspective but unhealthy from a long-term perspective can be avoided, simply by not interpreting pollution in terms of health impacts in the AQI banding. Air quality ranges from good to bad (or pollution from low to high). Which pollution level is good and what is bad is largely arbitrary and can be decided locally. The bands are usually derived from local legislation, policy targets, WHO guidelines, etc.
Sometimes bands are tweaked in such a way that there is a good frequency distribution of the classes over time. Though this might sound odd (and completely arbitrary) it is an approach well worth considering: e.g. if one provides an internet presentation with an AQI that is always the same (same band) the chances that such a website gets repeatedly visited is small. If one wants to use an AQI as a communication tool one has to make sure that it is attractive and informative. In this chapter the policy/standards based indices in general will be called type 2.a, those specifically aimed at raising awareness type2.b.
Many European AQI-s are standard/policy based and they all slightly differ. Sometimes even within a country different indices are used. This is the consequence of regionally different air quality standards (dating from before common EU legislation) or local adaptations due to prevailing conditions. At some point countries have harmonised the use of AQI-s within their country (e.g. the French ATMO forcibly replaced local initiatives). At the EU level no common index exists. A bottom-up initiative to come to an EU index was the CITEAIR project, proposing a common AQI (CAQI). See annex A1.2. Though it is used in several EU projects it is not enforced by the EU. A Chinese example of a policy/standard based AQI is the RAQI used in the Pearl River delta area. See annex A1.2.
Type 2.a indices are less demanding on their delivery time (e.
g. need for forecasting) and communication channel. Though many operate in near-real time, providing yesterday’s AQI is still a good way of informing the public, NGO-s, or policy makers about air quality as well as a way of showing that one is accountable. AQI-s aiming at raising public awareness have to be This also makes the index unsuitable as a tool to compare the air quality in different cities in different countries.
The AQHI as the index is called was not meant for that purpose but it could be a useful side objective of having an index.
made in such a way that they not only report information but also allow the public to understand what drives air pollution. This implies for example separate indices for regional and urban background, for urban background and traffic; a short time resolution or time series to make sure that people can relate what actually happens (e.g. morning rush-hour to concentration developments); etc.
1.3. Issues related to all AQI types In the previous sections I distinguished health and policy based AQI-s. This provided a framework to explain some specific characteristics of the AQI-s from a communication perspective. Despite these differences several issues apply to both AQI types and these are discussed here.
In fact, many air quality policies or standards are ultimately motivated by public health concerns. If looked at in another way the differences between the various AQI-s are not that big and even policy/standard based indices can be used to alert the public in case of high air
pollution. The issues in common are:
1.3.1. The balance between continuity and being up to date It is important to avoid frequent changes to an AQI. All AQI-s need occasional maintenance but this should not happen too often. Especially major changes that lead to a different perception of the air quality (e.g. fundamental changes to the calculation grid) should be minimised.
Reasons for maintenance can be new epidemiological evidence, changed pollution levels, new legislation, etc. However, the index is a communication tool so people get used to its readings, its colours, etc. If changes are made they have to be clearly communicated. Many people don’t bother to understand all the details behind the AQI (and indeed that is why AQI-s are used to communicate complex information!) so if they are used to it but the underlying properties
change, misinformation is a result. Before an update is made one should consider what is worse:
that the AQI doesn’t reflect the latest views or that the public gets confused by the update.
One of the reasons that many European AQI-s were not updated when common EU legislation came into force is precisely this reason. People were used to their AQI and that was more important.
1.3.2. Time averaging Time averaging is mainly an issue for PM. Since most legislation is based on 24-h average concentrations AQI-s tend to work with these daily averages. If hourly updates are provided, they tend to use a moving 24-h average. Daily average information has a number of disadvantages: it is dull and one can hardly relate events in the real world to the concentrations or iAQI or AQI reported; if it is used for alerting people, the message arrives too late and stays too long (see box 3). For the CITEAIR CAQI an hourly calculation grid was developed in addition
to the daily average grid.4 In the recently revised UK AQI calculation another approach is taken:
when concentrations start to rise, a trigger value is introduced. The trigger alerts people that the concentrations are high a few hours before the AQI would reach the same conclusion, avoiding that the alert by the AQI is too late. Though this is an important improvement, it doesn’t resolve the problem of a moving average AQI being too high for too long. The US uses a different approach: their website AIRNOW does present hourly PM AQI maps. They are not a proper moving average but use a weighting that gives more weight to the past 4 hours if they differ considerably from the 12 hours before. This way a surrogate 24-hour average is calculated that avoids essentially lagging 12 hours behind. Deriving a corresponding calculation grid for hourly values appears to be the simplest way to capture rapidly changing situations, making an interesting AQI and providing appropriate (health) alerts.
In the course of 2014 some EMC-s in China started using hourly PM concentrations in their AQI.
We recommended and encourage this. It was a response to comments made by various EMC-s (a.o. Shanghai) that the AQI is sometimes out of sync with what the people see and experience (swift improvement or deterioration of the air quality). The only step left to take is to develop a proper hourly grid for PM to calculate appropriate iAQI-s. Currently they are occasionally far too high (which might undermine the switch to hourly PM data in the iAQI).