The coronavirus disease of 2019 (COVID-19) is a viral infection generated by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). This disease manifests clinically as a severe acute respiratory syndrome with symptoms of fever, dry cough, dyspnea, respiratory disorders and pneumonia that can lead to progressive respiratory failure and death (Gorbalenya et al. 2020; WHO 2005; Zhou et al. 2020; Zhu et al. 2020). Human-to-human transmission of SARS-CoV-2 was confirmed in Wuhan (China) on January 20, 2020 (Kucharski et al. 2020). However, global mobilization of infected people in advance of the implementation of control measures (Jenson 2020; Kenyon 2020) resulted in numerous epidemic chains that produced new outbreaks in different nations. SARS-CoV-2 has inevitably spread throughout South America (WHO 2020). The first confirmed case in Latin America occurred in Brazil on February 26, 2020 (WHO 2020). In the following days, cases were reported in Mexico (February 28), Ecuador (February 29), Argentina (March 3), Chile (March 3) and Colombia (March 6). The first confirmed case in Peru occurred on March 6, 2020 (MINSA 2020).

Air pollution has decreased in cities around the world because of restrictions ("closures") imposed to limit the spread of SARS-CoV-2 (Coccia 2020; Kucharski et al. 2020; Muhammad et al. 2020; Saadat et al. 2020). Recently, researchers have reported that partial or complete lockdowns implemented in response to the pandemic have caused decreases in air pollutant concentrations in different cities and megacities in China (Bao and Zhang 2020; He et al. 2020; Wang and Su 2020; Xu et al. 2020a, b), India (Mahato et al. 2020; Sharma et al. 2020), Malaysia (Abdullah et al. 2020), Brazil (Dantas et al. 2020; Nakada and Urban 2020), Australia (Sánchez-García and Leon 2020), Canada (CBC 2020), Italy (Collivignarelli et al. 2020), Spain (Tobías et al. 2020) and the United States (CNBC 2020). In several cities in China, measures to suppress the spread of COVID-19 have led to a remarkable improvement in air quality. (He et al. 2020) reported that the PM2.5 (particulate material smaller than 2.5 μm) concentration decreased by 25% from that over the same period in 2019 and that this pollution reduction is beneficial to public health in the short term as long as the measures implemented by COVID-19 are maintained. Within a few days of forced restrictions, reduced pollution levels were observed in cities across India (Mahato et al. 2020). Also, reductions were observed on the order of 60% and 40% in the PM10 (particulate material with aerodynamic diameter smaller than 10 μm), e.g. Sale, Morocco (Otmani et al. 2020), New Delhi, India (Mahato et al. 2020), Milan, Italy (Collivignarelli et al. 2020), and PM2.5 levels, e.g. Milan, Italy (Collivignarelli et al. 2020), Zaragoza, Spain, Shanghai, China, New York, USA, Rome, Italy, Los Angeles, USA (Chauhan and Singh 2020). On the order reducctions of 53% and 30% for NO2 levels were observed, e.g. Milan, Italy (Collivignarelli et al. 2020), Sao Paulo, Brazil (Nakada and Urban 2020)) and CO levels, e.g. Milan, Italy (Collivignarelli et al. 2020), Rio de Janeiro, Brazil (Dantas et al. 2020), New Delhi, India (Mahato et al. 2020). The results of another study in India showed that reduced PM2.5 levels resulting from lockdown measures have counteracted increased pollutant concentrations from unfavorable weather conditions (Sharma et al. 2020). In Malaysian cities, reductions of between 30 and 60% in PM2.5 have been observed (Abdullah et al. 2020). In Sao Paulo, Brazil, drastic reductions in the concentrations of NO (nitrogen monoxide, up to 77%), NO2 (up to 54%) and CO (up to 65%) have been observed in urban areas during partial lockdowns compared to the five-year monthly means (Nakada and Urban 2020). Also, in other studies, spaceborne NO2 column observations from two high‐resolution instruments (TROPOMI and OMI), reveal unprecedented NO2 decreases over China, South Korea, western Europe, the United States (Bauwens et al. 2020) and Spain (Petetin et al. 2020) as a result of public health measures enforced to contain the coronavirus disease outbreak (Covid‐19). These decreases observed have mainly resulted from reductions in urban road transport in many regions, with additional contributions from decreased industrial and commercial activity.

The highest infection rates have been reported in areas of high population density (Chauhan and Singh 2020). Over 30% of the population of Peru is concentrated in the Lima metropolitan area (LMA) (INEI 2020). The government of Peru implemented a complete national lockdown that drastically reduced anthropogenic activities (ED 2020). In this study, the impact of the sanitary measures implemented in the LMA to reduce the spread of the COVID-19 on the air quality are quantified by comparing the concentrations of particles (PM10 and PM2.5) and NO2 during pre-lockdown period (February 1 and March 15 of 2020), historical period (March 16 to April 30 2017–2019) and lockdown period (March 16 to April 30, 2020). The results of this study can provide indicators of compliance with measures against COVID-19 and are therefore useful for both air pollution and public health regulatory agencies.

Materials and methods

Study area

The LMA is made up of the provinces of Lima (capital of the Republic of Peru) and Callao (Fig. 1), forming an extensive and populous urban area; it is one of the five megacities of South America, with a population of approximately 10 million inhabitants (population density of 3278.9 inhabitants/km2 for Lima and 6815.8 inhabitants/km2 for Callao). It is located in the central and western zone of the South American continent (longitude 77°W and latitude 12°S) in a large alluvial plain formed by the valleys of the Chillón, Rímac and Lurín rivers (IGN 2014). The LMA climate is subtropical, with a mean annual temperature of 19 °C and a relative humidity of 80% (SENAMHI 2016). The amount of rainfall barely reaches 10 mm per year, but cloudy skies are common throughout the year.

Fig. 1
figure 1

Road map of LMA, population density and station locations of Lima air quality monitoring network; details on the network can be found in Table S1 of the supplementary material

Air pollution database and statistical analysis

For each station, data from February, March and April from 2017 to 2020 were used to calculate the mean levels of each pollutant at different time scales. The period between 1 February 1 and March 15 of 2020 was considered the pre-lockdown period and that between March 16 and April 30 of 2020 was defined as the lockdown period. Likewise, the mean of the data from March 16 to April 30 in the years 2017, 2018 and 2019 was used for the historical period. Hourly mean concentrations were used in the analysis. The relative change (%) in the concentrations was assessed by calculating and comparatively analyzing the concentration variations over the pre-lockdown, lockdown and historical periods. Before analysis, the data were validated to rectify duplicate entries and gaps.

The LMA has a real-time air quality monitoring network (AQMN) (SENAMHI 2020). The data from the monitoring network are the responsibility of the National Meteorological and Hydrological Service (Servicio Nacional de Meteorología e Hidrología—SENAMHI), an entity attached to the Ministry of the Environment (Ministerio del Ambiente—MINAM). These stations measure atmospheric pollutant concentrations of PM10, PM2.5 NO2 and O3 on an hourly basis. PM10 and PM2.5 are monitored at five of the six stations in the network under study while ozone and NO2 is only one stations (see Table S1 for more information related to station names, measurement principles and instrumentation in supplementary material). Meteorological variables (that is, the temperature, relative humidity, wind speed and direction) are also measured. The stations are located in urban environments that are mainly impacted by emissions from vehicular traffic and residential emissions (Silva et al. 2017, 2018).

Spatial distribution analysis

In addition to the temporal analysis, a spatial distribution analysis of NO2 levels was performed for in pre-lockdown and lockdown periods using satellite monitoring data. The data are obtained from remote sensing provided by the Copernicus Sentinel-5 Precursor Tropospheric Monitoring Instrument (S5p/TROPOMI), developed by the European Space Agency (ESA) (ESA 2018), at a spatial resolution of 0.01° × 0.01° and daily temporal resolution. In both cases, the data were downloaded in NetCDF format and then processed in the software R (Team 2020) and/or Python (Python 2020).

Statistical analysis

The descriptive statistical analysis was performed in MS-Excel (Excel® 2017) and using the open source software programming language R (Team 2020) with the Openair package (Carslaw and Ropkins 2012, 2020). The analysis of variance (ANOVA) is a well-known statistical approach for analyzing differences between group means (for air pollutant concentrations and meteorological variables). Different time series during the lockdown and non-lockdown periods were compared using ANOVA to statistically test whether the means were the same across the groups.

Results and Discussion

After the first confirmed COVID-19 case, occurred in Peru on March 6, 2020 (see Fig. 2), the government, through the Ministry of Health (Ministerio de la Salud—MINSA), put into operation a National Preparedness and Response Plan for the Risk of Introduction of the Coronavirus to strengthen surveillance, containment and response systems (MINSA 2020). On March 11, a state of health emergency was decreed at the national level (see Fig. 2). From March 12 to 15, the number of positive cases in Peru increased to a total of 71 cases and were concentrated in the LMA (MINSA 2020). On March 15, a state of emergency was decreed (ED 2020), making Peru the first country in South America to take strict measures to prevent an increase in positive cases (see Fig. 2). These measures included mandatory social isolation (via a national lockdown) and a complete lockdown of the border, starting on March 16. On March 18, a national curfew was enacted to tighten the mandatory social distancing measures because people were not conforming to the lockdown restrictions, and the use of private vehicles was prohibited from March 19. Despite all the sanitary measures implemented, the number of confirmed cases in Peru surpassed 100,000 on May 20 (MINSA 2020).

Fig. 2
figure 2

Evolution of confirmed COVID-19 cases in Peru (red) and LMA (brown), national electricity demand (blue) and percentage of change in public transport use (green)

The impact of the measures implemented during the state of emergency was to freeze production activities, as reflected by electricity consumption (COES 2020). Figure 2 shows the mean daily electricity consumption in Peru: the application of isolation and immobilization measures is coincident with a clear decrease of approximately 40% in the energy demand. This decrease in production activity is not counteracted by overconsumption by people forced to stay at home. The decrease in the electricity demand reflects the freezing of production activities and attendant potential sources of air pollutant emissions.

By April 30, 2020, that is, six weeks after the declaration of the national state of emergency on March 16, 2020, LMA contamination, in terms of the PM10, PM2.5 and NO2 concentrations, had decreased (Fig. 3, Table 1). Likewise, compared with the historical period (the three previous years), similar behavior was observed. These observed changes are analyzed in the following sections.

Fig. 3
figure 3

Box plot (mean, median and 90th and 10th percentiles) by pollutant and station of LMA air quality monitoring network (see Fig. 1) for the following periods: A: Mar 16-Apr 30, 2017–19 (three-year mean, in red), B: Feb 1-Mar 15, 2020 (pre-lockdown, in green); C: Mar 16-Apr 30, 2020 (during lockdown)

Table 1 Mean concentration in (µgm−3) recorded by LMA air quality monitoring network for the following periods: A: Mar 16-Apr 30, 2017–19 (three-year mean); B: Feb 1-Mar 15, 2020 (pre-lockdown) and C: Mar 16-Apr 30, 2020 (during lockdown)

A 40% decrease in the PM10 level between the pre-lockdown and lockdown periods was observed. A similar trend was observed for the PM2.5 level, in terms of the magnitude of the relative change. The ANOVA results revealed a statistically significant decrease in the PM10 and PM2.5 concentrations from the pre-lockdown to the lockdown: the p-value was smaller than the significance level of 0.01 at a 99% confidence level. The highest decrease for both PM10 and PM2.5 levels was observed at the STA station. The smallest relative changes were observed at the CDM stations, which are located in a commercial area with large malls, small businesses, restaurants, etc. and are therefore highly impacted by emissions from vehicle traffic. Sustained commercial activity produced a smaller decrease in PM concentrations because of the lockdown for the CDM area than other zones. These other zones are mainly residential areas with unpaved streets, for which the reduction in vehicular motion reduced vehicle emissions as well as particle resuspension. Thus, the lockdown produced a decrease in the PM concentration of approximately 50%, which is similar in magnitude to the decrease in the NOx levels. This reduction can be explained in terms of the decrease in vehicular emissions by the lockdown measures, considering that the 2.2 million motor vehicles in the LMA make a total of nine million trips a day.

Ozone (O3) is the only pollutant for which an increase in concentration (of 59%) was observed during the lockdown period compared to the non-lockdown period, at the CDM station. This increase was shown to be statistically significant using ANOVA (the p-value was smaller than the significance level of 0.01 at a 99% confidence level). The increase in the O3 concentration can be explained in terms of the complex chemistry of O3 formation from mixtures of volatile organic compounds (VOCs) and NOx (Finlayson-Pitts and Pitts 2000). A reduction in NOx concentrations leaves a higher number of OH radicals available to react with VOCs, promoting ozone formation. O3 is also eliminated via rapid reaction with NO. Thus, a reduction in available NO increases O3 atmospheric levels. This behavior is consistent with a previous report that O3 formation in the LMA is hydrocarbon-limited (Silva et al. 2018).

The variability in the PM10, PM2.5, NO2 and O3 concentrations over the historical period (March 16 to April 30) from 2017 and 2019 was similarly analyzed. The same general trends were observed as compared to the pre-lockdown period of 2020, albeit with more extreme concentration decreases. The highest decreases in the PM10, PM2.5 and NO2 concentrations were 73%, 64% and 48%, respectively. A significant increase in variability was observed at the CDM station for O3 (166%) during the lockdown compared with the concentration for 2017–2019 and 2020. A comparison of the concentration values for the pairs of each pollutant and station for 2017–2019 and during the lockdown period in 2020 using ANOVA showed that the differences between these periods was significant (p-value < 0.01 at a 99% confidence level).

No statistically significant differences were found between the temperatures, relative humidity and surface winds recorded during the pre-lockdown and lockdown periods in the present study. In general, in LMA the highest PM10 concentrations are observed in the summer and early autumn (February-April). On the other hand, regarding the PM2.5 concentrations, the highest concentrations are recorded between late autumn and winter (May and September). This could be explained by the subsidence thermal inversion weakens in the middle of spring and early fall and because the humidity decreases, which is detrimental to the formation of secondary particulate matter and contrary to what occurs during the cold and humid period (May–September) (Silva et al. 2017).

Figure 4 shows the spatial variability in the NO2 column (µmol/cm 2) obtained from remote sensing provided by the Copernicus Sentinel-5 Precursor Tropospheric Monitoring Instrument (S5p/TROPOMI) for the pre-lockdown and lockdown periods, as well as the corresponding relative change (%).

Fig. 4
figure 4

Spatial distributions of LMA mean concentrations of NO2 (satellite data), during Feb 1-Mar 15, 2020 (pre-lockdown) and Mar 16-Apr 30, 2020 (during lockdown) and percentage relative change

NO2 distribution in LMA is influenced by its proximity to the sea and winds, see Fig. 4a. In general, in the coastal area of the city, there are lower concentrations due to the coastal wind that blows from the ocean to the mountains (Silva et al. 2017). The main sources of outdoor air pollution are automotive fleet for these the reason the area of greatest impact turns out to be the downtown of the city where most vehicle traffic is focused and where important economic activity takes place (Romero et al. 2020; Silva et al. 2017). Likewise, the wind disperses the atmospheric pollution emitted in the downtown area towards the mountain slopes. Restrictive measures to prevent the spread of COVID-19 clearly reduce the sources of NO2 emissions in the city (see Fig. 4b).

The relative changes in the LMA NO2 concentration are on the order of -40% and are homogeneously distributed over the domain, which is consistent with the surface records from the STA station (see Table 1). Significant decreases in NO2 levels in the 40% to 30% range are evident over most of the urban area of the domain. The decreases in NO2 levels are on the same order of magnitude and exhibit the same trend as the surface records.


The LMA has experienced significant reductions in air pollution since a complete national lockdown was decreed during the COVID-19 pandemic. These results confirm the effects of the measures of social confinement and the suspension of productive activities on the concentrations of atmospheric pollutants observed in others in urban areas around the world. This article shows that a strong lockdown, such as that implemented in Peru, produces significant improvements in air quality. It is important to incorporate the effects of the local and synoptic meteorological factors that could influence the observed variations in this type of analysis.

These relative changes in air quality have also been shown to have positive impacts in reducing the possible impact of air pollution as a cofactor of morbidity and mortality for COVID-19 and other respiratory diseases (Anjum 2020; Pradhan et al. 2020). Although the lockdown has had demonstrably positive effects on the control and prevention of the spread of COVID-19 and the reduction in air pollution, there have been strong economic and social impacts in countries that have implemented restrictive measures. It is also important to mention that in part of South America, the winter period is beginning, with a reduction in temperature and a worsening of ventilation conditions and dispersion of pollutants, which are factors to consider in the management and assessment of the pandemic and its impact on the environment and vice versa. Other factors, such as weather conditions and the territorial distribution of infections, should be investigated in detail during the total lockdown implemented due to the pandemic to provide effective physical distancing with a lower socioeconomic impact.