Disclaimer
The information, opinions, data, statements and recommendations contained in this report have been gathered, analysed and presented by a team of professionals with diverse backgrounds, all selected free from any discrimination based on race, colour, ancestry, national origin, ethnic group, identification, age, religion, marital or parental status,
physical or mental disability, gender, sexual orientation or genetic information.
Mobito and the authors of this report acknowledge the limited dataset, as well as the fact that part of the data contained has been retrieved from external sources, hence assume no legal liability for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed herein and do not represent that use of such information, apparatus, product or process would not infringe on privacy and proprietary rights.
This report and its content is the property of MobitoTM 2020 and are thus copyright-protected. All rights reserved.
Any redistribution or reproduction of part or all of the contents in any form is prohibited, other than the following:
You may print or download to a local hard drive extracts for your personal and non-commercial use only
You may copy the content to individual third parties for their personal use, but only if you acknowledge MobitoTM as the author of this material
You may not, except with our express written permission, distribute or commercially exploit the content of this report. Nor may you transmit it or store it in any other website or other form of electronic retrieval system.
Executive summary
Air pollution in urban areas is a long-standing problem that affects citizen’s lives and
highly concerns governments and policymakers across the globe. According to WHO,
ambient (outdoor) pollution is responsible for 4.2M deaths, annually[1]. Consequently,
legislators have unanimously prioritized the reduction of air pollution levels, placing it at
the top of their agendas. This is also evident in the UN's 17 Sustainable Development Goals
(SDGs) towards 2030. 3 out of the 17 goals (#3 Good Health & Wellbeing, #11 Sustainable
Cities and Communities, #13 Climate Action) involve, directly or indirectly, combating air
pollution in large urban areas[2]. Accordingly, the EU, in its effort to lead a coordinated and
more impactful response to the problem, has established certain health based stan-
dards and objectives for urban centers.
One of the fundamental reasons why air pollution levels are so high in cities is vehicle
density, commonly known as traffic. Problem seems to magnify especially when we con-
sider future projections which indicate that the global population is increasing and that
the urbanization trend will continue to dominate in the future[3] (source: BCG. Realizing the
Future of Mobility.) At Mobito, we help cities improve citizens' well-being by orchestrating
urban mobility activity and leveraging the power of data. The analysis that follows
demonstrates the how.
This report focuses on the city of Athens and compares pollution performance with other
major EU cities. Utilizing reliable air pollution sensors, and focusing on the measurements
of 3 key atmospheric particles (NO2, PM10, PM2), during the period Feb. 2019 - Feb. 2020,
we are offering an analysis of the impact of traffic and weather on levels of urban air
pollution.
The data confirms there is room to improve. Our analysis indicates that Athens is not in
line with air framework regulations. While comparing data extracted with values defined
by specific Air Quality Standards, as these were laid out by the EU, we were able to identify
several cases where Athens exceeded stated standards. In particular, Athens exceeded
the limit of PM 10 daily concentrations, 41 times against 35 permitted per year. The city
also exceeded the NO2 yearly concentration average limit by more than 20%.
Having established the correlation between traffic, air and weather with air pollution and
defined the extent of their effects, we bring forth some recommendations focused on
restrictions in traffic that would assist Athens to actively reduce pollution levels and align
with the EU standards. More specifically, and not-withstanding the limitations of data
used and methodological simplifications, we were able to pinpoint the following focal
points which policymakers ought to address:
For PM10 daily average concentrations:
Decrease traffic by 15% on the 40 days with temperature below 20°C and air humidity above 80% during the period 02/2019 to 02/2020.
This would have prevented PM10 concentration from exceeding the limit of 50 μg/m3 on 7 occasions, bringing the total exceedance events from 41 to 34, below the EU limit of 35.
For NO2 yearly average concentration, one of the following:
Decrease traffic by 36.5% on the 127 days with wind speed < 2 m/s during the period from 02/2019 to 02/2020.
Decrease traffic by 15.5% on all days of the year.
This would have brought the yearly NO2 concentration average down from 50.85 μg/m3 to 40 μg/m3, equal to the limit set by EU standards.
Furthermore, we provide a ranking of a selection of European cities by air quality, using
our own city compliance index and city composite compliance index. Athens ranks 9th
out of 11 ranked cities on both indices but could move up to 3rd by focusing on the reduc-
tion of traffic, as suggested in this report.
Air quality data sources
The air quality data used in this report was collected from the following sources:
European Air Quality Portal by the European Air Quality Agency, provides public archives of hourly measurements of air quality across Europe, including Athens. The sensors monitored are classified in three categories: traffic , industrial and background .
Luftdaten , a citizen science project, provides archives of air quality measurements across all Europe. We categorize these sensors as citizen sensors.
We aggregate air quality data with weather and traffic data:
Weather : we use yearly weather records from rp5.ru , which provides hourly records of weather at Athens' airport.
Traffic : we used the HERE traffic APIto record hourly traffic since December 2019 in the areas around the sensor locations.
Stations locations and compliance to EU standards
We monitor the air quality across different locations, providing good coverage of the city
of Athens.
The map below provides a visual representation of the measurements across stations
as compared to EU limits on air quality. The color-coding indicates extent of compliance
to the yearly average limits of PM10, PM2.5 and NO2 concentrations: red indicates
non-compliance.
The EU standards for air quality are set for hourly, daily and monthly averages of pollut-
ant concentrations (measured in μg/m3), as follows:
Hourly average concentration of NO2 not to exceed 200 μg/m3 more than 18 times over year.
Daily average concentration of PM10 not to exceed 50 μg/m3 more than 35 times over a year.
Yearly average concentrations of PM2.5 not exceed 25 μg/m3, of PM10 not to exceed 40 μg/m3 and of NO2 not to exceed 40 μg/m3.
To safeguard data accuracy of this report, compliance to these standards is assessed
by averaging measurements from data sources used. Emphasis is given to the impact
of traffic on air quality. Therefore, in what follows we restrict our analysis to sensors of the
traffic category, unless specified otherwise. These sensors are installed in a way that
makes them capable to capture traffic-related pollution.
The city compliance index:
Ranges from values 0 to 5. 0 corresponds to high air quality and full compliance and
values greater than or equal to 3 indicate non-compliance, at an increasing extent. The
city compliance index considers all stated European limits on PM2.5, PM10 and NO2 and
provides a score using the worst performance among these pollutants and limits. Values
above 3 indicate that the air quality is non compliant with at least one EU limit.
The city composite compliance index:
Ranges from values 0 to 6. Values below 1 correspond to full compliance, values above 1
indicate non-compliance to one EU limit, above 2 non-compliance to two EU limits and
so on.
For Athens, we find a compliance index of 3.81 and a composite compliance index of
2.81 indicating that Athens is not compliant on both counts.
Comparison to selected cities
The figures below compare Athens' compliance and composite compliance indexes to
a selection of European cities.
Athens' compliance index ranks number 9 out of 11 among these cities.
Athens' composite compliance index ranks number 9 out of 11 among these cities.
Air quality incidents
Let us have a closer look at the incidents leading to non-compliance to air-quality stan-
dards:
The daily limit of 50 μg/m3 for the PM10 density has been surpassed on 41 occasions : 6 more than the allowed limit of 35.
The yearly average density of NO2 is 50.85 μg/m3: more than 10 μg/m3 over the allowed limit of 40 μg/m3.
When did the PM10 daily incidents occur?
The worst three months are January, October and December
Which months drove the NO2 yearly average up?
The worst three months are May, June and October.
Driving factors
In order to be able to ensure the effectiveness of our policy recommendations regarding
the improvement of air quality, which would eventually elevate Athen’s position in the
corresponding standings and most importantly improve its citizens' well-being, we need
first to understand the factors which drive air pollution. To this end and considering that
sufficient quantity of data is one of the determinant factors affecting the relevance of
the results, we utilize data across all types of stations. It is thus assumed that the results
generated apply also to the subset of Traffic station measurements.
In the figure below, we illustrate the correlation between air quality and the following
variables:
Wind direction in degrees of separation from the North.
Wind speed in meters per second.
Feel-like temperature: a measure of temperature combining air temperature, wind speed and air humidity.
Current traffic , measured as a traffic jam assessment occurring at the time of the measurement.
Past traffic , means traffic over the last 12 hours before the air quality measurement.
As the graph illustrates, the three variables with the highest overall correlation to air
quality are past traffic, current traffic and wind speed .
As expected, current and past traffic have an important impact on air quality.
Regarding past traffic, it is interesting to understand in more detail the correlation to air
quality levels as a function of the hours passed after the traffic measurement.
As can be seen in the figure above, the correlation varies as the delay between traffic
and air measurements is increased.
For PM2.5 and PM10 concentrations, we see a peak in correlation at approximately 12
hours, suggesting that the microparticles generated by traffic build-up in the atmo-
sphere.
For NO2, on the other hand, the correlation is greatest with current traffic.
Driving factors
In what follows we identify the most effective traffic policy actions to decrease the
number of air quality incidents and improve the position of the city of Athens in our air
quality ranking.
Avoid PM10 daily average incidents
Our aim is to identify the conditions in which a decrease in traffic would have the biggest
effect in decreasing pollution. In this section, we focus on analysis PM10 and the traffic
decrease needed to reverse the non-compliance to the EU daily average concentration
limits.
As a sanity check, we can confirm from the above graph that high incidents of traffic do
indeed correlate with high atmospheric concentration of PM10. Accordingly, we expect a
decrease in traffic to reduce PM10 pollution levels.
When, and by how much, should we decrease traffic?
Next we look into the relationship between PM10 daily average concentration incidents
and contextual variables such as temperature and air humidity.
The figures above indicate that days where the temperature is lower than 20°C and air
humidity higher than 80% are days with a high likelihood of surpassing the EU daily aver-
age limit and therefore good candidates for action. Let's verify if it is the case:
Indeed, we observe that it is three times more likely for PM10 concentration to exceed the
EU limit on a day with average temperature below 20°C and average air humidity above
80%.
The plot below shows the distribution of such days per month of the year:
There were 40 such days in the last year, with 13 incidents. Avoiding 50% or 7 of them
would have been sufficient to meet the EU targets and bring total incident count below
35; Specifically, this could be achieved by reducing PM10 concentration by 15% on this
category of days. Let us understand by how much we should decrease traffic to achieve
this result. To this end, we analyse the evolution of traffic and PM10 concentrations by
hour:
It is apparent from the above plots that PM10 levels are highly correlated with the evening
and early morning hours. Looking at the traffic distribution, it appears that the morning
peak in pollution occurs as traffic levels rise, although the intensity of the rise in pollution
suggests that other effects than traffic might also be at play.
The evening peak in pollution happens as traffic cools off. It is possible that the elevat-
ed evening levels are the result of traffic-generated pollution building up in the atmosphere during the day.
Assuming that decreasing traffic is an efficient action to decrease PM10 concentrations levels, let us estimate what percentage of decrease in traffic could lead to 15% decrease in PM10 concentrations.
Our linear regression analysis suggests that a decrease of 15% in traffic levels on days with T<20°C and air humidity>80% could bring the PM10 concentration daily levels down by 15%. As a result, number of incidents would fall below the EU limit of 35 incidents per year.
Decrease NO2 concentration yearly average
Analogously to the analysis done for PM10 concentration levels, let's see if we can find
similar insights for NO2 yearly concentration levels.
Athens would need to bring down the yearly average from 50.85 μg/m3 to 40 μg/m3: a
decrease of 22%.
Even more evidently than in the case of PM10 concentration levels, the figure above illus-
trates the strong correlation between traffic and NO2 concentration levels. In this sense,
we are confident that a decrease in traffic is a move in the right direction to achieve our
goal.
Impact of actions suggested for PM10 compliance
Ideally, we would like the traffic response recommendation that we put forward for PM10
to also achieve the sought after result for NO2. This way a single policy action could
achieve both end goals. To this end we analyse the effects to NO2 pollution of: a
decrease traffic by 13.3% on days with T<20°C and air humidity>80%.
Similarly to the PM10 case, a linear regression model suggests that a decrease of 15% of
traffic levels on such days could lead to a decrease of 26.0% of NO2 concentrations
levels over the days of interest which translates to a decrease of 2.73% of the NO2 con-
centration yearly average to 49.46 μg/m3 , still above the limit of 40 μg/m3. In order to
meet the EU target, another 19.1% of decrease of NO2 emissions is necessary.
Identify other days to act against NO2 pollution
Let us identify which days are more likely to have a high level of NO2 concentration:
The figures above suggest that days where the wind speed is less than 2 m/s are good
candidates for action.
There were 127 such days in the past year, and indeed days with low wind speed are
highly correlated with bad air quality. A decrease of 45.7% in NO2 concentration during
such days would be necessary to achieve compliance to EU standards.
To estimate what percentage of decrease in traffic is required to reach this goal, we use
our multi-variate model to predict NO2 concentration levels as a function of traffic,
weather and station location. We find that a decrease of 36.5% in traffic levels around
the locations of the air pollution sensors would lead to a decrease of 45.9% of NO2 levels.
This is an ambitious goal but it would bring Athens in full compliance with EU standards
on air quality.
Of course, decreasing traffic on days with low wind would in turn have an effect on PM10
concentration levels, further improving the city's air quality. We would need additional
traffic data to analyse the relation between traffic and PM10 concentrations on days with
low wind, temperature greater than 20°C and air humidity lower than 80%.
Decrease traffic on all days of the year
Alternatively, the traffic response could be applied to all days of the year. If we were to
look at this time period, and under the same assumptions, we would need to achieve a
19.1% decrease in NO2 concentration levels in order to fall below permitted limits.
Conclusions
As mentioned earlier the aim of this report is not to convey strict guidelines to national
governments and local authorities on how to solve the air pollution problem. Rather, its
purpose is to reveal the key factors that affect a city’s air pollution levels, identify and
validate their link, and suggest intermediate steps or associated KPIs which would simply
indicate the direction of a more targeted and successful policy.
Based on the amount of data at our disposal and the already stated assumptions, as
well as considering compliance with EU standards on air quality, we have concluded that
regarding Athens, Greek officials should:
Decrease traffic by 15% on days with T<20°C and air humidity<80% to meet the requirement on daily averages of PM10 concentration to decrease daily average incidents to 34.
Either one of the following:
Decrease traffic by another 36.5% on days with wind speed < 2 m/s to meet the requirement on NO2.
Decrease traffic by 15.5% on all days of the year.
This would decrease Athens compliance index to just below 3, say 2.99, improving
Athens's ranking from 9th position to 3rd !
It would also decrease Athens' composite compliance index to 0.71, improving Athens's
ranking from 9th position to 3rd !
As noted this study is meant to showcase the ability of Cities to use pollution sensor
measurements to guide policy actions and improve citizen's well being, while staying
compliant to EU stated pollution limits. Our study is largely meant to be indicative and
can be largely improved by incorporating more data. As a next step, we are keen to
automate this process and offer it as a tool to Cities that are interested to take control of
their City data and enact the right policies for a more sustainable future.
Sources:
[3] source : “BCG. Realizing the Future of Mobility.”
Comments