Temporal air quality (NO₂, O₃, and PM₁₀) changes in urban and rural stations in Catalonia during COVID-19 lockdown: an association with human mobility and satellite data

Abstract

In this study, changes in air quality by NO₂, O₃, and PM₁₀ in Barcelona metropolitan area and other parts of Catalonia during the COVID-19 lockdown with respect to pre-lockdown and to previous years (2018 and 2019) were evaluated. Selected air monitoring stations included 3 urban (Gràcia, Vall d’Hebron, and Granollers), 1 control site (Fabra Observatory), 1 semi-urban (Manlleu), and 3 rural (Begur, Bellver de Cerdanya, and Juneda). NO₂ lockdown levels showed a diminution, which in relative terms was maximum in two rural stations (Bellver de Cerdanya, − 63% and Begur, − 61%), presumably due to lower emissions from the ceasing hotel and ski resort activities during eastern holidays. In absolute terms and from an epidemiologic perspective, decrease in NO₂, also reinforced by the high amount of rainfall registered in April 2020, was more relevant in the urban stations around Barcelona. O₃ levels increased in the transited urban stations (Gràcia, + 42%, and Granollers, + 64%) due to the lower titration effect by NO_x. PM₁₀ lockdown levels decreased, mostly in Gràcia, Vall d’Hebron, and Granollers (− 35, − 39%, and − 39%, respectively) due to traffic depletion (− 90% in Barcelona's transport). Correlation among mobility index in Barcelona (− 100% in retail and recreation) and contamination was positive for NO₂ and PM₁₀ and negative for O₃ (P < 0.001). Satellite images evidenced two hotspots of NO₂ in Spain (Madrid and Barcelona) in April 2018 and 2019 that disappeared in 2020. Overall, the benefits of lockdown on air quality in Catalonia were evidenced with NO₂, O₃ and PM₁₀ levels below WHOAQG values in most of stations opposed to the excess registered in previous years.

Graphical abstract

Supplementary Information

The online version of this article (10.1007/s11356-021-17137-7) contains supplementary material, which is available to authorized users.

Introduction

Monitoring studies of environmental pollution have always been necessary in order to evaluate the impact of air contaminants on human health and the environment. In the last decade, the fast-growing population around the world, especially localized in metropolitan areas, resulted in increments of industrialization, transport demand, and transport flow. Large anthropogenic emissions from these sectors led to various environmental concerns regarding poor-quality outdoor air, altered climate (Ramanathan and Feng 2009; Shakun et al. 2012), and harmful effects on human health (Kim et al. 2015; Anenberg et al. 2019). In the major metropolitan cities, nitrogen dioxide (NO₂), particulate matter (PM_2.5 and PM₁₀), and in minor concentrations, sulfur dioxide (SO₂), carbon monoxide (CO), and ozone (O₃) are among the most hazardous air pollutants (Cohen et al. 2017; Nuvolone et al. 2018).

However, in 2020, the outbreak of COVID-19, which started in China but quickly spread to many countries over the world and eventually turned into a global pandemic, caused a repercussion on the environmental panorama. On the 30th of January 2020, the World Health Organization declared a global health emergency (WHO 2020a; Sohrabi et al. 2020), and shortly after, many major human activities, in the field of transportation, industrial manufacturing, culture, and education, were globally constrained to prevent further spreading of SARS-CoV-2 virus. Pandemic lockdowns started in all parts of the world, generating reduced industrial production and energy consumption, lower road traffic, and therefore lower emissions of pollutants in the atmosphere (Isaifan 2020; Tobías et al. 2020; Mahato et al. 2020). This situation gives a unique opportunity to evaluate the impact of these restricted anthropogenic emissions on air quality. Moreover, air quality assessment might be very useful to better understand the incidence of COVID-19, according to recent studies that report a positive relation between air pollution and SARS-CoV-2 severity (Domingo et al. 2020; Marquès and Domingo 2022). Various studies that evaluate the effects of COVID-19 lockdown on air contamination have already been performed in different parts of the world (Delhi (Srivastava et al. 2020): London (Kumari and Toshniwal 2020), Milan (Altuwayjiri et al. 2020), Lima (Kumari and Toshniwal 2020), Ghaziabad (Lokhandwala and Gautam 2020), Nigeria (Zabbey et al. 2020), Tunisia (Chekir and Ben Salem 2021), Baghdad (Hashim et al. 2021), and Spain (Tobías et al. 2020; Briz-Redón et al. 2021)). All these studies agree on the association between contingency measures and improvement in air quality, but also in clean beaches and environmental noise reduction (Zambrano-Monserrate et al. 2020). Nevertheless, for the Iberian Peninsula, most of the studies are concentrated on the two cities of Madrid and Barcelona (Baldasano 2020; Tobías et al. 2020), and the few ones covering a larger region (e.g., Catalonia (Saez et al. 2020; Marquès et al. 2021)) are focused on the lockdown period. Thus, more in-depth evaluation on the effects of the lockdown on air contamination covering the largest period of time (i.e., pre, during, and post-lockdown), a wider region (i.e., the metropolitan region of Barcelona extended to Catalonia too), and different types of geographical locations (i.e., urban, semi-urban, and rural) will provide new insights on the reasons and social aspects related to such reductions.

Within this context, the objective of this study is to assess the changes in three of the most emitted air pollutants (i.e., NO₂, O₃, and PM₁₀) linked to COVID-19 lockdown restrictions in one of the most populated regions of Spain: Catalonia, including Barcelona and its metropolitan area. Given that the lockdown in Catalonia started on March 14th and lasted until June 21st (Mitjà et al. 2020), with April being the month with the strictest confinement restrictions, in this study, two distinct data analysis approaches were followed. First, a study covering the pre-lockdown, lockdown, and post-lockdown period was performed. Secondly, an exhaustive evaluation of the month of April (the period when the strictest lockdown was imposed) was performed. In both approaches, the data analysis compared the same time periods with the two immediately preceding years 2018 and 2019. Moreover, the analysis of air contamination has been performed in three types of air quality monitoring stations (urban, semi-urban and rural) and a control site, in order to evaluate the influence of geographical location on the monitored air pollution. In addition, the satellite observations of Spain in April 2018, 2019, and 2020 showed the total tropospheric column of NO₂ and O₃ to add more evidences of air pollution changes linked to the COVID-19 lockdown. This study is structured to provide the following information (i) of the overall situation regarding air contamination and COVID-19 lockdown in Barcelona and 5 other parts of Catalonia; (ii) of the air quality stations, the acquired data, and the methods used for their analysis; and (iii) discussion on the causes of air pollution reduction due to lockdowns, offering daily and hourly contaminant profiles and percentages of change of air contamination during lockdown compared to pre-lockdown and to previous years, associated to traffic and social aspects measured with mobility index.

Materials and methods

Lockdown scenario in Barcelona and Catalonia

The first case of COVID-19 disease in Catalonia (NE Spain) was registered on February 25, 2020, and the first death caused by the SARS-CoV-2 virus happened on March 6, 2020 (see Fig. 1). Soon after, the Spanish government declared a state of alarm due to COVID-19 health crisis, which started with the publication of the Royal Decree 463/2020 (Gobierno de España 2020a) on March 14 and imposed the lockdown of all non-essential industries and activities. Stores, hotels, and restaurants were ordered to close, together with shopping and administrative centers. Restrictions on mobility became obligatory, and remote working was imposed whenever possible. These measures became stricter on March 27, time when only essential services were allowed to remain open (Gobierno de España 2020b).

Fig. 1.

Dates of the scenario of COVID-19 health crisis (including start of state of alarm, beginning of lockdown, and de-escalation phases) of Catalonia and Barcelona and its metropolitan area

Transition toward a new normality began on May 4 with the start of the de-escalation. In Catalonia, such de-escalation was gradual and was organized in four different phases: 0 or preparatory phase, 1 or initial phase, 2 or intermediate phase, and 3 or advanced phase. The progress from one phase to another was specific for each region and was determined according to the capacities of the primary healthcare and hospital system, the epidemical situation and the implementation of collective protective measures. For this reason, there was a delay in the start of the different phases in Barcelona and metropolitan area with respect to other parts of Catalonia (see Fig. 1 to know the exact dates for these phases). The end of the four-phase of de-escalation was produced on June 21 with the end of the state of alarm and the beginning of the “new normality.”

NO₂, O₃, and PM₁₀ air pollution

In this study, air pollution by NO₂, O₃ and PM₁₀ was evaluated in Barcelona, its metropolitan area, and other parts of Catalonia. NO₂ is part of the group of nitrogen oxides (NO_x) that includes nitrogen monoxide, nitrous acid, and nitric acid. NO₂ primarily is emitted into the air from the burning of fuel of vehicles (cars, trucks, and buses), power plants, and combustion facilities and off-road equipment. The chemistry of NO₂ is complex (see Fig. 2), since it is subject to extensive further atmospheric transformations to form both ozone and particulate matter; for the latter, NO₂ is the precursor of other organic, NO₃⁻ and SO₄²⁻ particles currently measured as PM_2.5 or PM₁₀. Nitrogen dioxide exerts a range of health effects including effects on lung metabolism, structure, function, inflammation, and host defense against pulmonary infections (WHO 2020b). The current World Health Organization Air Quality Guideline (WHOAQG) (WHO 2020b) reference values for nitrogen dioxide are 40 and 200 µg/m³ for annual and 1-h mean NO₂ concentration, respectively.

Fig. 2.

Reactions among pollutants (including NO₂ and O₃) in the troposphere; daytime versus nighttime. R1, R2, and R3 indicate the main reactions: ozone formation from NO_x and VOCs (R1) and ozone suppression through NO_x titration at daytime (R2) and at nighttime (R3)

Ozone is the primary ingredient of photochemical smog and can produce harmful effects on human health, most of them associated with the respiratory system (i.e., asthma, chest pain, throat irritation, coughing, reduced lung function, and damaged lung tissue (WHO 2020b)). The formation processes of ozone are interdependent and complex, and their reaction and production rates are not linear. The major part of ozone is created by chemical reactions between oxides of nitrogen (NO_x) and volatile organic compounds (VOC) emitted by cars, power plants, industries, and natural (biogenic) sources in the presence of sunlight (hv) (see reaction 1 in Fig. 2). Elevated ozone in polluted regions is usually due to the ozone production with VOC and NO_x during daytime. However, ozone concentrations are depressed through the process of NO_x titration. Such process consists on the removal of O₃ through reaction with NO_x (NO + O₃ → NO₂ + O₂ or NO₂ + O₃ → NO₃ + O₂; see respective reactions 2 and 3 in Fig. 2). The titration process can happen both at daytime and at nighttime, in the immediate vicinity of very large emissions of NO_x. The WHOAQG reference value of ozone is 100 µg/m³ for 8-h mean O₃ concentration.

PM refers to a complex mixture of substances: organic plus elemental carbon, mineral dust (Al and Si oxides, CO₃²⁻, Ca, Mg, P, Fe, K, and some other trace elements), marine aerosols (SO_4marine²⁻, Cl⁻, and Na), and secondary inorganic phases (SO_4non-marine²⁻, NH₄⁺, and NO₃) (Querol et al. 2004). These substances can be in solid or liquid states and exhibit different properties (size distribution, gas–solid/liquid partitioning, toxicity, etc.). Different PM sources exist including building sector (construction, demolition, and domestic heating), traffic (motor emissions and tire, pavement, and brakes abrasion products), industry (high levels of sulfate, nitrate, and other burning products), and natural (i.e., marine aerosols and air masses, especially African dust) (Querol et al. 2004, 2008). In this study, only the fraction of particulate matter that passes through a size selective impactor inlet with a 50% efficiency cut-off at 10-µm aerodynamic diameter (i.e., PM₁₀) has been studied. The current WHOAQG reference values for this pollutant are 20 and 50 µg/m³ for annual and daily PM₁₀ concentration, respectively.

Air monitoring stations

Catalonia is located in the northeast corner of Spain, and its capital city is Barcelona (see Fig. 3). Holding more than seven million inhabitants, Catalonia is the most populated urban area in the Mediterranean coast, and Barcelona and its metropolitan area, with 5.5 million people, is the second-most populated city in Spain and the fifth-most populous urban area in the European Union. The metropolitan area of Barcelona is composed of multiple urban centers that are closely connected to each other but also to other external surrounding areas. Other Catalan provinces studied are Lleida (0.14 million inhabitants) and Girona (0.1 million inhabitants). The main pillars of the Catalan economy are tourism and hospitality industry. Tourism is enhanced by the excellent connections within the Catalan capital, such as the port, a high-speed train station, an international airport, and the second largest trade fair of Europe (Ajuntament de Barcelona 2020a). All of them, but especially the port and airport, involve a high flow of people, goods, and vehicles. The average traffic volume in the main accesses of Barcelona is approximately 1.08 million vehicles per day and constitutes one of the biggest emission sources of air pollutants (Ajuntament de Barcelona 2020a).

Fig. 3.

Map of Spain with a zoom in the region of Catalonia showing the eight air quality monitoring stations used in this study, also with a second zoom in the area of Barcelona

The air quality monitoring network in Barcelona and Catalonia is managed by the Generalitat de Catalunya (Spain) and is made up of 129 automatic remote stations distributed across 15 air quality zones (Gencat). In this study, we analyzed 8 of these stations, the ones containing full information of the pollution (NO₂, O₃, and PM₁₀) and meteorological parameters in the period of time studied, which were located in six distinct air quality zones (see Table 1). Among these eight air quality stations, three were urban (Gràcia, Vall d’Hebron, and Granollers), one semi-urban (Manlleu), and one control site (Fabra Observatory), all of them located in the province of Barcelona, and the remaining three were rural: Juneda and Bellver de Cerdanya in the province of Lleida and Begur (Costa Brava, NE Catalonia), in the province of Girona (Fig. 3). Information regarding the air quality monitoring stations, which includes the number of inhabitants and density, emission source, type of background, geocoordinates, altitude, province they belong to, and the air quality zone (AQZ) to which they correspond, are displayed in Table 1. As can be observed, the source of emission corresponds in all cases to background contamination except for Gràcia station, where traffic is responsible. Furthermore, it is important to highlight the particular features of Fabra Observatory, which is a control site located in the AQZ of Barcelona that shows especial characteristics since it is placed in Collserola Mountain at 415 m of altitude and receives less impact from the city. Such emplacement is useful to provide information of vertical contamination near the city of Barcelona, which complements superficial information registered in Gràcia and Vall d’Hebron.

Table 1.

Information of the air quality monitoring stations

Air quality monitoring station	Type of station	Emission source	Inhabitants	Density (inhabit/km2)	Area (km2)	Latitude	Longitude	Altitude (m)	Province	Air quality zone (AQZ)
Gràcia	Urban	Traffic	120,907	60,300	4.19	41º 23′ 55″ N	2º 9′ 12″ W	57	Barcelona	Area of Barcelona
Vall d’Hebron	Urban	Background	5687	7700	0.74	41º 25′ 33″ N	2º 8′ 52″ W	136	Barcelona	Area of Barcelona
Granollers	Urban	Background	62,419	4073	14.9	41º 35′ 55″ N	2º 17′ 13″ W	133	Barcelona	Vallès Oriental
Observatori Fabra	Control site	Background	–	–	–	41º 25′ 6″ N	2º 7′ 26″ W	415	Barcelona	Area of Barcelona
Manlleu	Semi-urban	Background	20,573	1194	17.2	42º 0′ 11″ N	2º 17′ 14″ W	460	Barcelona	Plana de Vic
Begur	Rural	Background	3925	190	20.7	41º 57′ 31″ N	3º 12′ 46″ W	200	Girona	Empordà
Bellver de Cerdanya	Rural	Background	2138	22	98.14	42º 22′ 5″ N	1º 46′ 36″ W	1060	Lleida	Pirineu Oriental
Juneda	Rural	Background	3475	73	47.3	41º 32′ 38″ N	0º 49′ 47″ W	255	Lleida	Terres de Ponent

Air quality monitoring data

Hourly concentrations of NO₂, O₃, and PM₁₀ were measured from February 15 to August 31 of 2018, 2019, and 2020 in the above-mentioned eight permanent air quality monitoring stations. NO₂ concentrations were measured by means of chemiluminescence according to the UNE method 77,212:1993, using automatically operated MCV 30QL analyzers. Ozone concentrations were measured by means of UV photometry according to ISO FDIS 139464:1998, automatically operated with MCV 48 AUV analyzers. Finally, PM₁₀ concentrations were measured by means of gravimetric determination, using manually operated high volume samplers MCV CAV-A/MS. The generated databases with all the concentrations measured were compiled by the Department of Air Monitoring and Control Service of the Generalitat de Catalunya. In this study, two periods of time were comparatively evaluated for the 3 years (2018, 2019, and 2020). First, the period compressed between the 15th of February and the 31st of August was chosen, as a representation of a period of time including pre-lockdown, lockdown, and post-lockdown and was used to evaluate the correspondence between contamination profiles and anthropogenic activity. Secondly, the month of April was studied in more detail, being this month a representation of a period of time when the strictest lockdown was applied, since the de-escalation process started at the beginning of May (see Fig. 1). The month of April was used to perform an exhaustive analysis of the effects of COVID-19 lockdown on air contamination.

For the analysis of the period of 15th February to 31st August, an initial database containing the experimental data measured every hour during 198 days was processed. Such database gave a vector of size (1 × 4752). A total number of 72 data vectors of this size were obtained, considering 3 years, eight air quality monitoring stations, and the levels of three air pollutants. Folding these 72 long data vectors produce 72 data tables sized (198 × 24), one per year and monitoring station, where the 198 rows correspond to the 198 days monitored, and the 24 columns correspond to the hour times when the contaminants were measured every day. To make all these data tables have the same size, the extra day of February 2020 was removed. All these data tables were used to evaluate the time trends of the three air pollutants during the pre, lockdown, and post-lockdown periods, to determine the differences among periods by calculating the percentages of change, to define the specific profiles in each location associated to the activities carried out in each area, and to study their correlation with the mobility index by calculating the Pearson’s correlation coefficients.

For the analysis of the month of April, an initial database containing the experimental data measured every hour during 30 days gave a vector of size (1 × 720). A total number of 72 data vectors of this size were obtained, considering 3 years, eight air quality monitoring stations, and three contaminants (3 × 8 × 3). Folding these 72 long data vectors produce 72 data tables sized (30 × 24), one per year and monitoring station, where the 30 rows correspond to the 30 days of April, and the 24 columns correspond to the hour times when the contaminants were measured every day. These data tables were used to perform a comparative analysis of the levels of the three pollutants in April 2020 with respect to the levels registered in April of the two immediately preceding years (i.e., 2018 and 2019). Only these 2 years were selected since they include the time when the implementation of important measures to improve air quality in the city of Barcelona started; introduction of shared transport systems and a cycling infrastructure; reduction of space for vehicles; and implementation of the low emission zones (LEZs (AMB)), which was first implemented on December 31, 2017 and finally put into permanent effect on January 1, 2020. Thus, the composition of the vehicle fleet and the mobility structure of Barcelona in 2018 and 2019 significantly differ to that of previous years and make those past years less comparable in the present study.

Raw data vectors containing all meteorological data are given in excel data files as Supplementary material to this manuscript. In all cases, occasional day missing data were replaced by the mean (average) of all the values in the same column of the data table.

Space observations data

Satellites in space provide global observation data for air quality monitoring over the Earth. For this study, data from satellite measurements of background tropospheric NO₂ and O₃ concentrations for the region of the Iberian Peninsula supplied by S-5P/TROPOMI-ESA were used. The Sentinel-5P mission, launched by the European Space Agency in 2017, is a low-orbit polar satellite used to monitor Earth’s atmosphere with a high spatiotemporal resolution using the TROPOMI. Concretely, it is a multispectral sensor that registers reflectance values at ultraviolet–visible (250–500 nm), near-infrared (675–775 nm), and short-wave infrared (2305–2385 nm) wavelengths which measure concentrations of key atmospheric constituents such as O₃, CH₄, CO, SO₂, CH₂O, NO₂, and aerosol properties (Veefkind et al. 2012). In this study, the measurements of Sentinel-5P were used to collect and plot data of NO₂ and O₃ from April 2018 and 2019 (under no pandemic) versus April 2020 (during the pandemic, under the strictest lockdown restrictions). To do that, satellite data were downloaded from https://scihub.copernicus.eu/ and further analyzed using Panoply software (NASA GISS).

Anthropogenic activity: mobility index

The mobility index is a parameter calculated by Google and provided in COVID-19 Community Mobility Reports (Google reports) that indicate the change in daily human mobility (including number of visits and length of stay at different places) from the start of the lockdown restrictions (February 15) until present, with respect to a referential value. Such referential value corresponds to the mean value calculated for the five weeks (January 3 to February 6 of 2020) previous to the lockdown. The MI is provided for six categories, the ones that are useful to indicate social distancing efforts together with access to essential services. These categories comprehend grocery and pharmacy (including grocery markets, food warehouses, farmer markets, specialty food shops, drug stores, and pharmacies), parks (including local and national parks, public beaches, marinas, dog parks, and public gardens), transit stations (including subway, bus, and trains stations), retail and recreation (containing restaurants, cafes, shopping centers, theme parks, museums, libraries, and movie theaters), residential (i.e., places of residence), and workplaces (i.e., places of work). The Community Mobility Datasets were created to be helpful to control the lockdown restrictions imposed by the governments and were constructed with anonymized sets of data from users who had turned on the Location History for their Google Accounts. Therefore, the overall MI data represents a sample of Google users at each part of the world and may not represent the exact behavior of all the population.

In this study, data of MI were tracked from Google (Google reports) for the region of Catalonia (Spain), covering the period of time compressed among February 15 and August 31, 2020, with the aim of better understanding the anthropogenic changes caused by the pandemic lockdown as well as the lockdown scenario in Catalonia.

Results and discussion

Global air contamination from satellite observations

As observed in Fig. 4a and b, in April 2018 and 2019, under normal human activity and no lockdown restrictions, recorded NO₂ levels were higher in the two most populated, highly transited, and industrialized cities of Spain: Madrid (5390 inhabitants/km²) and Barcelona (16,499 inhabitants/km²), revealing two hotspots with levels of NO₂ up to 100 μmol/m². Contrarily, in the same period of time in 2020, under the strictest lockdown, transportation restrictions and industry emission shutdown led to a clear decrease in NO₂ emission. According to S-5P/TROMPOMI satellite records and as observed in Fig. 4c, levels of NO₂ in the cities of Madrid and Barcelona were in the range of 0–40 μmol/m² on April 2020. Therefore, the two hotspots observable on April 2018 and 2019 (Fig. 4a and b) completely receded during the lockdown in April 2020, giving a map with homogeneous contamination distribution over the whole country (Fig. 4c). Similarly, Tobías et al. compared tropospheric NO₂ concentrations supplied by TROPOMI-ESA in the lockdown period with respect to the pre-lockdown. The authors reported a − 57% of decrease in 2020 (lockdown versus pre-lockdown) compared with a − 22% of decrease for the same period of time in 2019. In addition, the satellite images provided in their study also evidenced two hotspots of contamination in Spain, corresponding to Madrid and Barcelona, with NO₂ levels in the same order as ours (up to 200 μmol/m² in 2020).

Fig. 4.

Levels of background tropospheric NO₂ and O₃ measured by TROPOMI-ESA (Veefkind et al. 2012) in the Iberian Peninsula. Top panels: NO₂ tropospheric column measured on a April 2018, b April 2019, c April 2020 (during the strictest lockdown). Bottom panels: O₃ tropospheric column measured on d April 2019, e April 2020 (during the strictest lockdown)

Different concentration distribution changes were observed for O₃ (Fig. 4d, e). According to S-5P TROPOMI data, total O₃ tropospheric column showed an increase in the Iberian Peninsula in April 2020 with respect to April 2019. As observed in Fig. 4d, levels of O₃ were high in the upper northeast part of Spain, but low in the rest of the Peninsula. In contrast, on April 2020, the amount of O₃ increased over all the country, achieving levels up to ~ 0.16 mol/m². Such increase of O₃ linked to the lockdown has been also reported in the literature (Tobías et al. 2020). No georeferenced satellite images of O₃ on April 2018 neither of PM₁₀ are provided in this study since they are not available by the S-5P/TROPOMI-ESA.

Traffic reduction linked to lockdown

The restrictions in mobility imposed during the lockdown in Spain resulted in serious limitations on traffic all over the country, producing dramatic decreases in vehicular mobility, as reported by the General Directorate of Traffic (DGT).

Data of traffic density in Spain provided by the DGT is divided in two periods of time considering the traffic ratios registered on an equivalent period of time in 2019. Period 1 of time (March 9 to June 7) (Dirección General del Tráfico 2020a) corresponds to the period of time when the strictest lockdown was imposed, and period 2 of time (July 1 to August 31) (Dirección General del Tráfico 2020b) responds to the end of the state of alarm and the beginning of the “new normality.” The main recorded results for the two periods of time include (1) long-distance movements all over the country fell 72.48 and 13.26% (periods 1 and 2, respectively) for light vehicles, and 33.77 and 2.25% (periods 1 and 2, respectively) for heavy vehicles; (2) traffic across the borders of Spain decreased by 81.85 and 28.77% (periods 1 and 2, respectively) with Portugal and by 76.68 and 20.09% (periods 1 and 2, respectively) with France; (3) entries to and exits from the city of Barcelona decreased by 55 and 17% in periods of time 1 and 2, respectively; and (4) traffic across the distinct Spanish cities decreased by 37 and 12.11% in periods of time 1 and 2, respectively. As observed, traffic depletion was larger in period 1, when the strictest lockdown was imposed by the Spanish government and showed a recovery in period 2, after the end of the state of alarm.

Not only the travel across cities suffered a decrease but also traffic inside cities, especially in the city of Barcelona. Traffic data inside Barcelona was provided by Barcelona City Council (Ajuntament de Barcelona 2020b, c), at different moments of the pandemic. During the first two weeks of lockdown, the maximum diminution was registered: public transport conducted only 10% of the usual trips, taxis performed only 5% of their services, and travel by bicycles and other personal mobility vehicles decreased by 87% (Ajuntament de Barcelona 2020c). In phase 1 of de-escalation (see Fig. 1), the traffic inside the city of Barcelona was a bit higher compared to the first 2 weeks of lockdown: public transported conducted 19.9% of the usual trips, private vehicles performed only 47.3% of the movements, and travel by bicycles and other personal mobility devices only decreased by 23% (Ajuntament de Barcelona 2020b).

Meteorological characteristics during lockdown respect to previous years

Meteorological factors have a significant effect on atmospheric pollution. As stated by Gkatzelis et al. (2021), wind velocity, stability, and turbulence have an impact on the dilution, transport, and dispersion of chemicals. Sunshine activates the photochemical production of oxidants that constitute smog, while rainfall has an effect that eliminates from the atmosphere some particles and gases.

In this study, meteorological data were obtained from Barcelona open datasets (Opendata), supplied by the Meteorological Service of Catalonia (Meteocat). As observed in Table 2, some meteorological differences between April 2018, 2019, and 2020 were reported in the control site of Barcelona: Fabra Observatory. During the lockdown period, higher average temperatures and higher humidity were registered (+ 1.2 and + 0.3 ºC and + 3 and + 4.5%RH, respect to 2019 and 2018, respectively). However, the most important meteorological variation was the total amount of rain registered in the Catalan city. As observed in Table 2, the total rainfall registered in Fabra Observatory in April 2020 was 254.6 mm, a value + 6.75 and + 4.3 times higher than the rainfall recorded in 2019 and 2018, respectively. In fact, April 2020 achieved a historical record and stood as the more rainy April month registered in Fabra Observatory for the past 107 years, constituting a 454% of the meteorological average of this month ([CSL STYLE ERROR: reference with no printed form.]; Betevé).

Table 2.

Meteorological parameters registered in Fabra Observatory in April 2018, 2019, and 2020. Data supplied by Barcelona open data sets (Opendata)

Meteorological parameters registered in Fabra Observatory
	April 2018	April 2019	April 2020
Mean daily temperature (ºC)	13.6	12.7	13.9
Maximum daily temperature (ºC)	24.9	21.1	21.2
Minimum daily temperature (ºC)	5.1	3.1	6.8
Relatively daily humidity (%)	69	70,5	73.5
Rainfall (mm)	58.6	37.7	254.6
Mean daily insolation (MJ/m2)	22.5	18.3	21.2
Mean wind speed (m/s)	4.2	4.2	3.7
Average wind direction (º)	223	228	218
Maximum wind speed	18.2	21.1	20.2
Maximum wind direction	355	334	292

Finally, some other minor variations were detected regarding the insolation ratio (− 0.5 MJ/m² lower in 2020 respect to the previous 2 years), in wind speed, which was a bit lower in April 2020 with respect to April 2019–2018 (3.7 m/s versus 4.2 m/s), and in the wind direction.

Overall, the meteorological differences registered in April 2020 under pandemia with respect to the same time in the previous 2 years (under no pandemia) were fundamentally due to the amount of rain recorded during the lockdown, which was much higher than in the previous 2 years.

Time trend profiles and percentage changes of NO₂, O₃, and PM₁₀ for pre-, during, and post-lockdown and its relation with anthropogenic mobility

Mobility index (MI) of different human activities along with mean daily profiles of contaminants before (February 15 to March 13), during (March 14 to June 21), and after lockdown (June 22 to August 31) were studied to evaluate the impact of the anthropogenic mobility on air contamination (Fig. 5). In addition, the percentage changes of average concentrations of contaminants during lockdown and after lockdown with respect to the period of time before lockdown were calculated and are shown in Table 3. It can be observed that the levels vary according to urban, semi-urban, and rural sampling stations, with the highest NO₂ and PM₁₀ values in urban areas linked to a heavier traffic and mobility indices.

Fig. 5.

a Mobility index (MI) of different human activities before (light grey-shaded area), during (yellow-shaded area), and after (dark grey-shaded area) the lockdown being put into effect. b Daily average (24 h) concentrations of the three contaminants from 15 February to 31 August 2018 (blue), 2019 (green), and 2020 (red). Observe that concentration values in y-axes are different depending on contaminant and station

Table 3.

Average concentrations (μgm⁻³) and standard deviations of NO₂, O₃, and PM₁₀ for time periods March 14th–June 21 (during lockdown) and June 22^–August 31 (post-lockdown) together with the percentages of change (lockdown versus pre-lockdown and post-lockdown versus lockdown)

	February 15–March 13, 2020 (pre- lockdown) lockdown)	March 14–June 21, 2020 (lockdown)	% of change respect to pre-lockdown	June 22–August 31, 2020 (post-lockdown)	% of change respect to pre-lockdown
NO2
Gràcia	40.7 ± 5	21.5 ± 4	− 47%	26.0 ± 3	− 36%
Vall d’Hebron	27.7 ± 5	16.0 ± 4	− 42%	18.8 ± 2	− 32%
Granollers	38.0 ± 5	16.4 ± 4	− 57%	20.2 ± 3	− 47%
Fabra Observatory	10.3 ± 2	5.9 ± 2	− 43%	8.5 ± 1	− 17%
Manlleu	24.2 ± 2	10.5 ± 3	− 57%	10.4 ± 2	− 57%
Begur	4.9 ± 1	1.9 ± 0.4	− 61%	2.2 ± 0.4	− 55%
Bellver de Cerdanya	7.9 ± 2	2.9 ± 1	− 63%	4.8 ± 2	− 39%
Juneda	7.5 ± 1	5.5 ± 1	− 27%	5.2 ± 1	− 31%
O3
Gràcia	43.4 ± 5	61.6 ± 7	+ 42%	51.9 ± 4	+ 20%
Vall d’Hebron	54.1 ± 5	67.3 ± 6	+ 24%	62.5 ± 5	+ 16%
Granollers	35.0 ± 5	57.4 ± 7	+ 64%	56.4 ± 4	+ 61%
Fabra Observatory	73.3 ± 3	86.6 ± 9	+ 18%	85.3 ± 7	+ 16%
Manlleu	31.1 ± 7	46.8 ± 6	+ 50%	55.1 ± 6	+ 77%
Begur	67.5 ± 2	77.8 ± 9	+ 15%	78.1 ± 5	+ 16%
Bellver de Cerdanya	46.4 ± 7	51.4 ± 3	+ 11%	58.1 ± 7	+ 25%
Juneda	46.9 ± 12	58.0 ± 6	+ 24%	63.1 ± 6	+ 35%
PM10
Gràcia	28.3 ± 5	18.5 ± 3	− 35%	22.8 ± 3	− 19%
Vall d’Hebron	21.4 ± 3	13.0 ± 3	− 39%	17.8 ± 2	− 17%
Granollers	32.1 ± 4	19.5 ± 3	− 39%	24.2 ± 2	− 25%
Fabra Observatory	17.2 ± 3	12.5 ± 2	− 27%	16.7 ± 2	− 3%
Manlleu	32.1 ± 4	18.9 ± 4	− 41%	23.0 ± 2	− 28%
Begur	n.d	n.d	n.d	n.d	n.d
Bellver de Cerdanya	13.2 ± 3	9.9 ± 3	− 25%	15.1 ± 3	− 14%
Juneda	21.3 ± 4	15.3 ± 2	− 28%	20.8 ± 3	− 2%

n.d. no data available

Figure 5a illustrates the changes in MI in terms of different human activities for the periods before, during, and after lockdown in the three different regional areas where the air quality stations of this study are located (Barcelona, Girona and Lleida). It can be clearly observed that all the activities including transport, industries, social places, and educational sectors were running normally before lockdown (see curves in light-grey-shaded areas of Fig. 5a). However, after the beginning of the state of alarm and lockdown, the mobility index of all the human activities except for the residential (the latter showing an increment during lockdown) notably decreased (see curves in yellow-shaded areas of Fig. 5a). The decline of MI of human activities during the COVID lockdown reported in this study is in agreement with the findings of Zhang et al. (2020). In that study, the authors evidenced the same MI trend not only in Spain but in other countries (i.e., USA, France, Italy, Germany, UK, India, Bangladesh, and Pakistan), in which the MI of all human activities decreased a large extent (up to − 90% of drop) while the MI for residential activities significantly increased (up to + 30% of increment) since the start of the pandemic.

In Catalonia, the decline of MI was maximum in April (up to 100% decrease in retail and recreation), when the strictest lockdown was produced, and started to recover right up to 21 of June, when the state of alarm finished and the “new normality” started (see curves in dark-grey-shaded areas of Fig. 5a). Interestingly, in Girona and Lleida, the activity in parks after the lockdown suffered a substantial increase (up to 400% in Girona and up to 250% in Lleida) whereas in Barcelona, all the different type of human activities returned to normal levels.

In Fig. 5b, daily averages (24 h means) of NO₂, O₃, and PM₁₀ concentration (μg/m³) for the equivalent period of time (i.e., before, during, and after lockdown) are represented. The observed differences between pre-lockdown and lockdown periods were not observed in the previous 2 years, and thus, they are mainly attributed to the COVID-19 lockdown scenario rather than to seasonal cycles.

Concerning NO₂, averaged concentrations of this contaminant substantially decreased during lockdown period in 2020 in contrast to the same period of time in 2018 and 2019. Moreover, the differences in the amount of NO₂ were evident when comparing the periods of time pre-lockdown and lockdown during 2020. Interestingly, the pre-lockdown levels of NO₂ in some stations were considerably lower than in the previous 2 years with no known reason. However, this study only focuses on the evaluation of pre-lockdown versus lockdown levels, and for that, the percentages of decrease in the eight monitoring stations were calculated in Table 3: Bellver de Cerdanya (-63%) > Begur (− 61%) > Manlleu and Granollers (− 57%) > Gràcia (− 47%) > Fabra Observatory (− 43%) ~ Vall d’Hebron (− 42%) and Juneda (− 27%). The highest decrease in the two rural stations (Bellver the Cerdanya and Begur) is explained by the fact that they are widely populated in winter as the former is a ski resort and the latter a famous holiday and second residence emplacement, and most houses are heated by gasoil or wood burning, which are emission sources of NO₂ (Michael Alberts 1994; Saud et al. 2011). However, in both sites, people were asked to return to their main residence during lockdown, which was reflected in a high decrease in NO₂ levels. This did not happen in the third rural station (Juneda), since this is not a holiday spot. The urban and semi-urban areas had similar NO₂ decrease during lockdown and reflect the decrease in mobility observed in all urban areas. As expected, in the period of time right after the end of the state of alarm (June 22 to August 31), the levels of NO₂ incremented in all stations, but in no case they returned to the pre-lockdown levels (see last column of Table 3). Moreover, during the lockdown in 2020, the WHOAQG (EUR-Lex) daily reference value of 40 μg/m³ was not exceeded in any site, although this standard value was exceeded in the three urban stations (Gràcia, Vall d’Hebron, and Granollers) during the same period in 2018 and 2019. These results are in agreement with those reported by Baldasano, J.M. et al., who reported NO₂ levels below the WHOAQG reference value during the second half of March 2020 in 24 stations located in Madrid and 9 stations placed in Barcelona.

Concentrations of O₃ showed a substantial increase during lockdown, in the highly populated urban stations of Gràcia (+ 42%) and Granollers (+ 64%), and in the semi-urban station of Manlleu (+ 50%). In the rest of rural stations and in the control station, the registered percentages of change were lower, but still showing an increment of O₃ with respect to the pre-lockdown period: Juneda and Vall d’Hebron (+ 24%), Fabra Observatory (+ 18%), Begur (+ 15%), and Bellver de Cerdanya (+ 11%) (see Table 3). Increment in O₃ levels during the lockdown is related to the reported diminution of NO₂ levels and the suppression of the titration effect and was more evident in the most transited and populated stations: Gràcia (60,300 inhabit/km²), Granollers (4121 inhabit/km²), and Manlleu (1194 inhabit /km²) (see Table 1). Vall d’hebron, despite also being a populated station (7700 inhabit/km²), showed a lower increment of O₃ since this station does not receive the direct impact of traffic. After the end of the state of alarm and the return to the new normality, with increased traffic and NO₂ emissions, the concentration of O₃ started to decrease again, still showing percentages of increase with respect to pre-lockdown levels but in a lesser extent (see last column of Table 3). During the lockdown, the WHOAQG reference value of 100 µg/m³ for O₃ was slightly exceeded in one time in Observatory Fabra, and more times in previous years in Fabra Observatory and Begur.

Concerning PM₁₀, concentrations of this contaminant during lockdown decreased in all stations, but in a minor extent in comparison to NO₂. The percentages of decrease during the lockdown with respect to the pre-lockdown were as follows: Manlleu (− 41%) > Granollers and Vall d’Hebron (− 39%) > Gràcia (− 35%) > Juneda (− 28%) ~ Fabra Observatory (− 27%), and Bellver de Cerdanya (− 25%) (no data for Begur, see Table 3). Thus, the highest decrease was observed for the semi-urban station of Manlleu and the urban stations of Granollers, Vall d’Hebron, and Gràcia. In the stations of Fabra Observatory, Vall d’Hebron, Bellver, and Juneda, levels of PM₁₀ during lockdown in 2020 were lower than the WHOAQG annual reference value of 20 μg/m³, in contrast to 2018 and 2019, when the reference value was exceeded. However, in Gràcia, Manlleu, and Granollers, such reference value was slightly exceeded at some moments during the lockdown. The percentages of change and the levels of PM₁₀ obtained in our study are in agreement with those reported by Tobías, A et al., in a study performed in two air quality stations in Barcelona. In that study, the authors also reported PM₁₀ levels slightly over the WHOAQG limit value in the station placed in the urban center of Barcelona, which suggests a location very similar to our Gràcia station.

Pearson’s correlation coefficients of percentages of change of contaminants in the stations located in Barcelona (i.e., Gràcia and Vall d’Hebron) with respect to the mobility index, MI, in the same city were calculated for the period of time February 15 to August 31 and are summarized in Table 4. As it can be observed in the table, NO₂ showed a positive correlation with MI both in Gràcia (+ 0.51) and Vall d’Hebron (+ 0.38), which may suggest that the diminution of NO₂ levels was in a significant part caused by the MI reduction. In contrast, O₃ showed a negative correlation both in Gràcia (− 0.56) and Vall d’Hebron (− 0.41), indicating that the diminution of human mobility and traffic depletion contributed in an increase of O₃ levels in these two neighborhoods of Barcelona due to the lower titration effect. Finally, the correlation of PM₁₀ with MI was, as which occurred with NO₂, positive in the two locations, despite a bit lower (+ 0.41, in Gràcia and + 0.33, in Vall d’Hebron), also indicating that part of the diminution of PM₁₀ contamination can be attributed to traffic restrictions during lockdown.

Table 4.

Pearson’s correlation coefficient index among the mobility index (MI) and the percentage of change of contaminants for the period of time (February 15^–August 31)

	NO2	O3	PM10
	Pearson’s correlation coefficient	Pearson’s correlation coefficient	Pearson’s correlation coefficient
Gràcia	+ 0.51	− 0.56	+ 0.41
Vall d’Hebron	+ 0.38	− 0.41	+ 0.33

Hourly profiles and percentage changes of NO₂, O₃, and PM₁₀ during strictest lockdown: April 2020 versus April 2019 and 2018

Detailed evaluation of air contamination changes during the strictest lockdown was performed, focusing the study on data from April 2020 and comparing them to data acquired in April 2019 and April 2018, the latter used as basal concentrations. With that purpose, for these periods of time, hourly profiles of the contaminants and their percentages of change were calculated and evaluated to determine the sources of pollution and human impacts.

Hourly average profiles are represented in the plots of Fig. 6, each plot containing information of one contaminant, one station and 3 years simultaneously (in blue data from 2018, in green data from 2019, and in red data from 2020). Average values and the associated standard deviations were calculated for each hour as the mean ± SD of all the month of April (n = 30) and are represented in Fig. 6 with continuous lines and shaded areas, respectively.

Fig. 6.

Hourly average (30 days) concentrations of NO₂, O₃, and PM₁₀ together with their standard deviation (shaded areas) calculated during April 2018 (blue), 2019 (green), and 2020 (red) for the different air quality monitoring stations. Observe that concentration values in y-axes are different depending on contaminant and station

In order to obtain the percentages of change, the mean concentrations of each contaminant in each station were first calculated as averages of the whole month of April for each year (see Table 5). Then, the percentages of change were calculated as the % of variation among April 2019 versus April 2018 and April 2020 versus April 2019. The reason why the percentages of change were calculated in that way and not considering April 2020 versus April 2018 is the fact that, in most locations, air quality was better in 2019 with respect to 2018. The improvement in air quality observed in 2019 can be attributed to a combination of factors: on the one hand and most important, the implementation of LEZ, and on the other hand, the weather patterns of 2019. Such air quality improvement observed in 2019 with respect to 2018 mostly due to the implementation of emission control policies should not be observed in that much extent in 2020 with respect to 2019, since the LEZ restrictions which were put into permanent effect in January 1, 2020 were the same ones already applied in 2019.

Table 5.

Average concentrations (μg m⁻³) and standard deviations (n = 30) of NO₂, O₃, and PM₁₀ for April 2018, April 2019, and April 2020 together with the percentages of change (April 2019 versus 2018 and April 2020 and 2019)

	Average April 2018	Average April 2019	Average April 2020	% of change 2019 vs 2018	% of change 2020 vs 2019
NO2
Gràcia	49.2 ± 13	36.1 ± 10	19.7 ± 5	− 27%	− 45%
Vall d’Hebron	32.1 ± 10	25.7 ± 9	17.3 ± 6	− 20%	− 33%
Granollers	29.7 ± 7	29.7 ± 9	14.3 ± 5	0%	− 52%
Fabra Observatory	13.7 ± 3	13.2 ± 2	5.0 ± 1	− 4%	− 62%
Manlleu	15.2 ± 3	16.5 ± 4	10.2 ± 2	+ 9%	− 38%
Begur	3.0 ± 1	3.4 ± 1	1.9 ± 0.5	+ 13%	− 44%
Bellver de Cerdanya	7.0 ± 1	7.8 ± 1	2.9 ± 1	+ 10%	− 63%
Juneda	6.3 ± 2	8.2 ± 2	6.3 ± 1	+ 30%	− 23%
O3
Gràcia	50.9 ± 23	61.1 ± 20	68.2 ± 20	+ 20%	+ 12%
Vall d’Hebron	62.1 ± 23	77.2 ± 25	70.6 ± 20	+ 24%	− 9%
Granollers	55.5 ± 25	63.2 ± 31	64.9 ± 21	+ 14%	+ 3%
Fabra Observatory	88.8 ± 7	97.3 ± 8	92.2 ± 9	+ 10%	− 5%
Manlleu	52.8 ± 19	60.1 ± 26	49.2 ± 18	+ 14%	− 18%
Begur	101.0 ± 6	96.6 ± 7	83.2 ± 7	− 4%	− 14%
Bellver de Cerdanya	67.9 ± 18	70.2 ± 19	51.3 ± 13	+ 4%	− 27%
Juneda	60.0 ± 23	68.9 ± 26	58.0 ± 23	+ 15%	− 16%
PM10
Gràcia	31.7 ± 19	22.5 ± 10	16.8 ± 8	− 29%	− 25%
Vall d’Hebron	23.6 ± 13	27.2 ± 18	11.0 ± 6	+ 15%	− 60%
Granollers	27.3 ± 15	23.8 ± 9	17.7 ± 6	− 13%	− 26%
Fabra Observatory	n.d	15.4 ± 9	12.1 ± 7	n.d	− 21%
Manlleu	23.1 ± 13	18.4 ± 9	15.1 ± 7	− 20%	− 18%
Begur	n.d	10.2 ± 5	n.d	n.d	n.d
Bellver de Cerdanya	17.3 ± 13	10.4 ± 8	10 ± 6	− 40%	− 4%
Juneda	23.5 ± 17	18.2 ± 12	14.0 ± 7	− 23%	− 23%

n.d. no data available

Daily profiles calculated from hourly averages shown in Fig. 6 evidence that NO₂ had similar hourly profiles in all stations, showing two maxima. The first maximum appeared between 8:00 and 10:00 a.m. (2 h delayed in Fabra Observatory), coincident with the rush traffic hour and, thus, can be assigned to the increasing fuel combustion by vehicles. It is interesting to stand out that despite the lockdown and the traffic depletion, this maximum was still observed. The second maximum, a bit lower and wider, appeared around 22:00 p.m. in all stations. However, mean ranges of NO₂ differed among stations. Higher levels were registered in highly populated urban stations of Gràcia, Vall d’Hebron, and Granollers (~ 10–60 μgm⁻³), followed by semi urban station of Manlleu and control site Fabra Observatory (~ 5–30 μgm⁻³), ending with rural stations of Begur, Bellver de Cerdanya, and Juneda (~ 2–15 μgm⁻³). Thus, we observed a correlation between the NO₂ levels and the density of population of the air quality stations, the latter provided in Table 1. When comparing amounts of NO₂ among 2018 and 2019, in three stations (i.e., Gràcia, Vall d’Hebron and Fabra Observatory), the amount of NO₂ decreased; in Granollers’ station, there were no differences (i.e., 0% of change) among these 2 years; and in four stations (Manlleu, Begur, Bellver de Cerdanya and Juneda), there was an increment in NO₂ levels. In contrast, there was a uniform tendency in % of change of NO₂ levels among April 2020 versus April 2019 (see Table 5): in all stations, the amount of this contaminant was lower in 2020 (under pandemic lockdown) with respect to the same period of time in 2019 (under no pandemic). In particular, the percentages of change from higher to lower were as follows: Bellver de Cerdanya (− 63%) ~ Fabra Observatory (− 62%) > Granollers (− 52%) > Gràcia (− 45%) ~ Begur (− 44%) > Manlleu (− 38%) > Vall d’Hebron (− 33%) > Juneda (− 23%). As previously observed in “Time trend profiles and percentage changes of NO₂, O₃, and PM₁₀ for pre-, during, and post-lockdown and its relation with anthropogenic mobility,” the rural station of Bellver de Cerdanya showed the highest relative decrease of NO₂, again presumably due to the ceasing hotel and ski resort activities during eastern holidays. These changes in the concentrations of NO₂ in the Bellver de la Cerdanya rural station are however only in relative terms and local. Absolute changes show that the NO₂ concentration depletion was much more important in the Barcelona urban area than in the rural areas due to the lockdown situation. Moreover, the weather conditions of April 2020 in Barcelona urban area favored the cleansing of the atmosphere, including NO₂ gases, as they were especially rainy. The decrease of NO₂ reported in this study due to meteorological and lockdown restrictions is in accordance to previous air quality monitoring studies performed in the cities of Barcelona and Madrid (Baldasano 2020).

Moreover, in all stations, NO₂ depletion was more evident in the second maximum and in general for the second part of the day (12–24 h). In addition, it is worthy to stand out that in all stations, NO₂ concentration levels showed significantly lower standard deviation in 2020 with respect to the previous years (narrowed red-shaded areas in Fig. 6). In the lockdown period, the WHOAQG annual reference value of 40 µg/m³ was not exceeded in any site, whereas in previous years, it was exceeded in Gràcia, Vall d’Hebron, and Granollers stations (see Fig. 6), as previously observed when comparing the whole pre-, lockdown, and post-lockdown period for the 3 years (“Meteorological characteristics during lockdown respect to previous years”).

Hourly average profiles of O₃ showed a marked minimum between 8:00 and 10:00 a.m. (again 2 h delayed in Fabra Observatory), coincident with the maximum NO₂ concentration; a maximum around 16:00 p.m., coincident with the increase of solar radiation and simultaneous to the NO₂ minimum between the two maxima; and a minimum around 22:00 p.m., coincident with the second NO₂ maximum. The latter increase of O₃ at night can be attributed to the suppression of the titration effect (see Fig. 2).

Mean ranges of O₃ also varied among stations. Contrarily to NO₂, higher levels of O₃ were registered in Fabra Observatory and Begur stations (~ 60–120 μgm⁻³), followed by Gràcia, Vall d’Hebron, Bellver de Cerdanya and Juneda (~ 20–100 μgm⁻³), and finally with Granollers and Manlleu (~ 0–100 μgm⁻³). When comparing O₃ levels among 2018 and 2019, we observed that there was already an increment in ozone in April 2019 with respect to April 2018. Such increment was registered in all stations except for Begur, where a slight decrease was detected. Differences in O₃ in April 2020 versus 2019 showed two different tendencies. In two stations, the amount of O₃ was higher in April 2020 and the percentages of change, from higher to lower, were as follows: Gràcia (+ 12%) > Granollers (+ 3%). However, for the remaining six stations, the amount of O₃ was lower in April 2020 with respect to the previous year, in the following order: Bellver de Cerdanya (− 27%) > Manlleu (− 18%) > Juneda (− 16%) > Begur (− 14%) > Vall d’Hebron (− 9%), and Fabra Observatory (− 5%). It is important to highlight that the increase produced in Gràcia (and in a lesser extend in Granollers) occurred in the second part of the day (12–24 h, see Fig. 6), coincident with the previously reported decrease of NO₂. In the lockdown period, the WHOAQG 8-h reference value of 100 µg/m³ was not exceeded in any site, whereas in previous years, it exceeded in Fabra Observatory and Begur stations (see Fig. 6).

PM₁₀ hourly profiles showed a clear increase between 8:00 and 10:00 a.m. in most of stations (Gràcia, Granollers, Manlleu, Bellver de Cerdanya and Juneda). Mean ranges of PM₁₀ were similar for all the stations (~ 10–50 μg m⁻³), except for the control site (Fabra Observatory), which registered the lowest levels (~ 5^–30 μgm⁻³). The comparison among years evidenced a decrease of PM₁₀ in 2020 with respect to 2019 in all stations, despite that the diminution in Bellver de Cerdanya was little. Larger diminution was registered in the three more transited urban stations: Vall d’Hebron, showing a percentage of diminution of − 60%; Gràcia, with a diminution of − 25%; and Granollers of − 26%. This is due to the fact that PM levels are very dependent on the traffic influence (dust resuspension, erosion of road pavements, and brakes), and during the lockdown, the density of vehicles in the city of Barcelona decreased: − 90% in public transport, − 95% in taxis, and − 87% in bicycles and other personal mobility vehicles (see “Traffic reduction linked to lockdown” of this manuscript). Lower percentages of decrease of PM₁₀ in April 2020 with respect to April 2019 were registered in the other stations of Juneda (− 23%), Fabra Observatory (− 21%), Manlleu (− 18%), and Bellver de Cerdanya (− 4%). No data of concentration of PM₁₀ in Begur station in April 2020 were available, and thus, their percentage of change could not be calculated. Moreover, in the lockdown period, the WHOAQG daily reference value of 50 µg/m³ was not exceeded in any site, and the annual reference value of 20 µg/m³ was only slightly exceeded among 10–12 h in Gràcia, Granollers, Manlleu, and Juneda, while highly exceeded in the same stations and also in Vall d’Hebron and Bellver de Cerdanya in previous years (see Fig. 6).

There are quite a few other studies concerning changes in air quality during the COVID lockdown in many areas throughout the globe (Menut et al. 2020; Ropkins and Tate 2021; Filonchyk et al. 2021), and most of them report NO₂ diminution and small increase in O₃ and PM₁₀ diminution, the latter being generally a modest depletion in comparison to that of NO₂. For instance, Menut et al. (2020) reported a large reduction in NO₂ concentrations, a lower reduction in particulate matter, and a mitigated effect on ozone concentrations over western Europe. Filonchyk et al. (2021) reported reductions of tropospheric NO₂ approximately by − 10 to 19%, and reductions of PM₁₀ from − 8.5 to − 33.9% in 2020 with respect to 2019, in Poland, eastern Europe. Also, Ropkins and Tate (2021) reported NO₂ decreased from − 32 to − 50% and O₃ increase by + 20% across the UK. These findings are in accordance to the ones obtained in the present study. Therefore, the results hereby presented confirm the improvement of air quality due to the lockdown that has been observed worldwide, in the region of Catalonia (Spain).

However, a more exhaustive analysis of the acquired data using multivariate statistical and chemometric methods is pursued with the goal of the apportionment of the different sources of the three investigated air quality parameters (NO₂, O₃, and PM₁₀) and to describe how their temporal and geographical profiles changed during COVID-19. The results of this more exhaustive analysis will be hopefully reported when the COVID-19 pandemic situation finishes.

Conclusions

This study shows the impact of social movements on air quality by integrating different but complementary techniques: direct surface measurements of air pollutants, satellite observations, and mobility indexes. It also demonstrates the importance to accurately evaluate the time frame, from days to years, to determine the changes and evolution of NO₂, O₃, and PM₁₀ during the pandemic. In the 8 areas studied, considering urban, semi-urban, and rural, the concentration of NO₂, O₃, and PM₁₀ varied during the COVID-19 lockdown with respect to the pre-lockdown period and with respect to previous years. In the major part of air quality stations, the levels of these air contaminants were lower than the WHOAQG reference values. Only for O₃ the reference value slightly exceeded one time during lockdown in the control station (Observatory Fabra) and for PM₁₀ in four stations (Gràcia, Granollers, Manlleu, and Juneda). However, these standards were exceeded multiple times in most stations (especially urban areas) during the same period of time in the two previous years.

In this study, we observed a significant correlation among the levels of contaminants in the two stations located in Barcelona and the diminution in anthropogenic mobility registered in the same city (up to 100% decrease in retail and recreation). Pearson’s correlation coefficients were positive for NO₂ and PM₁₀, since lower mobility and traffic depletion resulted in lower levels of these contaminants. Conversely, a negative correlation was observed for O₃, as the reduction of NO₂ emissions went together with the lower titration effect.

The overall consequences of the COVID-19 lockdown regarding NO₂, O₃, and PM₁₀ air pollution extracted in the comparison of lockdown with respect to pre-lockdown periods in 2020 were analogous to those extracted in the comparison of April 2020 (time for the strictest lockdown) with respect to April 2019 (under no pandemic). For NO₂, the lockdown restrictions, together with the higher rainfall registered in 2020, especially in the Barcelona urban area, produced a decrease in the levels of this pollutant, especially in the urban stations in absolute terms, although also significant in relative terms because of the ceasing activities during eastern holidays in some rural stations close to ski hotels and resorts.

The effects of the lockdown regarding O₃ levels were opposed to those of NO₂. The traffic depletion originated by the lockdown resulted in a decrease of NO₂ levels, and thus, a suppression of the titration effect, resulting in higher amounts of ozone. Larger increments of O₃ were registered in the urban and most transited stations of Gràcia and Granollers. Levels of PM₁₀ also suffered a depletion due to the lockdown, as occurred with NO₂, but in a lesser extent. Largest decrease in PM₁₀ levels due to lockdown was registered in the three urban stations of Gràcia, Vall d’Hebron, and Granollers. Satellite S-5P/TROMPOMI images confirmed the results for NO₂, by showing two hotspots of contamination in Spain (Madrid and Barcelona) in April 2018 and 2019 that disappeared in April 2020, and for O₃ by showing higher levels of O₃ all over the country in April 2020. It is important to mention that the results presented in this study may be influenced also by the especial meteorological conditions observed in Observatory Fabra during the lockdown (i.e., extreme rain amount registered in April 2020 respect to 2019 and 2018). Therefore, the observed air quality improvement hereby presented might have a combined contribution of weather-driven and of lockdown-driven factors.

Overall, this study provides new evidences on the air quality improvement produced in the scenario originated by the COVID-19 pandemic, which represents a significant improvement in public health and quality of life. The experimental evidences about the improvement of air quality during the pandemic have made possible to see cities with clean and healthier skies, which were not observed for decades and confirmed the potential of applying traffic restriction policies in the near future.

Author contribution

Eva Gorrochategui contributed to data processing and writing the paper. Isabel Hernandez and Eva Pérez-Gabucio provided air quality data. Sílvia Lacorte and Romà Tauler contributed to data interpretation and paper revision.

Funding

This study was supported by the Ministry of Science and Innovation of Spain under the project PID2019-105732 GB-C21.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Ethical approval

Not applicable.

Consent to participate

The authors are informed and agree to the study.

Consent for publication

The authors agree to publication in the journal.

Conflict of interest

The authors declare no competing interests.