What is the actual impact of the COVID-19 lockdown upon air pollution in Spain?

11 May 2020

Reductions in air pollutants resulting from the lockdown are typically reported in comparison to the levels attained in previous years, which neglects the confounding effect of meteorological variability.

To calculate the actual nitrogen dioxide (NO2) reductions, BSC researchers have used artificial intelligence methods to estimate the NO2 levels that would have been attained in Spanish urban areas in the absence of the COVID-19 lockdown.

From March 14 to 29, the lockdown reduced NO2 levels by about 40%, and from March 30 to April 23, by 55% over urban areas on average.

In Madrid and Barcelona, the reduction was much more noticeable near trafficked areas (56% and 57%, respectively) than in urban background areas (40% and 43%, respectively)

Air pollution is strongly influenced by meteorological conditions. For example, nitrogen dioxide (NO2) levels can significantly increase under anticyclonic conditions. Because meteorological conditions can vary significantly from one year to the next, the quantification of air pollution reductions resulting from the lockdown can be significantly biased when mean pollution levels observed in previous years are taken as the reference. This is particularly true when analyzing the changes of air pollution at specific locations and short time periods (from days to weeks).

BSC has used artificial intelligence methods to account for the confounding effects of meteorology when quantifying the reductions. BSC researchers have trained machine learning models using past NO2 observations and meteorological data to predict the levels of NO2 that Spanish citizens would have experienced in the absence of the COVID-19 lockdown. The difference between these predictions and the actually observed ones provides a reliable quantification of the reductions due to the lockdown.

The results show that on average over all Spanish provinces, the lockdown reduced NO2 pollution by 40% in urban areas during its first phase, from March 14 to March 29. During the second phase, between March 30 and April 9, when the restrictions became more severe, the reductions reached up to about 57%. After this date and at least until April 23, when non-essential economic sectors were allowed to be active again, reductions remained at the same level. Although all provinces experienced a strong reduction of NO2 concentrations, geographical disparities are noticeable, with generally stronger reductions in the southern half of Spain.

Madrid and Barcelona

The most populated areas register large differences between stations located near traffic and those located in urban background areas. Two key examples of these differences are Madrid and Barcelona, where the reduction in NO2 pollution was much more noticeable near the trafficked streets (56% and 57%, respectively) than in the urban background areas (40% and 43%).

The role of traffic upon NO2 pollution

Road traffic is estimated to contribute between 66 and 69% of NO2 emissions in the urban areas analyzed. The remaining emissions of this pollutant are related to the residential sector (14 to 15%), the industrial sector (8-13%) and port activities (11-3%). NO2 has a short lifetime and a simple chemical composition, which makes it more susceptible to the effects of meteorology than other pollutants. The results of the study are consistent with reported traffic reductions of about 70-80%.

This study has been submitted to the Atmospheric Chemistry and Physics journal. It has been carried out by researchers of the Earth Sciences Department at the Barcelona Supercomputing Center, including Hervé Petetin who is the main author, Dene Bowdalo, Albert Soret, Oriol Jorba, Kim Serradell and the ICREA and AXA Professor Carlos Pérez García-Pando. These results are being used in studies assessing the impact of the COVID-19 lockdown upon health led by researchers from ISGlobal.


Caption: NO2 reduction in stations located in urban areas (squares) and stations located in road arteries (triangles). The size of the symbols is proportional to the mean annual NO2 concentrations. Not enough NO2 data were available in some Spanish provinces and islands (in dark gray).