Air Quality and Industrial Emissions in the Cities of Kazakhstan

: Industrial emissions are of major concern, especially in developing countries. Hence, there is a need for studies that investigate the trends in industrial emissions in these countries. The purpose of this study is to discuss trends in industrial emissions in Kazakhstan and the air pollution level in its industrial cities. Data on emission limit values from the permitting documents of twenty-one power plants and nine metallurgical enterprises of Kazakhstan were analyzed. Eight cities (out of fourteen) had a “high” level of atmospheric air pollution according to the Air Pollution Index in 2019. Most of the considered enterprises increased their emission limit values compared to the previous permitting period. In some cities there is a lack of monitoring stations, indicating the need for improving the spatial coverage of the air quality monitoring network in the industrial cities of Kazakhstan. The location of industrial plants far outside the cities could reduce the exposure of the urban population to air pollution. Kazakhstan urgently needs to adopt stringent emissions standards for coal-ﬁred power plants and heavy industrial plants. The national air quality standards and deﬁnitions of air pollutants need to be updated based on the latest scientiﬁc knowledge.


Introduction
The industry is one of the major sources of air pollution and greenhouse gases (GHG) emissions [1]. Globally, there is an increasing trend of GHG emissions from industry [2]. Industrial GHG emissions have declined in OECD countries but have increased in the Asian region [2]. Geng et al. [2] highlight the need for studies to learn and share experiences for developing countries. Kazakhstan has experienced considerable economic growth in the last two decades, mainly due to the export of fossil fuels and metals, but it has also undergone many years of environmental degradation because of the poor management of its significant natural resources [3]. The carbon intensity of GDP in Kazakhstan (0.6 kg per PPP $ of GDP in 2016) is two times higher than the world average value (0.33 kg per PPP $ of GDP) and three times higher than that in the European Union (0.2 kg per PPP $ of GDP) [4]. During the time of the USSR, coal-fired power plants were built to meet the needs of heavy industry and due to the presence of coal mines in the northern, central, and eastern regions of Kazakhstan. In Kazakhstan, there were two industrial sectors, namely power generation and metallurgy, which are the major sources of emissions, as they are responsible for 37% and 30% of the country's gross industrial emissions, respectively [5].
Industrial plants in Kazakhstan are required to obtain a "permit for emissions" setting emission limit values (ELV) [6]. ELVs are determined with dispersion models based on the prerequisite for achieving "environmental quality standards" at the border of the sanitary protection zone and in nearby residential areas [6]. Such an approach generally leads to less stringent emission limits than ELVs, which are based on internationally defined best available techniques (BAT) for installations (e.g., EU Industrial Emissions Directive) [7]. In To assess the air pollution level, indices that are related to the MAC values are employed in Kazakhstan. The most important is the Air Pollution Index (API). For its calculation, the average values of the concentrations of various pollutants are divided by their respective MAC and benchmarked by a factor related to the MAC value of SO 2 . In this study, API values for the cities of Kazakhstan were obtained from the information bulletin on the state of the environment of the Republic of Kazakhstan for 2019 for the cities of Kazakhstan [33]. The quality in the city is estimated by four classes: Low, Increased, High, Very High (Table 2).

Description of Selected Enterprises and Their Permitting Documents
Permitting documents of industrial enterprises are taken from the reports of the state environmental expertise on the website of electronic licensing of Kazakhstan http:// elicense.kz/ (accessed on 23 April 2020). The ELVs (emission limit values) that are reported in the permitting documents of major industrial enterprises in the power and metallurgy sectors are analyzed. We evaluated the efficiency of environmental regulations and changes in recent years by comparing the existing (valid at the time of this study) with the previous (valid in the previous permitting period) ELVs of the major industrial enterprises.
The power plants selected for this study have a total available capacity of 12.9 GW, which is 68% of the total available capacity in the country (18.9 GW) (Appendix A, Table A1). Three major state district power stations called GRES are thermal power plants that produce mainly electricity. Aksu GRES, Ekibastuz GRES-1, and Ekibastuz GRES-2 have the available capacity of 2450 MW, 4000 MW, and 1000 MW respectively, representing 39% of the total available capacity in the country. The remaining eighteen power plants are combined heat and power plants (CHP), producing not only electricity but also serve as sources of heat in centralized heat supply systems of the respective cities. District heating systems based on coal CHPs are common in most of the cities in Kazakhstan.
All selected power plants use coal as a main source of fuel (except for Almaty CHP-1). In all power plants, coal is flared in the pulverized state; ignition is carried out with fuel oil. In all coal power plants of Kazakhstan (except for one station) emissions control devices for SO 2 and NO 2 emissions are not employed. In most of the power plants in Kazakhstan, a method for wet cleaning of the flue gas from Russian manufacturers named "emulsifiers" is used to control ash. In the permitting documents of many power plants in Kazakhstan, it is indicated that ash collection is represented by "emulsifiers" with an efficiency of about 99%. However, the collection efficiency of fine dust (PM 2.5 ) is not reported. More advanced pollution control devices for PM, such as fabric filters, are not yet employed. Information from approved ELVs from coal combustion for selected power plants is presented in Appendix A (Table A1).
Nine metallurgical enterprises produce a major share of the country's total metals output: 65% of ferroalloys, 100% of copper, 100% of iron and steel, 95% of zinc, 94% of lead. Information from approved ELVs for selected metallurgical plants is presented in Appendix A (Table A1).

Description of the Selected Cities
The cities selected in this study have major industrial plants: either coal power plants or metallurgical plants. Fourteen cities were selected in this study. The sum of gross emissions from stationary sources of the fourteen cities is 1369 thousand tons, whereas the total national emissions from stationary sources is 2483 thousand tons. Thus, the selected cities represent 55% of national emissions from stationary sources, while representing only 30% of the population of the country. Two cities have a population of more than 1 million people (Nur-Sultan and Almaty), five cities between 300 thousand to 500 thousand people, and seven cities have less than 200 thousand people (Table 3).

Air Quality in the Cities of Kazakhstan
Eight cities had a "high" level of air pollution according to the Air Pollution Index data (more or equal to seven) in 2019: Nur-Sultan, Karaganda, Temirtau, Aktobe, Balkhash, Ust-Kamenogorsk, Zhezkazgan, Almaty ( Figure 1). The two most populous cities of Kazakhstan (Almaty, Nur-Sultan) were highly polluted. Two cities (Pavlodar, Semei) had an "increased" level of air pollution. Only four of selected cities of Kazakhstan had a "low" level of air pollution.

Air Quality in the Cities of Kazakhstan
Eight cities have a "high" level of air pollution according to the Air Pollution Index data (more or equal to seven) in 2019: Nur-Sultan, Karaganda, Temirtau, Aktobe, Balkhash, Ust-Kamenogorsk, Zhezkazgan, Almaty ( Figure 1). Two most populous cities of Kazakhstan (Almaty, Nur-Sultan) are highly polluted. Two cities (Pavlodar, Semei) have an "increased" level of air pollution. Only four cities of selected cities of Kazakhstan had a "low" level of air pollution. Over the recent years (2015-2019), deterioration in air quality is observed in seven cities ( Figure 1). Air quality worsened in large cities: in Almaty API increased from 7.6 in 2015 to 8 in 2019, in Nur-Sultan from 4.2 to 7. An increase in API is also observed in Temirtau (API increased from 7.9 to 9) and in Zhezkazgan (an increase in the IPA from 7.  Over the recent years (2015-2019), deterioration in air quality was observed in seven cities ( Figure 1). Air quality worsened in large cities: in Almaty the API increased from 7.6 in 2015 to 8 in 2019, in Nur-Sultan from 4.2 to 7. An increase in API was also observed in Temirtau (API increased from 7.9 to 9) and in Zhezkazgan (an increase in the IPA from 7. The average annual concentration of TSP is extremely high in the cities in Kazakhstan: in ten cities (out of twelve), it was higher than 100 µg m −3 ( Figure 3). The WHO has not established limit values for TSP, and the average annual concentrations of pollutants were compared with the nationally adopted average daily MAC only. Five cities exceeded the nationally adopted MAC for TSP (150 µg m −3 ). In Temirtau, the exceedance of the national MAC values was recorded for all the analyzed pollutants (TSP, NO2, and SO2). Ekibastuz and Pavlodar, which rank second and third in gross emissions, have lower TSP, NO2, and SO2 concentrations than other cities. This could be explained by the industrial enterprises being located far from the air quality monitoring stations, meteorological conditions, possible mistakes in the air quality data, and other factors. For example, the Pavlodar aluminum smelter with CHP-1 is not covered by the air quality monitoring network, and the largest power plants in Aksu and Ekibastuz are 3 and 30 km, respectively, away from the monitoring site. The relatively low air pollution levels in Ekibastuz, Aksu, and Rudnyi are partly due to the presence of only 1-3 monitoring stations. The World Bank [13] recommended that Kazakhstan extend its air quality monitoring network, especially in industrial cities.
The average annual concentrations of NO2 and SO2 in 2019 in fourteen cities in Kazakhstan are presented in Figure 3. Exceedance of the nationally established daily MAC values for SO2 (50 µg m −3 ) and NO2 (40 µg m −3 ) was recorded in three cities: Almaty, Ust-Kamenogorsk, and Temirtau. The levels in these cities are clearly beyond the acceptable range, as they are higher than those in other cities in Kazakhstan. These three cities have coal-fired power plants, and two of them (Ust-Kamenogorsk and Temirtau) also have metallurgical enterprises. If we compare the concentration values with the WHO standard  The average annual concentration of TSP was extremely high in the cities in Kazakhstan: in ten cities (out of twelve), it was higher than 100 µg m −3 ( Figure 3). The WHO has not established limit values for TSP, and the average annual concentrations of pollutants were compared with the nationally adopted average daily MAC only. Five cities exceeded the nationally adopted MAC for TSP (150 µg m −3 ). In Temirtau, the exceedance of the national MAC values was recorded for all the analyzed pollutants (TSP, NO 2 , and SO 2 ). Ekibastuz and Pavlodar, which rank second and third in gross emissions, had lower TSP, NO 2 , and SO 2 concentrations than other cities. This could be explained by the industrial enterprises being located far from the air quality monitoring stations, meteorological conditions, possible mistakes in the air quality data, and other factors. For example, the Pavlodar aluminum smelter with CHP-1 is not covered by the air quality monitoring network, and the largest power plants in Aksu and Ekibastuz are 3 and 30 km, respectively, away from the monitoring site. The relatively low air pollution levels in Ekibastuz, Aksu, and Rudnyi were partly due to the presence of only 1-3 monitoring stations. The World Bank [13] recommended that Kazakhstan extend its air quality monitoring network, especially in industrial cities.
The average annual concentrations of NO 2 and SO 2 in 2019 in fourteen cities in Kazakhstan are presented in Figure 3. Exceedance of the nationally established daily MAC values for SO 2 (50 µg m −3 ) and NO 2 (40 µg m −3 ) was recorded in three cities: Almaty, Ust-Kamenogorsk, and Temirtau. The levels in these cities were clearly beyond the acceptable range, as they were higher than those in other cities in Kazakhstan. These three cities have coal-fired power plants, and two of them (Ust-Kamenogorsk and Temirtau) also have metallurgical enterprises. If we compare the concentration values with the WHO standard (24-h limit value for SO 2 = 20 µg m −3 ), the situation becomes more dramatic, as the annual average concentration was exceeded in seven cities in Kazakhstan-by a factor of 6. To compare variations in concentrations in winter and summer, we used the monthly average concentrations of TSP, SO 2 , and NO 2 in December, January, February, June, July, and August in 2019 (Table 4). Transitional months were excluded, as the heating season varies greatly between cities. Differences between winter and summer were significant, and the differences varied by city and pollutant. To compare variations in concentrations in winter and summer, we used the monthly average concentrations of TSP, SO2, and NO2 in December, January, February, June, July, and August in 2019 (Table 4). Transitional months were excluded, as the heating season varies greatly between cities. Differences between winter and summer were significant, and the differences varied by city and pollutant.
Seasonal variations in the air pollution can be highly affected by meteorology. Two meteorological parameters (temperature and relative humidity) by months and by cities is presented in the Appendix A, Tables A2 and A3. Climate of Kazakhstan is continental, with cold winters and hot summers. Average temperature significantly varies from north to south regions of Kazakhstan. The average winter temperature in the selected cities located in the north and central areas was −12-14 °C and it was −4 °C in Almaty (only city considered in this study which is located in the south). Heating is a basic need for survival in Kazakhstan, therefore amount of fuel consumption is higher in the winter time due to additional demand for heating. In all selected cities there are coal-fired CHPs, except for three cities.
The concentration of TSP increased in seven cities in the summer compared to the winter. This is unexpected result because the level of emissions from CHPs and residential heating is generally higher in the winter-time. On the other hand, the level of emissions from metallurgical enterprises and transportation could be stable throughout the year. CHPs in Almaty and Nur-Sultan annually burn around 3.7 and 3.2 million tons of coal respectively. The operation of the CHPs depends on the outside temperature. As an example, the average daily coal consumption in Almaty CHP-2 was more than twice as high in the winter period compared to summer, 9424 tons/day in January 2019 and 3999 tons/day in July 2019. Stable atmospheric conditions and temperature inversions could lead to winter-time episodic pollution. In Almaty the level of TSP is 17.6% lower in winter compared to summer in 2019 (Table 4). On the contrary, independent air quality monitoring from airkaz.org website depicts that average level of PM2.5 concentration in winter in Almaty was 110 µg m −3 , which is nearly 6 times higher than in summer (18 µg m −3 ). Such differences in seasonal variations of TSP and PM2.5 from two sources: official Kazhydromet's TSP values and civil activists' PM2.5 measurements needs further investigation. TSP values cannot be directly compared to PM2.5, the ratio PM2.5/TSP can vary by seasons,   Seasonal variations in air pollution can be highly affected by meteorology. Two meteorological parameters (temperature and relative humidity) by months and by cities are presented in Appendix A, Tables A2 and A3. The climate of Kazakhstan is continental, with cold winters and hot summers. The average temperature significantly varies from the northern to southern regions of Kazakhstan. The average winter temperature in the selected cities located in the north and central areas was −12-14 • C and it was −4 • C in Almaty (the only city considered in this study which is located in the south). Heating is a basic need for survival in Kazakhstan, therefore the amount of fuel consumption is higher in the winter time due to the additional demand for heating. In all selected cities there are coal-fired CHPs, except for three cities.
The concentration of TSP increased in seven cities in the summer compared to the winter. This is an unexpected result because the level of emissions from CHPs and residential heating is generally higher in the winter-time. On the other hand, the level of emissions from metallurgical enterprises and transportation could be stable throughout the year. CHPs in Almaty and Nur-Sultan annually burn around 3.7 and 3.2 million tons of coal respectively. The operation of the CHPs depends on the outside temperature. As an example, the average daily coal consumption in Almaty CHP-2 was more than twice as high in the winter period compared to summer, 9424 tons/day in January 2019 and 3999 tons/day in July 2019. Stable atmospheric conditions and temperature inversions could lead to winter-time episodic pollution. In Almaty the level of TSP was 17.6% lower in winter compared to the summer in 2019 (Table 4). On the contrary, independent air quality monitoring from the airkaz.org website depicted that the average level of PM 2.5 concentration in winter in Almaty was 110 µg m −3 , which was nearly 6 times higher than in the summer (18 µg m −3 ). Such differences in seasonal variations of TSP and PM 2.5 from two sources: official Kazhydromet's TSP values and civil activists' PM 2.5 measurements need further investigation. TSP values cannot be directly compared to PM 2.5 , the ratio PM 2.5 /TSP can vary by seasons, with fine particles originating from the combustion of fuels and coarse particles from soils and, road dust resuspension. Another possible explanation could be that in the cities under consideration, the average relative air humidity in the summer months was 16-47% lower than in the winter (Table A3). At high air humidity, the process of deposition of fine particles could occur faster due to their coagulation.
On the contrary, SO 2 declined in the summertime compared to the winter period in eleven cities. NO 2 concentration was lower in the summertime in eight cities. Higher levels of NO 2 and SO 2 levels in the winter time were not a surprise, especially in the areas with coal-fired CHPs and households burning coal for heating.
No data is available on the industrial activity by seasons; therefore explaining the factors contributing to the seasonal variations needs further investigation.

Emissions Trends from Stationary Sources
Between 2009 and 2018, NO x , CO, and SO 2 emissions increased significantly by 32%, 10%, and 8%, respectively (Table 5). Paradoxically, there was a decline in TSP emissions by 32% over this period but an increase in inorganic dust (SiO 2 70-20%) by 67%. This could be partially explained by the fact that, in Kazakhstan, when coal is burned at power plants methodologies prescribe that ash dust be standardized as inorganic dust (SiO 2 70-20%). The increase in the emissions of inorganic dust (SiO 2 70-20%) in Kazakhstan could be due to the increase in coal consumption at coal-fired power plants over the same period. Emissions of TSP in Kazakhstan are estimated mainly at metallurgical enterprises and in the construction industry. The decline in the TSP emissions from stationary sources for the period 2009-2018 may be explained by the fact that some industrial plants started to report TSP as inorganic dust (SiO 2 70-20%), but this assumption needs further verification. Kazakhstan inherited a system for setting emissions and hygienic standards for pollutants from the Soviet Union. There was no separation of dust into PM 2.5 and PM 10 , therefore MACs were developed for certain types of dust, typical for the main types of industries. No attention was paid to the size of particles. The concept of PM 2.5 and PM 10 appeared in the legislative documents of Kazakhstan only in 2012, but there are still no requirements for reporting and estimating dust by particle size. The concept of inorganic dust with different contents of silicon dioxide is used only in post-Soviet countries. In the sanitary rules of Kazakhstan, an explanation is given of the cases to normalize which types of dust: In fact, when industrial plants report emissions of dust, the composition of the dust is not checked by the authorities. All these types of dust affect the concentration of TSP, PM 2.5 , and PM 10 in the ambient air. To eliminate confusion, when standardizing dust, which is currently named as "inorganic dust (SiO 2 70-20%)", "inorganic dust (SiO 2 > 70%)", "inorganic dust (SiO 2 < 20%)", "coal ash from power plants" and "TSP" should be unified in accordance with international definitions of pollutants. Those definitions of dust (inorganic dust (SiO 2 70-20%), inorganic dust (SiO 2 > 70%), inorganic dust (SiO 2 < 20%), coal ash from power plants) are outdated, as they have been used since the USSR period.

Links between Air Quality in the Cities and Industrial Emissions
To establish links between air quality in the cities with the activity of the industrial plants, the available data on air pollution levels (API) were compared with the data on industrial emissions. Emissions from stationary sources by cities reported in the Statistical Publication "Environmental Protection and Sustainable Development of Kazakhstan" [9,10] were employed. Gross emissions stationary sources were used because emissions by each pollutant by cities are not available.
For fourteen cities, there was no correlation (R 2 = 0.0256, p = 0.58) found between API level and industrial emissions (Figure 4). In Kazakhstan, the availability of monitoring systems is insufficient; particularly in small industrial cities. In Ekibastuz, Aksu, and Rudnyi there are only one to three monitoring stations. Pavlodar aluminum smelter is not covered by the monitoring system of Pavlodar city. In Aksu and Ekibastuz, monitoring stations are located at a distance of 3 and 30 km from the enterprises.

Links between Air Quality in the Cities and Industrial Emissions
To establish links between air quality in the cities with the activity of the industrial plants, the available data on air pollution levels (API) were compared with the data on industrial emissions. Emissions from stationary sources by cities reported in the Statistical Publication Environmental protection and sustainable development of Kazakhstan [9,10] were employed. Gross emissions stationary sources were used because emissions by each pollutant by cities are not available.
For fourteen cities, there was no correlation (R 2 = 0.0256, p = 0.58) found between API level and industrial emissions (Figure 4). In Kazakhstan, the availability of monitoring systems is insufficient; particularly in small industrial cities. In Ekibastuz, Aksu, and Rudnyi there are only one to three-stations. Pavlodar aluminum smelter is not covered by the monitoring system of Pavlodar city. In Aksu and Ekibastuz, monitoring stations are located at a distance of 3 and 30 km from the enterprises. If we exclude four cities (Pavlodar, Ekibastuz, Aksu, Rudnyi) where industrial plants are located outside of the city and where there is a lack of monitoring stations, the situation is dramatically different ( Figure 5). API level and gross emissions were correlated (R 2 = 0.4791, p = 0.027) for the remaining ten cities, although the correlation was not strong. This could indicate a contribution of industrial emissions in the air pollution of the cities.
The results also showed the importance of high spatial coverage of the air quality monitoring network, especially in polluted industrial cities. The World Bank (2013) also highlighted the lack of an air quality monitoring network in the industrial cities of Kazakhstan [13]. Despite the high level of industrial emissions in Aksu and Ekibastuz cities (more than 200 thousand tons), the API level was low (API two and three), possibly because the industrial plants are located outside the cities. Results indicated that the location of industrial plants far outside the cities could reduce the exposure of the population of the cities to air pollution.
If we exclude four cities (Pavlodar, Ekibastuz, Aksu, Rudnyi) where industrial plants are located outside of the city and where there is a lack of monitoring stations, the situation is dramatically different ( Figure 5). API level and gross emissions were correlated (R 2 = 0.4791, p = 0.027) for the remaining ten cities, although the correlation was not strong. This could indicate a point towards a contribution of industrial emissions in the air pollution of the cities. The results also show the importance of high spatial coverage of the air quality monitoring network, especially in polluted industrial cities. The World Bank (2013) also highlighted the lack of an air quality monitoring network in the industrial cities of Kazakhstan [13]. Despite the high level of industrial emissions in Aksu and Ekibastuz cities (more than 200 thousand tons), the API level was low (API two and three), possibly because the industrial plants are located outside the cities. Results indicate that the location of industrial plants far outside the cities could reduce the exposure of the population of the cities to air pollution.

Other Possible Factors Contributing to Air Pollution
Topography and meteorological factors could play an important role in air pollution. Almaty and Ust-Kamenogorsk cities are located in the bowl of mountain ranges. It causes a large number of windless days. The rest of the considered industrialized cities are located mainly in the steppe zone.
Winter smog episodes in Nur-Sultan have common meteorological conditions and anticyclonic conditions (with low wind speed, high ground-level pressures, freezing weather conditions with the air temperature reaching −30 °C at night and −20 °C in the daytime over several days) [12].
There are also natural sources of air pollution, such as dust blown from soils. Desert and semi-desert areas of Kazakhstan are mainly located in the areas of the Caspian Sea and southern parts of Kazakhstan. Issanova and Abuduwaili [36] and Nobakht et al. [37] note that sandstorms in Kazakhstan are common in South Kazakhstan (the Syr Darya and Ile river valleys), the southern part of Lake Balkhash. Zhang et al. [38] find that the main

Other Possible Factors Contributing to Air Pollution
Topography and meteorological factors could play an important role in air pollution. Almaty and Ust-Kamenogorsk cities are located in the bowl of mountain ranges. It causes a large number of windless days. The rest of the considered industrialized cities are located mainly in the steppe zone.
Winter smog episodes in Nur-Sultan have common meteorological conditions and anticyclonic conditions (with low wind speed, high ground-level pressures, freezing weather conditions with the air temperature reaching −30 • C at night and −20 • C in the daytime over several days) [12].
There are also natural sources of air pollution, such as dust blown from soils. Desert and semi-desert areas of Kazakhstan are mainly located in the areas of the Caspian Sea and southern parts of Kazakhstan. Issanova and Abuduwaili [36] and Nobakht et al. [37] note that sandstorms in Kazakhstan are common in South Kazakhstan (the Syr Darya and Ile river valleys), the southern part of Lake Balkhash. Zhang et al. [38] found that the main source of Aeolian Dust in Central Asia is the Aral Sea (southwestern Kazakhstan), which has significantly dried up since the 1960s. The selected cities in this study are located in the northern parts of Kazakhstan (except for Almaty) in the steppe and forest-steppe zones and those areas could be less affected by frequent sandstorms.
Transport could be one of the important sources of air pollution, particularly in densely populated large cities. The number vehicles per 100 persons increased in Kazakhstan from 4.7 units in 1990 to 19.2 units in 2019. The rapid increase in the number of passenger cars can be attributed to inefficient public transport and a lack of alternatives for the population. None of the cities in Kazakhstan have high-speed public transport modes (e.g., metro, LRT). There is a metro system in Almaty, but it has only 9 stations with 11.5 km length. Two densely populated cities (Almaty and Nur-Sultan) have the highest number of registered vehicles: 514 and 300 thousand vehicles respectively [39].
Selected cities, except for Almaty and Nur-Sultan, are small industrial cities. Seven cities (out of fourteen) have a low population (less than 179 thousand people) and have less than 41 thousand registered vehicles. The number of registered vehicles by cities was compared with the API in the selected cities ( Figure 6). Transport in Almaty and Nur-Sultan could have a contribution to air pollution deterioration, while in other cities non-transportation sources may dominate in the contribution to air pollution. The correlation between the number of registered vehicles and API was low (R 2 = 0.151, p = 0.189). Temirtau, Zhezkazgan, and Balkhash have high levels of air pollution, despite the very low population and fewer transport vehicles. Coal combustion for household heating is also an important source of air pollution. The households Survey showed that solid fuels (coal and firewood) are mainly used by rural households: 55% of rural households in Kazakhstan used solid fuels and only 17% of urban households used solid fuels [40]. The data on the share of households using coal by cities were compared to with the Air Pollution Index by cities (Figure 7). It could be observed that despite the low share of households using coal in Almaty (1%) and Nur-Sultan (6%), those cities remain highly polluted. Generally in the selected cities, the share of households using coal was less than 25%. There is no available data for small cities such as Semei, Rudnyi, Aksu, and Ridder. Out of fourteen cities, only two cities have an access to network gas: Aktobe and Almaty. Households could be significant sources of PM2.5, SO2, NOx emissions in the cities of Kazakhstan. However, in Kazakhstan, household emissions are not estimated, since, according to Article 10 of the Environmental Code [6], they are referred to as the general use of natural resources, which is free and not subject to regulation. The volume of household emissions is currently not established in Kazakhstan.
Emissions inventory and source-apportionment studies need to be conducted in the cities of Kazakhstan to quantify the impact of the sources of air pollution. The impact of Out of fourteen cities, only two cities have access to network gas: Aktobe and Almaty. Households could be significant sources of PM 2.5 , SO 2 , NO x emissions in the cities of Kazakhstan. However, in Kazakhstan, household emissions are not estimated, since, according to Article 10 of the Environmental Code [6], they are referred to as the general use of natural resources, which is free and not subject to regulation. The volume of household emissions is currently not established in Kazakhstan.
Emissions inventory and source-apportionment studies need to be conducted in the cities of Kazakhstan to quantify the impact of the sources of air pollution. The impact of topography and meteorology factors also needs further investigation.

Emission Limit Values of Selected Industrial Enterprises
In the nine major metallurgical enterprises considered, there was an overall increase in ELV by 124 thousand tons per year (23% increase). A reduction in the ELV was observed for only three enterprises, totaling 11 thousand tons, while the ELV increased by 135 thousand tons for five enterprises (Figure 8).     The used data proved the inefficiency of the air protection policy at major thermal power plants. Authorities permit enterprises to increase their consumption of coal without obliging them to implement special measures for reducing their emissions of harmful substances into the atmosphere.
Permitting documents often do not contain activity data. Therefore, it was not possible to compare emissions per unit of output with the best practices in the sector. Permits contain data on coal consumption by power plants per year (at maximum production output). Therefore, ELVs per unit of coal use have been analyzed in this study (Table A1). It could be observed that there were large variations in emissions per unit of coal consumed across power plants. As an example, allowed emissions, i.e., the ELV of SO 2 were between 9 to 34 kg/ton of oil equivalent (toe), allowed emissions of ash were between 3 to 24 kg/ton. Such variations in emissions intensities across power plants could be related to gaps in methodology for the determination of the ELV, which do not require emissions per unit of activity to be estimated and compared with best practices in the sector. There is no practice of benchmarking the environmental performance of industrial plants in terms of their emissions per unit of activity data due to the lack of data, lack of expertise, and absence of such a requirement in the legislation.

Discussion
Cities selected for this study were heavily polluted in 2019: the concentration of TSP exceeded 100 µg m −3 in ten (out of twelve) cities, and the concentration of SO 2 exceeded 20 µg m −3 in seven (out of fourteen) cities. Annual average SO 2 concentration levels were very high in Almaty (128 µg m −3 ), in Ust-Kamenogorsk (90 µg m −3 ), and in Temirtau (58 µg m −3 ). Eight cities (out of fourteen) have a "high" level of air pollution according to the Air Pollution Index (API). Winter to summer variation was considerable, and the differences varied by the cities and by the pollutants. Seasonal variations are hard to explain, due to the lack of data on industrial activity by seasons.
Non-industrial emission sources were also discussed in this study. Transport is unlikely to be a major contributor to air pollution in the selected cities, except for Almaty and Nur-Sultan.
For fourteen cities, there was no correlation found between air pollution level and industrial emissions. If we exclude four cities where industrial plants are located outside the city and where there is a lack of monitoring stations, API level and gross industrial emissions would correlate (R 2 = 0.4791), although not strongly. Results indicate the need for expanding the coverage of the air quality monitoring network in the industrial cities of Kazakhstan.
Despite a high level of industrial emissions in Aksu and Ekibastuz cities (more than 200 thousand tons), the API level was low, possibly because the industrial plants are located outside the cities. Results indicate that the location of the industrial plants far outside of cities could reduce the exposure of the population of the cities to air pollution.
In this study, it was demonstrated that the current emission permitting system does not provide an effective mechanism for emissions reduction. Most of the considered enterprises increased their emission limit values. Although the primary legislation, in theory, provides a framework for environmental protection, there are several loopholes in the secondary legislation: the methods for determining ELVs are linked to weak environmental quality standards, and there is a lack of monitoring and law enforcement. The increase in allowed emissions for coal power plants was explained by the growth of heat and electricity consumption and investment programs to increase production capacities. Power plants were not required to reduce emissions or introduce emission control technologies for reducing NO 2 and SO 2 emissions. Considering the quality of ambient air in the cities, it would be rational to assume that no increase in emissions should be permitted for coal power plants.
The air quality standards adopted in Kazakhstan are outdated. For example, enterprises can legally generate emissions that lead to a 500 µg m −3 concentration of SO 2 at the border of the sanitary protection zone, which is twenty-five times higher than the 24-h average WHO limit value (20 µg m −3 ). Outdated methodologies for estimating and presenting data, such as the summing of pollutants with different levels of toxicity or using several definitions for dust, cause information on emissions and air quality levels to remain hidden and unclear. There is an urgent need to harmonize environmental quality standards and methodologies in Kazakhstan to collect and present environmental data in accordance with the latest scientific knowledge and international practices.
Stringent emissions standards for coal-fired power plants have been successful in other countries. In 2014, China introduced "ultra-low" emission standards for coal-fired power plants; by 2017, almost all coal-fired power plants in China had installed NO x and SO 2 control devices [23]. Between 2014 and 2017, China's annual power emissions of SO 2 , NO x , and PM reduced substantially by 65%, 60%, and 72%, respectively [23]. Stringent emissions standards for coal-fired power plants must be introduced in Kazakhstan.
Kazakhstan is planning to transition to an emission permit system based on technologyintegrated OECD practices. With the constant pressure from powerful industrial associations, it remains uncertain whether authorities would be able to push stricter industrial emissions regulations. Authorities have to ensure that Kazakhstan's list of BAT and technical emissions standards correspond with those from Europe.
Due to scarcity of funding for expensive laboratory equipment and lack of capacity, source apportionment with chemical analysis of PM particles has not been conducted for the cities of Kazakhstan so far.