Spatiotemporal variations of atmospheric black carbon concentration and its correlation with meteorological and environmental factors in Xinjiang, China, from 2010 to 2022

MERRA-2 reanalysis data

The MERRA-2 dataset assimilates satellite aerosol (AOD) datasets based on the Goddard Earth Observing System (GEOS)-5 model and its data assimilation system (ADAS−5.12.4). The AOD datasets include the AOD dataset of the moderate resolution imaging spectrometer (MODIS), the corrected AOD dataset of the advanced very high resolution radiometer (AVHRR), the AOD dataset of the multi-angle imaging spectroradiometer (MISR), and the AOD dataset of the aerosol robotic network (AERONET). In addition, the MERRA-2 dataset also assimilates several remote sensing-retrieved datasets and the observation datasets from sondes and aircraft.

The MERRA-2 dataset includes 72 vertical layers from the surface to about 80 km and a variety of datasets, including clouds, radiation, hydrological cycles, and ozone, which is useful for comprehensive analysis of land, air, and sea. The spatial distribution data of atmospheric BC concentrations since 1980 (spatial resolution: 0.5° × 0.625°; format: NetCDF) could be extracted from the MERRA-2 dataset downloaded from NASA’s Earth Science Data website (https://science.nasa.gov/earth/data/). The data format conversion, cropping, and regional statistics were completed in ArcGIS 10.8, to obtain the average atmospheric BC concentrations in Xinjiang from 2010 to 2022.

Ground-based data

In this study, the accuracy of the atmospheric BC concentrations in Xinjiang of the MERRA-2 dataset was verified by using the ground-based observation data (October 1, 2010–December 31, 2022) provided by the Akdala Atmospheric Background Station (47°10′N, 87°58′E, 562 m a.s.l., Fig. 1). The sampling instruments (AE-31 and AE-33 aethalometers) were installed on the roof of a building of the Akdala Station. The inlet was 3 m away from the ground. Because the light absorption by BC accounts for 90%–95% of the total light absorption by aerosols, the aethalometers use the absorption characteristics of BC to measure the mass concentration of BC (Wang et al., 2012). The aethalometers could obtain BC concentration data of 7 wavelengths (370, 470, 520, 590, 660, 880, and 950 nm). The BC concentration measured at 880 nm was selected as the observed concentration (minute-level data), and then the monthly average BC concentrations were calculated. Since one grid of MERRA-2 BC data was enough to cover the observation station, the monthly average data of a MERRA-2 grid close to the observation station were selected for comparison, to verify the applicability and accuracy of the MERRA-2 BC data. The correlation was strong if the correlation coefficient (R) was in the range of 0.6–0.8, moderate if the R was in the range of 0.4–0.6, and weak if the R was in the range of 0–0.4.

Meteorological and environmental data

In this study, the meteorological data and environmental pollutant concentration data (January 1–December 31, 2022) provided by the Akdala Atmospheric Background Station were used to explore the correlation between atmospheric BC concentration and meteorological and environmental factors. The daily meteorological data included wind speed (m/s), land surface temperature (°C), humidity (%), snow depth (cm), sunshine hours (h), etc., and the daily environmental pollutant concentration data included PM₁₀ (ug/m³), PM_2.5 (ug/m³), NO₂ (ppb), SO₂ (ppb), CO (ppb), and O₃ (ppb) concentrations.

Research methodology

Evaluation of the applicability of MERRA-2 BC data in Xinjiang, China

The linear regression of the observed BC concentration at the Akdala Station and the BC concentration at the corresponding MERRA-2 grid (N = 131) showed that R = 0.49 (p < 0.05), the RMSE and MAE were smaller than those of China and northwest China, and the RAD was greater than that of northwest China (Cao, 2022) (Table 1). Therefore, the spatiotemporal variation characteristics of BC concentration in Xinjiang can be explored using the MERRA-2 reanalysis data for it has high reliability (RAD = 0.65). However, due to the impact of the spatial resolution of the reanalysis data, the observed BC concentrations were higher than the MERRA-2 data. Besides, because the Akdala Station is the atmospheric background station for northern Xinjiang and one of the regional background stations of the Global Atmosphere Watch (GAW) network of the World Meteorological Organisation (WMO), the BC concentrations are much lower than the average value in other regions (Cao, 2022).

Variation of atmospheric BC concentration from 2010 to 2022

The annual average BC concentration in Xinjiang was 195.40 ± 15.55 ng/m³, and the multi-year average change rate was −0.05% (Fig. 2). The annual average BC concentration was the highest (231.99 ng/m³) in 2012 and the lowest (174.50 ng/m³) in 2022. The variation range of the annual average BC concentration in Xinjiang was 174.50–231.99 ng/m³. The annual average BC concentration in Xinjiang is lower than the averages of western (0.38 µg/m³) and eastern (2.36 µg/m³) China from 1980 to 2020, and also significantly lower than that of the Beijing-Tianjin-Hebei region, Sichuan Basin, and northern Henan (average BC concentration from 1980 to 2020: 3.35–5.01 µg/m³) (Cao, 2022). The Beijing-Tianjin-Hebei region and northern Henan have developed industry and large populations. The changes in BC concentration are closely related to human and industrial activities. The Beijing-Tianjin-Hebei region and northern Henan are developed and have a large population, while Xinjiang has a vast area, sparse population, and low GDP, so the BC concentration in Xinjiang is low (Tiwari et al., 2013; Mao et al., 2016). The BC concentration in Xinjiang showed a decreasing trend in the past 13 years, indicating an improvement in regional BC pollution, but there was an increase in BC concentration in some regions of Xinjiang. Specifically, the BC concentration in Xinjiang showed a significant increasing trend from 2010 to 2012, especially, from 2011 to 2012, the BC concentration increased by 17.90%. From 2013 to 2015, there was a rapid downward trend, with a cumulative decline of 22.52%. After 2016, there was a fluctuating trend. Anthropogenic emissions are key factors affecting atmospheric BC concentrations. For example, Xinjiang’s energy industry output continued to increase from 2010 to 2012. Xinjiang’s GDP in 2012 reached 750.531 billion yuan, increasing by 40.0% compared with that in 2010. This growth rate was much higher than the average growth rate of 11.2% from 2010 to 2022. Rapid economic growth may lead to increasing atmospheric BC concentrations. It is worth noting that in 2020, due to the impact of COVID-19, there was a significant decrease in human activities, so the BC concentration in 2020 was 17.20% lower than that in 2019.

The seasonal variation of BC concentration in Xinjiang was obvious (Fig. 3A). The atmospheric BC concentration in winter (145.52 ± 39.31 ng/m³) was higher than that in autumn (124.95 ± 28.82 ng/m³), spring (74.05 ± 9.96 ng/m³), and summer (73.41 ± 5.69 ng/m³). With the increase of temperature, the number of windy weather increases in spring (the average number of days in a month with a maximum wind speed of >5 m/s in spring was 24.4 days), which is beneficial for pollutant diffusion, so the BC concentration in spring and summer is low. In addition to the influence of wind speed, the wet deposition of summer precipitation and high vegetation coverage was significantly strengthened (the average monthly precipitation was 11 mm in summer and 5.9 mm in winter). Thus, the atmospheric BC concentration remained low in summer.

Figure 3B shows the monthly average BC concentration in Xinjiang. The BC concentration was the highest in December (292.99 ± 197.82 ng/m³) and the lowest in June (135.69 ± 65.30 ng/m³). The BC concentration showed an upward trend from August, reached the highest value in December, decreased rapidly from January to June, and increased slightly in July. The seasonal variation of BC concentration in Xinjiang was consistent with that of other regions of China, i.e., Tianjin and Beijing (Qin et al., 2019; Zhao et al., 2021). The seasonal variation of BC emissions in certain regions is the main cause of the seasonal variation of atmospheric BC concentration in China (Zhao, 2015). The increase in the coal consumption for heating in winter, coupled with low wind speed, frequent emergence of inversion layer, and low boundary layer height, inhibit the diffusion of BC and aggravate atmospheric BC pollution. In summer, the temperature rises, convective weather occurs frequently, and weather destabilization intensifies. Therefore, there are good diffusion conditions in both the vertical and horizontal directions, resulting in a decrease in BC concentration. However, in autumn, activities such as large-scale crop harvesting increase the BC concentration.

Spatial distribution characteristics of atmospheric BC concentration

The analysis results of spatial distribution of annual average BC concentrations in Xinjiang from 2010 to 2022 using the MERRA-2 dataset (Fig. 4) showed that from 2010 to 2022, there were two areas with high BC concentrations: Urumqi-Changji-Shihezi urban agglomeration and Kashgar. The annual average BC concentration in Kashgar (515.18 ± 203.09 ng/m³) was higher than that in Urumqi-Changji-Shihezi urban agglomeration (444.84 ± 126.17 ng/m³), and much higher than that in Xinjiang (196.53 ± 57.90 ng/m³). The monthly average BC concentration in Kashgar was 2.09–2.77 times that of Xinjiang and 1.35–1.52 times that of Urumqi-Changji-Shihezi urban agglomeration (Fig. 5). The Urumqi-Changji-Shihezi urban agglomeration has developed industries and a dense population, with an urbanization rate of 86.98%. Although the Urumqi-Changji-Shihezi urban agglomeration only accounts for 5.7% of Xinjiang in area, it holds 83% of Xinjiang’s heavy industries, its coal consumption accounts for 60% of Xinjiang’s total, and the pollutant emissions account for one-third (Zhu et al., 2023). It was also found that in southern Xinjiang, the spatial distribution of BC concentration was highly consistent with the spatial distribution of industries and population. With Korla, Aksu, Kashgar and Hotan as important nodes, the high-BC-concentration areas surrounded the Tarim Basin. On the one hand, the special topographic and climatic conditions of the Tarim Basin leads to the formation of a unique layout of cities, and most of the human activities are concentrated in the cities, resulting in the highly consistency in distribution between high-BC-concentration areas and cities. On the other hand, the Tarim Basin is surrounded by mountains in the south, north, and west. The airflow from the east enters the Tarim Basin from the southern part of the Tianshan Mountains. After the airflow reaches the edge of the basin, it is not easy to spread due to the high mountains in the north, south, and west, resulting in the accumulation of pollutants in Aksu, Hotan, and Kashgar at the edge of the basin. Especially, Kashgar is located in the north of the westernmost Pamir Plateau, which is greatly affected by the frequent sand and dust in the Tarim Basin. Therefore, pollutants are easy to be transported to Kashgar, which is the main reason for its very high BC concentration.

From 2010 to 2022, the monthly average atmospheric BC concentration in Xinjiang showed significant seasonal differences (Fig. 6). The atmospheric BC concentration was high in autumn and winter and low in spring and summer. The area of high-BC-concentration areas (the northern slope of the Tianshan Mountains and the cities surrounding the Tarim Basin) was the largest in winter (November–February of the next year), and gradually decreased from November to February. The BC concentration also gradually decreased. The high-BC-concentration areas gradually decreased from March until it re-appeared in October. In spring and summer, all regions of Xinjiang have good pollutant diffusion conditions, with high wind speed and increased precipitation (the average number of days in a month with a maximum wind speed of >5 m/s in spring and summer was 23.5 days). It was worth noting that the seasonal atmospheric BC concentrations in the Urumqi-Changji-Shihezi urban agglomeration and Kashgar were higher than those in other regions of Xinjiang, and the monthly average concentrations were much higher than those of Xinjiang (Fig. 7). This indicates obvious spatial differences. Atmospheric pollution in the Urumqi-Changji-Shihezi urban agglomeration is more likely to occur in autumn and winter (Li, 2003; Ma et al., 2010; Wei, Meng & Li, 2012). On the one hand, the high concentration of energy-intensive industries (with coal as the main energy) and human activities results in large emissions of pollutants. On the other hand, the heating season is long in winter. Due to the long-term impact of the Mongolian high, the weather is dry, the wind is slow, and the atmosphere is stable, which provide good conditions for the formation of an inversion layer. However, the inversion layer, coupled with the special topography, is not conducive to the diffusion of pollutants, (Yang et al., 2011; Zhao, Li & Yang, 2011; Li et al., 2012a; Li et al., 2012b), which makes atmospheric pollution more likely to occur in winter. Kashgar is located in the hinterland of the Eurasian continent, adjacent to the Taklamakan Desert and other deserts in Central Asia. PM₁₀ is the primary pollutant in Kashgar, and the average annual PM₁₀ concentration is 3.4 times the PM₁₀ concentration limit of China (24-hour average PM₁₀ concentration limit: 150 µg/m³). Due to there is a lot of sand and dust in spring and summer, dust particles may transport BC to Kashgar, leading to a high BC concentration.

Influences of meteorological and environmental factors on atmospheric BC concentration

Jeong & Park (2013) showed that meteorological conditions caused particulate matter concentrations to decrease by 4% in summer but increase by 7% in winter in East Asia. The results of the RF model predicting the influence of meteorological factors on the atmospheric BC concentration (Fig. 8) showed that the correlation coefficient between the atmospheric BC concentration and the simulated value was 0.76, and the importance of meteorological factors to the atmospheric BC concentration was ranked as follows: snow depth >land surface temperature >humidity >sunshine hours >wind speed. Therefore, the atmospheric BC concentration at the Akdala Station was affected by all meteorological factors.

The snow depth was 0–24.04 cm at the Akdala station (average: 11.33 cm), which was divided into six grades (Fig. 9A), to explore the variation of atmospheric BC concentration with snow depth. It was found that the atmospheric BC concentration decreased with the increase of snow depth, and the average BC concentration in the snow cover season was 169.62 ng/m³, which was much lower than that in the snow cover-free season. The number of days without snow cover accounted for 71.78% of the total, and the number of days with 0–5, 5–10, 10–15, 15–20, and 20–25 cm snow depth accounted for 4.93%, 1.64%, 11.51%, 6.03%, and 1.37%, respectively.

The land surface temperature ranged from −22.81 to 38.19 °C (average: 10.18 °C), which was divided into 7 grades (Fig. 9B), to explore the variation of atmospheric BC concentration with land surface temperature. The number of days with land surface temperature at −30 to −20 °C, −20 to −10 °C, −10–0 °C, 0–10 °C, 10–20 °C, 20–30 °C, and 30–40 °C accounted for 1.09%, 11.51%, 20.82%, 16.71%, 14.52%, 20.55%, and 14.79%, respectively. Taking 0 °C as the cut-off point, when the land surface temperature was lower than 0 °C, the atmospheric BC concentration decreased with the increase of land surface temperature; when the land surface temperature was above 0 °C, the BC concentration increased with the increase of land surface temperature. The atmospheric BC concentration was the highest at 10–20 °C. The land surface temperature was below 0 °C in winter. Most areas of Xinjiang are located in the rear of the Mongolian high, with stable stratification and slow winds, which inhibits the diffusion of pollutants. Besides, due to the blocking by the Tianshan Mountains, there is a temperature inversion in northern Xinjiang, which promotes the accumulation of pollutants. In winter, the demand for heating increases, industrial coal-fired boilers and automobile exhausts emit large amounts of pollutants, causing an increase in the concentration of BC.

The relative humidity ranged from 15.63% to 92.83% (average: 60.08%), which was divided into 5 grades (Fig. 9C), to explore the variation of atmospheric BC concentration with humidity. The atmospheric BC concentration decreased with the increase of humidity in a certain range. The days with humidity at 0%–20%, 20%–40%, 40%–60%, 60%–80%, and 80%–100% accounted for 1.09%, 12.88%, 35.07%, 38.08%, and 12.88%, respectively. High relative humidity causes aerosols in the atmosphere to absorb moisture and settle, which reduces the atmospheric BC concentration. Currently, climate of Northwest China tends to be warmer and wetter due to the influence of climate warming, which accelerates the deposition and transformation of air pollutants, causing a decreasing in the annual variation of atmospheric BC concentration.

The results of the RF model predicting the influence of environmental factors on the atmospheric BC concentration (Fig. 10) showed that the correlation coefficient between atmospheric BC concentration and simulated value was 0.71, and the importance of environmental factors to atmospheric BC concentration was ranked as follows: NO₂ >PM₁₀ >SO₂ >CO >PM_2.5 >O₃.

The NO₂ concentration at Akdala Station ranged from 0.02 to 4.26 ppb (average: 1.84 ppb), which was divided into five grades (Fig. 11A), to explore the variation of atmospheric BC concentration with NO₂ concentration. It was found that the atmospheric BC concentration increased with the increase of NO₂ concentration. The days with NO₂ concentration of 0–1 ppb, 1–2 ppb, 2–3 ppb, 3–4 ppb, and 4–5 ppb accounted for 10.69%, 49.66%, 34.83%, 3.45%, and 1.38%, respectively.

The PM₁₀ concentration at Akdala Station ranged from 1.3 to 31.8 µg/m³ (average: 17.71 µg/m³), which was divided into 7 grades (Fig. 11B), to explore the variation of atmospheric BC concentration with PM₁₀ concentration. It was found that the atmospheric BC concentration increased with the increase of PM₁₀ concentration. The number of days with PM₁₀ concentration of 0–5 µg/m³, 5–10 µg/m³, 10–15 µg/m³, 15–20 µg/m³, 20–25 µg/m³, 25–30 µg/m³, and 30–35 µg/m³ accounted for 1.94%, 10.28%, 20.28%, 29.72%, 25.83%, 9.72%, and 1.94%, respectively.

The SO₂ concentration at Akdala Station ranged from 0.43 to 2.33 ppb (average: 1.38 ppb), which was divided into four grades (Fig. 11C), to explore the variation of atmospheric BC concentration with SO₂ concentration. It was found that the atmospheric BC concentration increased with the increase of SO₂ concentration. The number of days with SO₂ concentration of 0–1 ppb, 1–1.5 ppb, 1.5–2 ppb, and 2–3 ppb accounted for 0.86%, 74.43%, 24.14%, and 0.57%, respectively.

Black carbon is a typical primary aerosol that is emitted directly from natural and anthropogenic sources. Factors such as local pollutant sources, emission intensity, meteorological conditions, and geographical environment lead to strong spatiotemporal variation of BC pollution. In terms of meteorological conditions, the unique arid climate and geographical characteristics of Xinjiang is conducive to the diffusion and wet deposition of BC in a certain temperature range (−10–10 °C) and high humidity conditions, and the BC concentration may be high when the depth of the snow cover is 0–10 cm. The concentrations of PM, SO₂, CO, and O₃ in the atmosphere are the results of anthropogenic emissions. Through the analysis of the influence of major air pollutants on atmospheric BC concentrations, it was found that NO₂, PM₁₀, and SO₂, which ranked in the top three by importance, were all products of industrial and anthropogenic emissions, and the atmospheric BC concentration increased with the increase of NO₂, PM₁₀, and SO₂ concentrations.