INTRODUCTION

Air quality is one of the main environmental factors that has a direct impact on human health [13, 19]. In accordance with the updated recommendations of the World Health Organization (WHO), more than 90% of the world’s population as of the end of 2019 lives in regions with high concentrations of chemicals in the air [28]. The problem of air pollution is most acute in cities and industrial zones, where industrial enterprises, urban heating systems, and automobile internal-combustion engines become sources of harmful impurities [10].

In Russia, with a share of the urban population of 74.74% [9], observations of the state of atmospheric air in cities are carried out at stationary and mobile posts of Roshydromet (Federal Service for Hydrometeorology and Environmental Monitoring). Currently, regular monitoring of the content of 57 pollutants in the atmospheric air is carried out at 612 posts in 221 out of 1117 cities [2, 5].

A similar approach has been implemented, in particular, in the countries of the European Union (EU). An observational network of approximately 5000 urban and suburban air-pollution monitoring stations [24] operates as part of the AirBase system managed by the European Environment Agency. The received data are openly published on the Internet in real time on a specialized portal European Air Quality Index (https:// www.eea.europa.eu/themes/air/air-quality-index/index).

At the same time, the possibilities of remote observations of the state of the atmosphere have considerably expanded in recent years. In particular, a number of satellites have been launched into the Earth’s orbit with the main purpose of measuring the vertical distribution of various chemicals in the atmosphere: ozone, nitrogen dioxide, methane, carbon monoxide, water vapor, aerosols, etc. These are SCISAT-1 (Science Satellite), 2003, Canada; Aura, 2004, United States; GOSAT (Greenhouse Gases Observing Satellite), 2009, Japan; GOSAT-2, 2018, Japan; etc. [12, 15, 22, 26].

In this regard, one of the most promising instruments for remote studies of the chemical composition of the Earth’s atmosphere is the TROPOMI (Tropospheric Monitoring Instrument) spectrometer installed on the Sentinel-5P satellite (the launch date was October 13, 2017). TROPOMI is intended for daily global monitoring of the atmosphere within the EU Copernicus program [25]. The integration of the TROPOMI data archive into the Google Earth Engine cloud platform significantly expands the analytical capabilities for applying the information [14].

The purpose of this study is to assess air pollution within the territory of the 20 largest Russian cities based on TROPOMI data for 2019–2020.

MATERIALS AND METHODS

In terms of space, a city is a combination of residential buildings, industrial facilities, and open spaces (water bodies, green areas) [4]. This study considers the 20 largest cities in Russia (as of January 1, 2020) [9], which collectively contain 25.64% of the country’s population (Table 1). The administrative boundaries of cities in vector form were obtained by digitizing master plans (Federal State Information System for Territorial Planning, Ministry of Economic Development of the Russian Federation, data for 2020; https://fgistp.economy.gov.ru). The use of relatively stable administrative boundaries makes it possible to compare the year-to-year level of pollution in individual cities and analyze the causes of negative or positive changes in detail. The percentage of urbanized areas is calculated as the ratio of the built-up area to the official area, while the built-up area was determined based on the OpenStreetMap (https://www.openstreetmap.org) data as a total 200-m buffer zone around existing buildings and structures.

Table 1. Largest cities in Russia by population (as of January 1, 2020)
Full size table

The TROPOMI spectrometer takes pictures in ultraviolet (UV), visible (VIS), near-infrared (NIR) and short-wave-infrared (SWIR) ranges of the electromagnetic spectrum. The data make it possible to determine the total content in the vertical column of the troposphere of such chemicals as ozone (О3, mol/m2), nitrogen dioxide (NО2, mol/m2), sulfur dioxide (SО2, mol/m2), carbon monoxide (CO, mol/m2), methane (CH4, ppbV), and formaldehyde (HCHO, mol/m2). In addition, the aerosol pollution index is determined. The capture band is 2600 km wide, and the spatial resolution varies depending on the analyzed substance: 7 × 3.5 km for SO2, NO2, HCHO, and aerosols; 7 × 7 km for CO and CH4; and 28 × 21 km for O3 [25].

Access to TROPOMI data and basic analytical operations were carried out using the application programming interface (API) of the Google Earth Engine (GEE) cloud platform [14]. The TROPOMI data in the GEE are contained as thematic sets that correspond to individual chemical compounds (https:// developers.google.com/earth-engine/datasets/catalog/ sentinel-5p). Each set is represented by a data cube of the L3 processing level: for the convenience of analysis, the L2 level data is divided into a regular grid for each orbit without aggregation by components. The default quality thresholds used in the resulting datasets are 80% for the aerosol index, 75% for NO2, and 50% for all other substances. Thus, there are wide possibilities of statistical and spatial analysis available for each pixel in the data cube without the need for additional preparation of the source data.

The TROPOMI data were analyzed in GEE using the built-in Code Editor interface with JavaScript scripts written on the basis of client libraries to perform the following operations:

• loading data filtered by date (ee.Filter.calendarRange(), ee.Filter.date()), geographic area (ee.FeatureCollection.filterBounds()), and condition (ee.Filter.eq(), ee.Filter.gte(), ee.ImageCollection.select());

• aggregation of the given properties in the collection, calculation of the mean, minimum, and maximum values of the selected property (ee.ImageCollection.max(), ee.Reducer.mean(), ee.FeatureCollection, aggregate_min(), ee.FeatureCollection.aggregate_max());

• zonal statistics within the boundaries of selected cities (ee.Image.reduceRegions());

• combining images of different times into one image (ee.Image.cat());

• arithmetic calculations: addition, subtraction, multiplication, division (ee.Number.subtract(), ee.Number.add(), ee.Number.multiply(), ee.Number.divide());

• auxiliary operations (ee.Feature.setGeometry(), ee.Feature.set(), ee.Feature.get(), ee.Algorithms.If(), ee.FeatureCollection.map(), print()Map.addLayer()); and

• export of results (Export.table. toDrive(), Export.image.toDrive()).

The final statistics was calculated in MS Excel 2019. Thematic maps were made in QGIS v.3.20.

To estimate the integral air pollution in cities in relation to TROPOMI remote measurements, the universal normalized pollution index TAQI (Tropomi Air Quality Index) was developed and calculated based on the following formula:

$$\text{TAQI} = \sum\limits_{i = 1}^n {{{c}_{i}}{{a}^{i}}} ,$$

where i is the pollutant number, ci is the normalized content of the ith pollutant [0; 1], n is the number of indicators involved in the calculation, and ai is the dimensionless coefficient that takes into account the hazard class of the substance.

In this study, the TAQI index takes into account the following pollutants: carbon monoxide (CO), formaldehyde (HCHO), nitrogen dioxide (NO2), sulfur dioxide (SO2), and aerosols (aerosol pollution index).

The pollutant content is normalized in the range [0; 1] to integrate absolute values for various pollutants that have significantly different dimensions.

The coefficient that takes into account the hazard class according to the requirements of Roshydromet [6] is determined as follows: 1.3 corresponds to hazard class 2 (highly hazardous substances) used for HCHO, 1.0 corresponds to hazard class 3 (moderately hazardous substances) used for NO2 and SO2, and 0.85 corresponds to hazard class 4 (low-hazardous substances) used for CO.

For ease of comparison of results for different years, the TAQI values are normalized to a scale from 0 to 100. Five main categories can be conditionally distinguished in the interpretation of the TAQI values:

• category 1: TAQI value from 0 to 10, insignificant level of air pollution, almost a background state;

• category 2: TAQI value from 11 to 20, low level of air pollution;

• category 3: TAQI value from 21 to 30, elevated level of air pollution;

• category 4: TAQI value from 31 to 50, high level of air pollution; and

• category 5: TAQI value of 51 or higher, very high levels of air pollution.

RESULTS OF THE STUDY

The main limitation for the use of the TAQI index is the simultaneous availability of the TROPOMI data for all pollutants. For this reason, the completeness of the initial data for each city was assessed at the first stage of the study according to two criteria: the coverage area and the frequency of observations. The minimum threshold values were set to 90% of the city’s area coverage and three measurements per month. The assessment results showed (Table 2) that the full set of measurements for all the cities is available only from April to September of each year. In November, December, and January, due to lack of sunlight, measurements of no more than four substances are available. In February, March, and October, the availability of measurements is uneven and depends on climatic conditions: cities in the southern regions (Volgograd, Krasnodar, etc.) have the best indicators. Average monthly data availability (at threshold level) for selected chemicals for 2019 and 2020 was 99.9% for the aerosol pollution index, 83.9% for carbon monoxide, 69.3% for formaldehyde, 59.1% for nitrogen dioxide, and 43.1% sulfur dioxide; the density of observations for individual months is shown in Tables 3 and 4.

Table 2. Quantity of chemicals measured simultaneously by TROPOMI in the cities under consideration
Full size table
Table 3. Density of TROPOMI observations for individual months in 2019 (covering at least 90% of the cities’ area)
Full size table
Table 4. Density of TROPOMI observations for individual months for 2020 (covering at least 90% of the cities’ area)
Full size table

The calculation of mean monthly TAQI values for 2019–2020 (Tables 5, 6) shows that the highest pollution was observed in October 2019 and 2020 (47 and 44, respectively). The maximum value is noted in September 2019 and 2020 in Moscow (77 and 67, respectively). June 2019 is characterized by the lowest TAQI values (21). The minimum TAQI values are recorded in June 2019 for Omsk (9) and August 2019 for Izhevsk (12). A general seasonal dynamics can be noted: the level of pollution ubiquitously decrease in June and July and increase in September and October.

Table 5. Distribution of the TAQI index by months for 2019
Full size table
Table 6. Distribution of the TAQI value by months for 2020
Full size table

If TAQI is calculated for a common period of available data for all cities (from April to September), almost half of the cities (nine) had an elevated level of pollution in 2019; the number of cities with a high level of pollution is significantly smaller (five) (Fig. 1). In 2020, the situation is slightly different: the number of cities with high levels of pollution decreases (to four), while the number of cities with low levels of pollution increases (12) and there is one city with a very high level of pollution. In 2019, the highest air pollution is noted for Novosibirsk and Krasnoyarsk (TAQI values of 49 and 44, respectively); in 2020, for Moscow, Krasnodar, and Chelyabinsk (59, 41, 41). The minimum pollution level is noted in 2019 for Kazan, Voronezh, and Samara (13, 14, 14); in 2020 for Volgograd, Perm, and Omsk (12, 13, and 14).

Fig. 1.
figure 1

Distribution of cities by pollution level.

Full size image

Regarding the dynamics of the TAQI value for 2019–2020 (observation period from April to September), several groups of cities can be distinguished, in which the following oppositely directed trends are observed:

• an increase in TAQI values in the range from 10 to 16 (Omsk, Yekaterinburg, Moscow, Voronezh);

• an increase in TAQI values in the range from 1 to 7 (Saratov, Tyumen, Perm, Samara, Ufa, St. Petersburg);

• a decrease in TAQI values in the range from 3 to 9 (Novosibirsk, Kazan, Krasnodar, Nizhny Novgorod, Chelyabinsk, Togliatti); and

• a decrease in TAQI values in the range from 11 to 29 (Rostov-on-Don, Krasnoyarsk, Izhevsk, Volgograd).

It can be noted that Voronezh is characterized by a consistently low level of pollution (an increase in TAQI from 14 to 16), while the level of pollution for Moscow within the boundaries before July 1, 2012 is consistently high (increase in TAQI from 43 to 59).

In general, formaldehyde (35.7%) and nitrogen dioxide (26.4%) make the main contribution to the structure of pollutants (Fig. 2) for the cities under consideration in 2019–2020. Sulfur dioxide also accounts for a significant share (16.4%). The contribution of carbon monoxide and aerosols averages 10.8 and 10.6%, respectively. At the same time, there is a significant variability in the combinations of pollutants in cities. For example, Yekaterinburg leads in the share of aerosols (up to 53.1% in 2020), Omsk in the share of carbon monoxide (36.2% in 2019), Izhevsk in the share of formaldehyde (74% in 2020), Krasnoyarsk in the share of nitrogen dioxide (63% in 2020), and Saratov in the share of sulfur dioxide (45.1% in 2020).

Fig. 2.
figure 2

Share of various pollutants in the composition of atmospheric air pollution in the cities under consideration.

Full size image

RESULTS AND DISCUSSION

Space monitoring methods for assessing the content of pollutants in the atmosphere have been used in world practice for more than 45 years [1, 20, 27]. Studies of urban areas often estimate the concentrations of individual components, such as nitrogen dioxide [11, 23]. A similar approach is practiced in Russia; examples include remote sensing of nitrogen dioxide in the air of Moscow and St. Petersburg [8, 16], nitrogen dioxide in the air of Moscow and Moscow oblast [7], or sulfur dioxide in the air over the Norilsk industrial zone [3]. One current trend is considered to be an assessment of the dynamics of reducing pollutant emissions in urban agglomerations associated with the COVID-19 pandemic [18].

An important advantage of the integral atmospheric pollution index TAQI based on the TROPOMI data is the possibility of both real-time and retrospective analysis using the developed tools, as well as a complete simultaneous coverage of almost any area with a high (almost daily) frequency of observations. Among the drawbacks of remote analysis using TAQI, one can note the low frequency of TROPOMI observations during the period of lack of sunlight (from October to March), a limited set of pollutants available for monitoring, and dependence on the presence of clouds. The latter drawback is to a certain extent offset by a decrease in the potential for atmospheric pollution during the period of cyclonic activity.

The comparison of the results with official estimates of the level of pollution based on the complex atmospheric pollution index (API) (Table 7) shows a rather low degree of convergence. Apparently, this is due to the following differences:

Table 7. Comparison of air pollution levels based on TAQI and API
Full size table

• the area of the objects under consideration: the placement of ground sensors does not always takes into account the background territories that are adjacent to the residential and industrial zones and officially included in the administrative unit; this can potentially give higher API values in large cities with a compact urbanized zone;

• the density and frequency of observations: ground-based sensors are not able to cover all sources of pollution and provide a uniform spatial distribution of the observation network, which can potentially underestimate the API values in large cities; and

• the composition of the measured substances: the calculation of the API, in particular, involves benzo(a)pyrene, which is characterized by increased toxicity (hazard class 1), which makes a significant contribution to the total API value even at low concentrations.

Nevertheless, the results of TAQI calculations for 2019–2020 (Figs. 3, 4) allow us to identify some patterns of air pollution in the largest cities of Russia, namely:

Fig. 3.
figure 3

Air pollution in the largest cities of Russia for April–September 2019. The map displays the information on the calculated index within the boundaries of Moscow after July 1, 2012 ([9]; Russian Oikonyms Dataset, https://yastatic.net/s3/milab/ 2021/toponyms/data/RussianOikonymsDataset.zip).

Full size image
Fig. 4.
figure 4

Air pollution in the largest cities of Russia for April–September 2020. The map displays the information on the calculated index within the boundaries of Moscow after July 1, 2012 ([9]; Russian Oikonyms Dataset, https://yastatic.net/s3/ milab/2021/toponyms/data/RussianOikonymsDataset.zip).

Full size image

• the contribution of natural factors to pollution can be noted on the example of the increase in the content of aerosols in certain months for various cities; the increase in this case is caused, as a rule, by natural fires (European part of Russia, Siberia) and dust storms (southern regions of Russia);

• anthropogenic factors associated with the COVID-19 pandemic have a dual manifestation: on the one hand, most cities show no noticeable decrease in pollution due to restrictions introduced in certain periods, with the exception of some cities (for example, Moscow in April 2020); on the other hand, there is an increase in the content of formaldehyde in a number of cities (Moscow, Ufa, Samara, Kazan) during the pandemic, which can be associated with the population refusing from public transport in favor of private cars; and

• the generally observed decrease in the level of pollution in summer and an increase in autumn can be explained by the vacation season and the summer season, when, first, there is a widespread and rather noticeable shift in the anthropogenic load to suburban areas (in summer) and, then, the restoration of economic activity in cities (in autumn).

The issue of determining the boundaries of urban settlements remains debatable. The approach adopted in this paper with the use of official administrative boundaries allows us to compare air quality from year to year, as well as correlate the share of urbanized (built-up) territories with the level of atmospheric air pollution. In this regard, a positive relationship between the level of urbanization and the level of pollution can be noted; there are cities that show an elevated pollution rate while the share of buildings within official boundaries is low (for example, Perm and Ufa in 2019 and Yekaterinburg in 2020).

The reliability of the proposed index is based on the validation of the original TROPOMI measurements [21]. Nevertheless, it is planned in further studies to carry out a more detailed comparison of the obtained values with the results of ground-based measurements of individual chemical compounds and aerosols, as well as with integral estimates of the pollution level obtained by traditional methods. In addition, it is planned to expand the list of Russian cities under study.

The proposed TAQI index organically fits into the multisensor paradigm for assessing the air quality of the urban environment [17, 29] and can be implemented as a useful addition to an existing ground monitoring system. At the same time, the tools for remote assessment of the level of air pollution developed during the study can be integrated into specialized information systems as their functional extension.

CONCLUSIONS

The study allows us to draw the following conclusions.

(1) The data from the TROPOMI target equipment (Sentinel-5P satellite) integrated into the Google Earth Engine cloud platform allow the assessment of urban air pollution by substances such as carbon monoxide (CO), formaldehyde (HCHO), nitrogen dioxide (NO2), sulfur dioxide (SO2), and aerosols (aerosol pollution index). The integral air quality index TAQI has been developed based on the TROPOMI data.

(2) The results of TAQI calculations for the 20 largest cities in Russia in terms of the number of cities in the period from April to September 2019 and 2020 (simultaneous availability of measurements of all five substances) showed that 15 cities have low or high levels of pollution. Formaldehyde (35.7%) and nitrogen dioxide (26.4%) make the main contribution to the structure of pollutants. Sulfur dioxide also accounts for a significant share (16.4%). The contribution of carbon monoxide and aerosols averages 10.8 and 10.6%, respectively.

(3) The quality of atmospheric air in cities is affected by both natural (fires, dust storms) and anthropogenic (seasonal migrations of the population, epidemiological restrictions) factors. The impact of the COVD-19 pandemic on reducing emissions is local; the impact is neutralized over time, and the initial level of pollution is quickly restored.

(4) The proposed TAQI index, including the tools developed for its calculation, can be used to assess air pollution for a territory of any area. Within the actively developing multisensor paradigm, space monitoring data are a useful information addition to existing ground-based measuring systems.