Due to the high infection rate of the COVID-19 virus that caused the first pandemic of the 21st century, lockdown measures were implemented in several cities around the world to contain the virus.
In urban areas, the first visible effect was the drastic reduction of traffic and hence a decrease in the levels of traffic-driven regulated pollutants, namely nitrogen dioxide (NO2). This pollutant plays an important role in tropospheric chemistry. Reductions in NO2 have been linked to increases in other pollutants such as ozone (O3) in traffic sites. These two pollutants are identified as key atmospheric pollutants causing various adverse health effects by the World Health Organisations (WHO).
In the short-term (daily levels), no threshold (below which no adverse health effects are expected) is associated with these pollutants. This implies that any reduction in the short-term airborne concentrations of these pollutants would also reduce the air quality health burden.
The atmospheric composition and dynamics are complex with pollutant levels being influenced by a range of physical and chemical processes (for example, meteorology). For this reason, it is difficult to determine the driving factors for changes in air quality. To isolate changes in pollutant concentrations related to the implementation of COVID-19 measures, we normalise the data by determining the meteorology driven component. In order to do this meteorology normalisation, we used a machine learning algorithm (Random Forest) to analyse the relationships between meteorological parameters and the pollutants (NO2 and O3) from all the stations forming part of the national air monitoring network, operated and maintained by the Environment and Resources Authority (ERA).
Based on these relationships the normalised pollutant concentrations were obtained and then analysed to determine breakpoints using structural change analyses whereby differences greater than a threshold symbolise a drastic change typically related to the implementation of air quality interventions.
14 March and 8 May 2020 were identified as breakpoints in the Msida NO2 dataset (refer to Figure), while for Żejtun breakpoints were noted on 14 March and 23 May 2020. 14 March 2020 coincides with the implementation of the closure of schools as lockdown started.
On the other hand, 8 May 2020 refers to the gradual easing of the lockdown measures. The breakpoints indicate that a statistically significant change in the pollutant levels were noticed with the implementation/lifting of the measures.
Overall, we noted reductions in the monthly mean NO2 concentrations of up to 48% between February and June 2020 compared to the same period between 2008 and 2017. In contrast, results suggest increases in monthly mean tropospheric (ground) O3 concentrations of up to 41% in Msida.
This is generally consistent with published literature. For example, Sicard et al. (2020) suggest an average reduction of about 52% in NO2 concentrations in European cities (Nice, Rome, Turin and Valencia) as compared to 2017-2019 while O3 was shown to increase by about 17%.
and from the were invited to submit findings of this study in a special edition entitled “” in the journal Frontiers in Sustainable Cities.
