Is COVID-19 Lockdown Cleaning the Skies Over Milan?
How COVID-19 controls are impacting air pollution levels in Lombardy
It’s been more than one month since the Italian government adopted the most sweeping measures outside China to face the Coronavirus outbreak, locking down the most populous, richest and most productive region of the Bel Paese: Lombardy.
Consequential travel bans and the shutdown of all non-essential businesses and industries had a huge impact on Italy’s fragile economy, but also gave researchers a unique chance to collect data and study the environment we live in under never-seen-before conditions.
As a matter of fact, blocking most of our daily activities almost canceled out the human component from the pollution equation and allowed us to study how the environment reacts when the human contribution becomes neglectable, in a way that would simply be impossible to consider under normal circumstances.
Do our daily commutings, activities and habits have an impact on the quality of the air we breathe?
Everyone who lives in Lombardy could easily answer this question just by looking at the bright sky and enjoying the clear, starry nights as never before. As men of science, though, we seek data to support our empirical experience, trying to explain a phenomenon that remains much more complex than what we see with our naked eye.
Air Pollution in Lombardy
Air pollution: what is it and why do we care?
We refer to air pollution as a heterogeneous mixture of chemical components in the air. When we talk about pollutants we can refer to gas compounds—like Carbon Dioxide and Nitrogen Dioxide — as well as atmospheric aerosol particles, groups of solid and liquid matter suspended in the air with different diameters like PM10 or PM2.5.
Long-term health effects from breathing polluted air include heart disease, lung cancer, and respiratory diseases such as emphysema. Even in lower concentrations, these chemicals have toxic characteristics that may result in health effects from prolonged exposure. PM2.5 particulates, for instance, are so small they can infiltrate lungs in 30 minutes time, causing severe breathing problems.
According to the WHO, 4.2 million people die every year as a result of exposure to ambient air pollution, while 91% of the world’s population lives in places where air quality exceeds guideline limits.
Breathing poison: why the Po Valley is one of the most polluted places in Europe
Given the deadly poisonous nature of these air pollutants, the European Union has developed an extensive body of legislation which establishes health-based standards and objectives for a number of pollutants present in the air.
In 2019, Milan exceeded the legal limit for the whole year before the end of February. At the end of the same year, the city recorded 135 days (37% of the entire 2019) with pollution levels far above the guideline threshold.
Milan shares the bad scores with an endless list of cities: Venice, Turin, Brescia, Lodi, Bergamo, Pavia, Monza, etc.
All these cities are located in the same geographical spot: the Po Valley. As of today, this is considered the worst area in Europe for air quality.
The Po Valley or Pianura Padana is known as the most industrialized part of Italy, with the main factories located in what was known as the Industry Triangle between the cities of Turin, Milan, and Genoa. The majority of the fields in the Po Valley are cultivated, both for human consumption and to feed animals under intensive breeding.
The big stain of air pollution over the Po Valley is indeed connected to factories, cars and intensive farming, but is also the consequence of a very unlucky geographical conformation. In this flatland, surrounded by the Alps, the wind is rare and prolonged episodes of climatic inversion are frequent. This causes emissions to stay still on the valley instead of being blown away.
Air Pollution and Lockdowns: Preliminary Analysis
Questions
Now that we introduced the main aspects of air pollution and the characteristics of our reference environment, we can use this knowledge to collect proper data and try to answer some questions.
- Is the air quality changed after the lockdown?
- How did it change? Which pollutant changed the most?
- Why did it change? Do data suggest a reasonable guess for this behavior?
The Data
To answer these questions we rely upon the Regional Environmental Protection Agency of Lombardy (ARPA). ARPA has ground-level sensors that constantly monitor several weather parameters like temperature, wind intensity, precipitation, as well as important air pollution indicators like the aforementioned particulate matter. Data can be asked through this form.
Pollutants are monitored in different locations, multiple times over the same day. The dataset we built averages these observations to get a single value for each day, from January 1st of 2018. The analysis performed is therefore not spatial and doesn’t take into account night-day variations.
Here is the summary of collected features:
- Year: the year when the data has been collected (2018,2019,2020)
- Day: the day of the measurement (1–365)
- Pollutants: the daily concentration of pollutants (µg/m^3)
- Wind: daily average intensity of wind (m/s)
- Rain: amount of precipitations over the day (mm)
- Temperature: average temperature of the day (°C)
- Lockdown: the day is a day of lockdown (boolean)
- LockCount: amount of days since the beginning of quarantine (0-ongoing)
Pollutants monitored are Nitrogen Dioxide (NO2), Carbon Monoxide (CO), PM10, PM2.5, Ozone (O3), Benzene, Ammonia (NH3), and Sulfur Dioxide (SO2).
The harmonized dataset can be found here:
1) Is the air quality changed after the lockdown?
Before checking the data, let’s have a look at the satellite images coming from the European Space Agency:
ESA’s Sentinel-5 highlights a clear descending trend of Nitrogen Dioxide’s level in the troposphere. This is particularly evident in the Po Valley, but affects also Spain and Central Europe. Similar behavior has been shown in Hong Kong, Bejing, Shangai and Wuhan.
This already suggests that some changes are definitely going on, in an area that is far too wide to be explained with just local climatic factors. However, these variations might still be due to seasonal trends.
And that’s why we need to check the data.
Let’s look at the NO2 trend in Lombardy over the past three years, for instance.
As we can see from the size of the area highlighted in yellow, the NO2 drop doesn’t appear to be just seasonal. The concentration of Nitrogen Dioxide measured during the 2020 quarantine is lower than what we experienced in the same period of 2019 and 2018.
Performing an ANOVA test over this time windows shows that there is indeed a significant difference between the three groups, while the Tukey’s test for multiple comparisons of means states that this difference is meaningful comparing 2018–2020 and 2019–2020, while it is not for 2018–2019.
The air quality, then, did change during the lockdown.
2) How did the air change?
Proving that the concentration of nitrogen dioxide we breathe now is different from what we breathed before doesn’t necessarily imply that the air pollution dropped because of the lockdown, neither that it happened for all the pollutants involved.
Let’s test the other pollutants then. A p-value less than the significance level (0.05) can be found also for Carbon Monoxide, Ozone, Benzene, Ammonia and Sulfur Dioxide. However, Tukey’s test on these components shows that only Benzene and Sulfur Dioxide has a significant drop in 2020.
PM10 and PM2.5 both tested negative with ANOVA, so the guess here is that they didn’t change much over the years (statistically speaking: we can’t reject the test’s null hypothesis, therefore it is not safe to say that they change over the years).
3) Why did the air change?
Now that we know that some pollutants experienced an unprecedented drop during the lockdown — while others didn’t—it goes without saying that it would be nice to understand why this happened.
This is indeed a trivial question. Correlation doesn’t imply causality, and there are so many factors to be included that it’s really hard to assess a direct cause-effect path on such drops. Causality also requires a deep understanding of the subject that simply can’t be reached here.
We can still use correlation and literature to make a reasonable guess, though, excluding as many pressure factors as we can.
Analyses of variance tend to exclude seasonal trends as the main cause of these air pollution drops, but they might still be due to local climatic factors. The correlation table clearly shows that pollutants have a significant inverse dependency with wind, temperature and — for PM10 — precipitations. However, running some tests over February and March for 2018,2019 and 2020 shows that meteorological conditions have been pretty much the same over these years.
Given that we didn’t experience important local climatic deviations, what we have left is the anthropogenic activity. Human activities widely contribute to air, land and water pollution in many different ways.
Conversely from what we saw for the climatic factors, the vast majority of these intensive anthropogenic activities faced a historical drop after the national shutdown. Banning commuting means fewer vehicles around, which means fewer nitrogen oxides, particulate matter, carbon monoxide and benzene. Blocking factories means less industrial processes and less energy consumption (and again, less road and non-road transport).
Among all the concauses we analyzed here, the intensity of human activity is the only parameter that changed terrifically compared with the previous years. This is a direct consequence of lockdown and allows us to move a reasonable guess about the positive impact of the quarantine on the overall quality of the air we breathe.
Assuming lockdown as the main cause behind the air quality improvement, we still have to find a reasonable explanation for those pollutants which concentrations didn’t change much with the beginning of quarantine. This can be motivated considering that different human activities emit different pollutants, and not all the activities have been affected by the lockdown in the same way.
Let’s consider particulate matter, for instance. Here, residential and commercial combustion lead an important role in emissions. Shutting down the industry cut commercial component but raised the residential one, as people now spend much of their time at home. This might have caused an emissions shift from commercial to residential sources that kept PM10 and PM2.5 locked around the average despite the lockdown.
Another relevant example is NH3. Agriculture is the largest source of ammonia emissions by far: more than 90% of NH3 emissions in Europe originate from animal farming, manure processes and synthetic fertilizers. Agricultural activities haven’t suffered the lockdown as badly as the industrial ones, and this might explain why ammonia hasn’t dropped so much.
Conclusions
Studying ongoing phenomena in a complex environment is never an easy task. Nevertheless, statistics provide clever instruments helping us in formulating reasonable guesses to be validated or rejected in the future.
In this article we focused our guesses on air quality in Lombardy, showing how it significantly improved in comparison with the same period of the previous years.
While an in-year drop in air pollution can be associated with seasonal phenomena, a statistically significant reduction over the year needs further explanation. Since we didn’t find relevant variations in local climatic factors, the only thing that appears to be changed drastically is human activity, which has been greatly limited in the last month due to the imposed COVID-19 lockdown. Even though, not all the activities have been shut down completely and some pollutants like PM10 and NH3 still remain at a high concentration level.
We hence stated that the COVID-19 lockdown might be in fact the reason why the air over Lombardy looks now clean as never before.
***UPDATE***
ARPA Lombardia published the results of a preliminary survey on the variation of pressure factors and the trend of air quality data generated by the restrictive measures introduced to combat the spread of the virus. This survey seems to confirm the link between improved overall air quality and national lockdown.
References
- WHO on Air Pollution
- ARPA Lombardia e Piemonte
- Legambiente
- Common Air Pollutants and Their Health Effects [NSW Government, 2013]
- NIH on Air Pollution
- Medlineplus on Air Pollution
- Air Pollution Over the Plain of Lombardy [NASA, 2000]
- Nitrogen Dioxide and Fine Particles Are Threatening Po Valley Air Quality [Vitesy, 2019]
- COVID-19: nitrogen dioxide over China [ESA, 2020]
- Anthropogenic Air Pollution Sources [F.Popescu and I.Ionel, 2010]
- Health Matters: Air ollution [UK Government, 2018]
- Agriculture, Ammonia, and Air Pollution [IASS fact sheet, 2016]
Note from the editors: Towards Data Science is a Medium publication primarily based on the study of data science and machine learning. We are not health professionals or epidemiologists, and the opinions of this article should not be interpreted as professional advice. To learn more about the coronavirus pandemic, you can click here.
