Spatiotemporal Variation in Absorption Aerosol Optical Depth over China

Abstract

Absorbing aerosols can absorb solar radiation, affect the atmospheric radiation balance, and further have a profound influence on the global and regional climates. The absorption aerosol optical depth (AAOD) as well as the absorption Angstrom exponent (AAE) across China over 2005–2018 were systematically studied through the Ozone Monitoring Instrument (OMI) dataset. The monthly AAOD samples from the OMI generally showed a good correlation (~0.55) compared to the monthly data from AERONET at four typical sites (North: Xianghe, East: Taihu, South: Hongkong Polytechnic Univ; Northwest: Sacol) across China. The ensemble annual average of the OMI AAOD at 388 and 500 nm is 0.046 and 0.022, with minor changes during 2005–2015, and a relatively fast increase after that. The winter and spring seasons depict the maximum mean AAODs, followed by autumn, whereas summer shows minimum levels. On the contrary, the high AAE values appear in summer and low values in winter. The order of the annual average AAOD₅₀₀ from 2005 to 2018 is the Tarim Basin (TB, 0.041) > the Yellow River Basin (YRB, 0.023) > Beijing and Tianjin (BT, 0.026) > the Sichuan Basin (SB, 0.023) > Nanjing and Shanghai (NS, 0.021) > the Pearl River Delta (PRD, 0.017), whereas the AAE_388–500 exhibits the opposite trend except for the TB (3.058). From 2005 to 2018, the AAOD rises by nearly 1.5–2.0 fold in the six typical regions, implying a severe situation of dust and/or BC aerosol pollution in the last several years. The monthly mean AAOD₃₈₈ over the TB, the SB, the YRB, BT, the PRD, and NS is estimated to be smallest at 0.072, 0.024, 0.026, and 0.027 in July, 0.024 in June, and 0.025 in September, respectively, whilst largest in January for NS, the YRB and BT, April for the TB, February for the SB, and March for the PRD with 0.055, 0.077 and 0.067, 0.123, and 0.073 and 0.075, respectively. The monthly averaged AAOD₅₀₀ in each region is consistently about half of the AAOD₃₈₈. The highest AAE appears in June while the lowest values are in December and January, and the daily AAE values in episode days slightly decrease as compared to non-episode days. Our study indicates that northwestern China plays an important role in the overall AAOD as a result of dust aerosols stemming from desert areas. Moreover, the meteorological conditions in winter and early spring are associated with more energy consumption conducive to the accumulation of high black carbon (BC) aerosol pollution, causing high alert levels of AAOD from November to the following March.

1. Introduction

Aerosols are tiny water droplets and particulates in the air. They are considered an important factor affecting global climate change, radiation balance, the cloud formation process and human health [1,2,3]. Aerosol optical depth (AOD) is the elementary optical parameter of aerosols [4]. As the non-scattering part of the AOD, the absorption aerosol optical depth (AAOD), on behalf of aerosol column optical absorption, is an important parameter for assessing radiative forcing and atmospheric heating [5,6,7]. The Angstrom exponent (AE) provides information about the aerosol dominance size in the atmospheric vertical column. As an indispensable part of the AE, the absorption Angstrom exponent (AAE) is another key parameter, and it can be applied to reveal the spectral dependence of aerosol absorption [8]. Additionally, the AAE can be used to infer the main particle types in absorbent aerosols, for example mineral dust, BC, and brown carbon (BrC), as the absorption wavelength dependence is an intrinsic property closely interrelated with aerosol composition. BC and BrC chiefly come from incomplete anthropogenic combustion and biomass burning, and dust is predominantly derived from natural wind erosion, sometimes from landscape changes or soil disturbance [1,9]. The AAE value of pure BC is near 1 theoretically, while the presence of dust or BrC increases the AAE value because of its enhanced absorption of ultraviolet/near-ultraviolet wavelengths [10]. The AAE depends on the emitting source and the mixing state of the absorption aerosol [11,12,13,14]. To understand the effect of absorptive aerosols on the environment profoundly, it is necessary to analyze the AAOD and AAE.

Remote sensing techniques including ground and satellite networks play the vital role in aerosol monitoring and characterization, as well as allowing a deeper understanding of the optical properties of atmospheric aerosols. Currently, the way to measure aerosol absorption characteristics routinely and globally is radiation measurements, e.g., the Aerosol Robotic Network (AERONET) based on ground observation [15,16,17,18,19], and the Ozone Monitoring Instrument (OMI) based on satellite remote observation [20,21,22,23,24,25,26,27]. Nonetheless, AERONET is unable to determine the regional-scale AAOD variation trends with relatively finite monitoring stations and a narrow monitoring range while the OMI may misestimate or lose data for some days such as for excessive rain. Shin et al. developed a method for classifying dust and non-dust aerosol types and determined the contributions of BC-related absorption to the non-dust AAOD from AERONET observations [16]. Sun et al. compared the variations in black carbon AAOD and dust AAOD in China [17]. Dehkhoda et al. inverted the AAOD of black carbon from AERONET observations over the world from 2000 to 2018 [18]. Russell et al. utilized AAOD, single-scattering albedo (SSA), and the AAE from the AERONET data to distinguish the respective proportion of BC, dust, and organic matter in the entire AAOD, and pointed out that the AAE between 325 and 1000 nm is about 1.45 for biomass burning, whereas it is 2.3 for Saharan dust aerosols [19]. The OMI utilizes the greater sensitivity of radiation measured at the top of the atmosphere in the near ultraviolet region to different loads and aerosol types, and derives extinction AOD and AAOD using inversion programs generated by the near-ultraviolet aerosol inversion algorithm at three shortwave lengths of 354, 388 and 500 nm [20,21]. Buchard et al. used the OMI aerosol index and AAOD to evaluate the NASA MERRA aerosol reanalysis [22]. Zhang et al. reduced the uncertainties of BC emissions by assimilating the OMI observation results of AAOD with the chemistry model of the Goddard Earth Observing System [23]. Their research group also discussed what factors have controlled the AAOD growth trend in the USA during the past decade [24]. Zaman et al. compared the mean annual spatial distribution of satellite-based OMI-AAOD (0.024 for Dhaka; 0.023 for Bhola) with the ground-based AERONET-AAOD (0.110 for Dhaka; 0.073 for Bhola) [25]. Ali et al. found that the annual patterns of OMI-AAOD and AERONET-AAOD over Saudi Arabia are almost similar [26]. Zhang reported that the daily AAE_388–500 for dust events over the Tarim Basin based on the OMI is depicted to be in a range of 3.2–3.6, with an average of 3.3 [27].

To the best of our knowledge, few investigations have focused on the long-term and large-scale variation characteristics of detailed AAOD and AAE values over China, especially in typical regions, and the reasons for AAOD and AAE variations should be studied. Herein, the accuracy of the OMI AAOD with ground-based AERONET AAOD is validated, then the spatial–temporal characteristics of the AAOD and AAE across China during the period of 2005–2018 on the basis of the OMI data are studied systematically. This study aims to evaluate the present AAOD situation, AAOD and AAE characteristics with regard to annual, seasonal, and monthly variations, and the divergence in AAE values for episode days and non-episode days across China, which will hopefully benefit governments’ efforts to mitigate air pollution.

The paper is arranged as follows: in Section 2, the data and method for the AAOD and AAE are introduced; in Section 3, the validation of the OMI dataset with AERONET followed by 14-year detailed characteristics of the AAOD and AAE, and the main factors leading to the variations are discussed. Finally, the main conclusions are summarized in Section 4.

2. Data and Methodology

2.1. OMI Data and Methodology

The OMI is a nadir-viewing near-ultraviolet/visible CCD spectrometer on NASA’s EOS Aura satellite [28]. Two particle algorithms including OMAERUV (OMI Aura near UV) and OMAERO (OMI multi-wavelength) were applied in OMI observation [29]. Sub-pixel cloud pollution is the main factor affecting the quality of aerosol products. Nevertheless, AAOD is less affected by cloud pollution, as the cloud impact on the combined albedo of AOD and SSA inversion is partially offset [23]. Because the OMI is highly sensitive to aerosol absorption in near-ultraviolet observations, AAOD is the surest quantitative OMAERUV aerosol parameter, particularly over land [30]. The root-mean-square error estimate is about 0.01 [31], and the data are utilized due to their high reliability [21]. More details concerning the OMI’s data products, quality, accuracy, systematic bias, and drift have been introduced by scientific researchers, such as Kumar et al. [32], Adesina et al. [33], Torres et al. [21], and Boiyo et al. [34]. The aerosol data product contains the quality assurance labels, and applies the surest retrievals with the least impact from sub-pixel cloud contamination [30]. In this study, the Level 3 daily global gridded product OMAERUV produced by the OMI research team, including daily AAOD at 388 nm and 500 nm during 2005–2018, is used with a spatial resolution of 1° × 1° [35].

The AAE is described in Formula (1) [36].

$$\mathrm{A}\mathrm{A}\mathrm{E}=-{\displaystyle \frac{\mathrm{l}\mathrm{n}({\mathrm{A}\mathrm{A}\mathrm{O}\mathrm{D}}_{{\mathsf{\lambda }}_1}/{\mathrm{A}\mathrm{A}\mathrm{O}\mathrm{D}}_{{\mathsf{\lambda }}_2})}{\ln \left({\mathsf{\lambda }}_1/{\mathsf{\lambda }}_2\right)}}$$

(1)

where ${\mathsf{\lambda }}_1$ and ${\mathsf{\lambda }}_2$ denote two different wavelengths. In this study, the terms ‘AAOD₃₈₈’ and ‘AAOD₅₀₀’ will represent the values of AAOD at the wavelengths of 388 nm and 500 nm, respectively, while the term ‘AAE_388–500’ will be used to refer to the AAE for wavelengths 388 to 500 nm.

2.2. AERONET Data and Methodology

AERONET is a ground-based instrument network that supplies a lengthy, continuous, and easily accessible public-realm database of aerosol radiation, optics, and microphysical properties [37]. The AERONET inversion code supplies aerosol optical properties in the entire atmospheric column obtained from direct and scattered radiation measured by the Cimel solar/sky radiometer [38,39].

We utilized AERONET aerosol inversions data of AAOD at 440 nm with Level 1.5 and Level 2.0 quality assurance [40]. Cloud clearing and manual checks were performed in the measurements, and pre and post field calibrations were also carried out [41]. The total uncertainty of the on-site instrument AERONET AOD ranges from ±0.1 to ±0.2, which is related to the higher error (±0.2) within the ultraviolet spectrum scope [42]. To compare with OMI₅₀₀, herein, AERONET AAOD₄₄₀ was converted to AAOD₅₀₀ according to the following equation [25]:

$${\mathrm{A}\mathrm{A}\mathrm{O}\mathrm{D}}_{500}={\mathrm{A}\mathrm{A}\mathrm{O}\mathrm{D}}_{440}\times {\left({\displaystyle \frac{500}{440}}\right)}^{-{\mathrm{A}\mathrm{A}\mathrm{E}}_{440–870}}$$

(2)

where the term ‘AAE_440–870’ refers to the AAE for wavelengths 440 to 870 nm.

2.3. MERRA-2 Data

The MERRA-2 (Modern-Era Retrospective analysis for Research and Application-2) data reanalysis dataset is produced by NASA’s Global Modeling and Assimilation Office (GMAO) using the Goddard Earth Observing System Model (GEOS) version 5.12.4 [43]. At present, the MERRA-2 is the most significant long-term global reanalysis from 1980 to the present, representing the interactions of aerosols with other physical processes in the climate system [44]. A spatial resolution of 0.5° × 0.625° and 72 layers from the surface to around 80 km is provided in the MERRA-2 dataset. MERRA-2 level 3 supplies 1-hourly, 3-hourly, and monthly products in the NASA website [45].

2.4. MODIS-Aqua Data

The present study utilized level 2 AOD at 550 nm retrieved from MODIS (Moderate resolution imaging spectroradiometry) Terra at a spatial resolution of 1° × 1° for a period of 14 years (January 2005 to December 2018) to study the regional distribution of AOD over China. These satellite data products were obtained based on Giovanni maps from the NASA website [46]. Detailed information concerning the sensor, data products, retrieval algorithms, calibration, and uncertainties can be found elsewhere [47,48,49].

3. Results and Discussion

3.1. Validation of OMI AAOD with AERONET AAOD

There are approximately 35 unevenly distributed ground-based measurement AERONET sites in China, many of which only have several days’ or months’ measurements. In addition, ground-based observations are scarce in central and southwestern China, with more along the eastern coastal region. Unlike AERONET, the OMI AAOD can cover the whole of China and provide long-term AAOD values. Four AERONET sites with the most observation data were chosen for comparison of OMI remote sensing.

Table 1. Basic information of the four studied AERONET sites during 2005–2018.

Site	Xianghe	Taihu	Hongkong Polytechnic Univ	Sacol
Location	North China	East China	South China	Northwest China
Latitude (Degree)	39.8° N	31.4° N	22.3° N	35.9° N
Longitude (Degree)	117.0° E	120.2° E	114.2° E	104.1° E
Elevation (m)	36 m	20 m	30 m	1965 m
AERONET AAOD440	0.052	0.050	0.039	0.038
AERONET AAE440–870	1.288	1.310	1.272	1.553
AERONET AAOD500	0.044	0.042	0.033	0.031
OMI AAOD500	0.023	0.021	0.018	0.021
OMI AAE388–500	2.766	2.826	2.850	2.793

displays the detailed geographical location, longitude, latitude, and altitude of Xianghe, Taihu, Hongkong Polytechnic Univ, and Sacol stations, and summary data related to the AAOD and AAE. Figure 1 shows the comparison of monthly average AAOD₅₀₀ data between OMI products and AERONET during 2005–2018. The adjacent grid was extracted from the OMI according to the longitude and latitude of each site, and then the relevant data were employed for validation.

The correlation coefficient (R) for all sites is significant at the 0.01 level coherently. The R between the OMI and AERONET fluctuated around 0.55. Relative mean bias (RMB) is utilized to express the accuracy of AAOD data, and values greater than 0.5 are considered reliable [50]. Compared with the monthly average data, some more robust results were obtained for daily data as the sample size increased. For example, the R and RMB slightly increased from 0.54 and 0.53 to 0.56 and 0.58 in Xianghe from 2005 to 2018, respectively (Figure 2a). The total slope was 0.21–0.35, deviating from the predetermined 1:1 path (red dashed line), implying that the OMI product underestimated the AAOD₅₀₀ in Xianghe (RMB = 0.53), Taihu (RMB = 0.50), Hongkong Polytechnic Univ (RMB = 0.55), and Sacol (RMB = 0.68). The underestimation was caused by the overestimation in the estimated surface reflectance [26].

Correlation analysis on the daily dataset was also performed between the OMI AAOD₅₀₀ and AERONET AAOD₅₀₀ with Level 1.5 and Level 2.0 quality assurance (Figure 2). The sample size of Level 2.0 is only one-third of that for Level 1.5, and the R value has increased to a certain extent but the RMB has decreased from 0.61 to 0.48. Through comparison with ground-based observations from AERONET, OMI retrievals have achieved a great deal of research and approval. Ahn et al. reported a good correlation with the R of 0.81 between the OMI AOD retrieved using the OMAERUV algorithm and AERONET AOD over 2005–2008 [30]. Zhang et al. compared the inversion between the OMI and AERONET AOD at 27 stations over East and Central Asia and yielded a satisfactory level of agreement, with R about 0.77, especially with more accurate retrievals in South Korea and East and North China [51]. OMI SSA retrievals were quite consistent with the AERONET SSA retrieved on the basis of 269 ground-based stations at a global scope, with around 69% retrievals concordance under an absolute difference of ±0.05 [52]. As for AAOD, Bibi et al. evaluated the OMI retrievals in Karachi, Pakistan with AERONET AAOD, and the results revealed a higher positive correlation (R² = 0.78) [53]. There was a moderately positive correlation between the AERONET AAOD and OMI AAOD in a study conducted in Saudi Arabia (R² = 0.39) [26].

In Figure 3, 4452 daily 500 nm AAOD values for the time period of 2005–2018 were compared between the satellite-based OMI and ground-based AERONET at the above four sites. In general, the AERONET AAOD was higher than the OMI AAOD in all seasons and months. The seasonal distribution of AERONET AAOD exhibited the highest AAOD₅₀₀ values in spring (0.042 ± 0.028) and winter (0.041 ± 0.034), followed by autumn (0.031 ± 0.025) (Figure 3a). The summer displayed the lowest AAOD value (0.026 ± 0.025). OMI AAOD_500nm retrievals revealed a similar temporal pattern in the order of winter (0.027 ± 0.033) ~ spring (0.025 ± 0.020) > autumn (0.018 ± 0.028) ~ summer (0.019 ± 0.013). As shown in Figure 3b, the lowest AAOD values were in the months of July (0.017 ± 0.010) for AERONET and September (0.014 ± 0.010) for the OMI, respectively. Zaman carried out validation of the OMI AAOD₅₀₀ with AERONET measurement, and also found lower AAOD values compared to the AERONET-AAOD with a distinct annual cycle [25].

On the whole, although the OMI underestimated the AAOD value, the scatter plots portrayed well the correlation between the OMI AAOD data and AERONET, and the OMI was highly credible and applicable over the studied region.

3.2. Spatial and Annual Variation in OMI AAOD

Figure 4 presents the spatial distributions of the annual averaged OMI AAOD₃₈₈ and AAOD₅₀₀ in China during 2005–2018. Hu reported that the AAOD over East Asia is predominantly contributed to by dust and BC, whilst other aerosols make a small fraction of the contributions to the AAOD [54]. According to the measurement results of East Asian tropospheric aerosol motion research, Yang et al. estimated the mass absorption efficiencies of dust and BC at 550 nm to be 0.03 and 9.5 m² g⁻¹, respectively [55]. The highest AAOD generally occurs in the vast but sparsely populated northwest region with AAOD₃₈₈ above 0.85. Apparently, compared with other locations in China, aerosol absorption in northwest China has a greater contribution towards the total aerosol presence. From the inter-annual changes, the AAOD values in the economically developed and densely populated Beijing–Tianjin–Hebei (BTH) region, northern Henan, and the Sichuan Basin (SB) are also significantly higher, which may be due to regional emission sources and specific local features mixed with meteorological influence.

To explore the variations in the absorption of aerosols over China on a broader temporal scale, Figure 5 shows the annual variation in area-average BC (a) and dust (b) surface mass concentration, and BC (c) and dust (d) column mass density, over the study area in China, the Tarim Basin (TB), and Beijing and Tianjin (BT) over 2005–2018 dependent on MERRA-2 data [35]. Generally speaking, the BC surface mass concentration and column mass density began to rapidly increase after 2005 and reached their maximum in 2007, and then slowly decreased (Figure 5a,c). The phenomenon may be primarily associated with anthropogenic emissions. With the rapid industrial development and urbanization, China’s economy is accelerating in growth, and energy consumption such as coal and oil is also increasing. Subsequently, the government has adopted positive measures to control and prevent air pollution. The BC concentrations display a certain downward trend, but held firm at a relatively high level between 2008 and 2018, indicating the need for long-term implementation of BC emission-reduction policies. BC was mostly concentrated in the industrial and agricultural regions as a result of incomplete combustion of fossil fuels and biomass burning. As shown in Figure 5, BT is located in northern China with relatively high BC concentrations, and the TB is located in remote north-western area of China with low BC concentrations. As for dust surface mass concentration and column mass density in China (Figure 5b,d), that in the TB is much higher than that in BT and the national average. The peaks appeared in 2018 with annual average values of 373 μg/m³ and 6.2 × 10⁵ μg/m², respectively, revealing an extensively analogous tendency to the AAOD. The AAODs achieved in the TB are larger than those over BT, with dust stemming from the Taklimakan Desert contributing the most to the total AAOD.

As shown in Figure 6, the annual distribution provides an overview of the OMI AAOD over China. The averaged AAOD₃₈₈ and AAOD₅₀₀ in China are 0.046 ± 0.018 and 0.022 ± 0.009 during 2005–2018, respectively. The highest annual-averaged AAOD₃₈₈ and AAOD₅₀₀ was observed to be 0.055 and 0.026 in 2018. AAOD pollution was the lowest over China in 2008 with 0.043 for AAOD₃₈₈ and 0.021 for AAOD₅₀₀. The absorptive aerosol pollution became severe in the recent years of 2016–2018. High values of AAOD were mainly distributed near dust sources as well as BC sources. The AAOD across China increased by a factor of nearly 1.5, especially 2-fold in the northwest during the studied period, implying a severe situation of dust and/or BC aerosol pollution in the recent years of 2016–2018 (Figure 5).

In brief, the high-value center of OMI AAOD is predominantly scattered over the northwestern TB region, the BTH region and the surrounding northern part of Henan province, and the SB in China.

3.3. OMI AAOD over Typical Areas of China

The MODIS reanalysis data are utilized to distinguish sandstorms, fog, and haze, as well as cirrus clouds [56]. The mean AOD values at 550 nm area-averaged over China derived from the MODIS true-color images with a spatial resolution of 1 × 1 km in the TB (37.5–41.5° N, 79.5–86.5° E), the SB (29.5–31.5° N, 103.5–106.5° E), the Pearl River Delta (PRD) (21.5–23.5° N, 112.5–114.5° E), Nanjing and Shanghai (NS) (30.5–32.5° N, 118.5–121.5° E), the Yellow River Basin (YRB) (32.5–37.5° N, 114.5–117.5° E), and BT (38.5–40.5° N, 115.5–117.5° E) over 2005–2018 are all higher than 0.5, indicating the impact of aerosols from regional air pollution. The geographical locations of the six regions marked with white boxes are shown in Figure 7. The spatiotemporal properties of the AAOD and AAE are studied in the above six representative regions based on the latest OMI.

Figure 8 exhibits temporal variations of the AAOD₃₈₈, AAOD₅₀₀, and AAE_388–500 in the six typical geographic areas from 2005 to 2018. The effective monitoring days in the TB, the SB, the PRD, NS, the YRB and BT were 4187, 1401, 1425, 1976, 3032, and 2918, respectively. Hence, the temporal changes in the TB region were the most regular. Generally, the annual mean AAOD during 2015–2018 was in the order of TB > YRB > BT > SB > NS > PRD. It is not surprising that the maximum AAOD₃₈₈ was found over the TB, ranging from 0.073 to 0.140, followed by the YRB (0.053–0.077), BT (0.042–0.066) and the SB (0.039–0.050). However, compared to the aforementioned regions, the overall AAOD₃₈₈ of the PRD (0.029–0.051), and NS (0.036–0.062) was lower, implying lighter AAOD pollution, and smaller values are associated with lower population and aerosol loads.

The AAOD₃₈₈ values display large variabilities between 0–0.8, whereas the AAOD₅₀₀ values vary in the range of 0–0.4, nearly half of the AAOD₃₈₈. The AAOD values decrease as the wavelength increases, implying a positive AAE will be achieved. The distinctively higher annual mean AAOD (0.088 for AAOD₃₈₈ and 0.041 for AAOD₅₀₀) in the TB (

Table 2. Average and standard deviation of AAOD₃₈₈, AAOD₅₀₀, and AAE_388–500 at six representative regions in China. The maximum and minimum are highlighted in red and green, respectively.

	Season	TB	SB	PRD	NS	YRB	BT
AAOD388	Annual	0.088 ± 0.073	0.047 ± 0.046	0.036 ± 0.032	0.042 ± 0.045	0.058 ± 0.064	0.052 ± 0.059
	Spring	0.105 ± 0.052	0.053 ± 0.036	0.059 ± 0.038	0.042 ± 0.027	0.057 ± 0.034	0.056 ± 0.042
	Summer	0.073 ± 0.026	0.026 ± 0.016	0.026 ± 0.016	0.031 ± 0.019	0.033 ± 0.028	0.034 ± 0.025
	Autumn	0.077 ± 0.077	0.037 ± 0.054	0.029 ± 0.033	0.043 ± 0.073	0.062 ± 0.1	0.045 ± 0.083
	Winter	0.099 ± 0.115	0.065 ± 0.057	0.031 ± 0.026	0.051 ± 0.045	0.074 ± 0.067	0.063 ± 0.06
AAOD500	Annual	0.041 ± 0.035	0.023 ± 0.023	0.017 ± 0.016	0.021 ± 0.023	0.029 ± 0.033	0.026 ± 0.03
	Spring	0.049 ± 0.027	0.026 ± 0.018	0.028 ± 0.018	0.021 ± 0.013	0.028 ± 0.017	0.028 ± 0.021
	Summer	0.032 ± 0.011	0.012 ± 0.008	0.012 ± 0.008	0.015 ± 0.009	0.016 ± 0.013	0.016 ± 0.012
	Autumn	0.036 ± 0.037	0.018 ± 0.028	0.014 ± 0.017	0.022 ± 0.039	0.032 ± 0.052	0.023 ± 0.043
	Winter	0.048 ± 0.055	0.032 ± 0.03	0.015 ± 0.013	0.026 ± 0.024	0.037 ± 0.035	0.032 ± 0.031
AAE388–500	Annual	3.058 ± 0.272	2.821 ± 0.135	2.849 ± 0.119	2.825 ± 0.127	2.806 ± 0.144	2.770 ± 0.137
	Spring	3.088 ± 0.297	2.821 ± 0.127	2.871 ± 0.121	2.868 ± 0.115	2.842 ± 0.131	2.807 ± 0.129
	Summer	3.232 ± 0.191	2.879 ± 0.063	2.901 ± 0.093	2.877 ± 0.073	2.891 ± 0.086	2.859 ± 0.099
	Autumn	3.029 ± 0.201	2.796 ± 0.164	2.822 ± 0.101	2.78 ± 0.129	2.745 ± 0.139	2.731 ± 0.119
	Winter	2.825 ± 0.224	2.787 ± 0.152	2.806 ± 0.133	2.764 ± 0.137	2.759 ± 0.151	2.71 ± 0.137

) is intimately linked to dust episodes. The AAE_388–500 depicts regular annual variations, and ranges from 2.207 to 4.332 with the averages of 3.058, 2.821, 2.849, 2.825, 2.806, and 2.770 for the TB, the SB, the PRD, NS, the YRB and BT, respectively. The angstrom wavelength index can be used to quantify the spectral dependence of absorption [57]. The relatively higher AAE value in the TB is likely associated with dust aerosols due to a large dust source from the Taklimakan Desert in the TB region.

To better illustrate the current situation of absorptive aerosol loading, further comparisons in the year scale of the AAOD₃₈₈, AAOD₅₀₀, and AAE_388–500 are analyzed in the six typical regions (Figure 9). The AAOD over East Asia is mainly caused by dust as well as BC and BrC [29]. Most notably, BC contributes the most to total absorption; BrC and dust contribute significantly at shorter wavelengths or during dust events [54,55]. The yearly means of AAOD₃₈₈ and AAOD₅₀₀ display slight variations from 2005 to 2015, while from 2016 to 2018, yearly average AAODs increase obviously for the TB. This may be related to dust with a similar annual trend during 2005 to 2018 (Figure 5b,d) [17,27,58]. As shown in Figure 9a,b, the AAOD value over the TB is the largest among the six regions, and reaches its peak in 2018 with 0.140 and 0.065 for AAOD values at 388 and 500 nm, respectively. Additionally, NS, the YRB and BT all reach the yearly mean maximum in 2017 with 0.062, 0.077, and 0.066 for AAOD₃₈₈, and 0.031, 0.040 and 0.033 for AAOD₅₀₀, respectively. The AAOD value of the PRD is generally the lowest in the six regions, and reaches the smallest annual average in 2008. Apart from the PRD, the TB, the SB and the YRB values also descend to a lowest point in 2008, and then demonstrate an uptick again in recent years. The phenomenon may be mainly related to anthropogenic activities and natural changes. Along with rapid industrial development and urbanization, China has witnessed a fast-growing economy as well as energy consumption. Subsequently, as a coping mechanism against air-pollution hazards and to protect public health, the government and departments have taken various environmental measures to prevent and control air pollution [59]. The BC concentration fluctuates in a downward trend but remains at a high level during 2016–2018 (Figure 5a). On the other hand, the enhanced AAOD from 2016–2018 may indicate increased dust aerosol pollution over China, especially in the TB (Figure 5b,d), possibly induced by the enhanced frequency of sandstorms. In Figure 9c, the range of the AAE_388–500 in the TB, the SB, the PRD, NS, the YRB and BT is 2.936 to 3.116, 2.781 to 2.861, 2.798 to 2.890, 2.786 to 2.863, 2.750 to 2.852 and 2.722 to 2.802, respectively. The TB is significantly larger than other regions and reaches its peak value in 2015, indicating an increase in dust aerosols, possibly induced by the enhanced frequency of sandstorms. BT owns the lowest value among the six regions, and drops down to the lowest point in 2018. Meanwhile, the PRD, NS and the YRB all reach their minimum in 2017.

Despite large variations in the time scale, the AAOD₃₈₈ and AAOD₅₀₀ and AAE_388–500 over the six regions during 2005–2018 exhibit palpable seasonality, as demonstrated in Figure 10. Since the AAOD_388nm and AAOD₅₀₀ have a similar linear trend, herein, we only discuss AAOD₃₈₈ as a representative. In the TB, the highest AAOD₃₈₈ occurs in spring (0.105), and the magnitude has rapidly decreased in summer and autumn (Table 2). Drylands are characterized by high AAOD in spring, which can be attributed to the frequent dust activity [60,61]. The seasonally averaged AAOD₃₈₈ in the YRB peaks in winter (0.074), followed by autumn, spring, and summer. NS has a similar seasonal trend with 0.051, 0.043, 0.042, and 0.031 for winter, autumn, spring, and summer, respectively. In the SB and BT, the average AAOD₃₈₈ is alleviated to some extent in summer compared to the average values in the other seasons. For the PRD region, 0.06, 0.031, 0.029, and 0.026 are observed in spring, winter, autumn, and summer, respectively. The mean AAOD₃₈₈ in the PRD in the four seasons is analogous, and the variance is smaller as compared with the other five regions, indicating less absorptive aerosol loading. The largest seasonal averages of AAOD₃₈₈ for all six regions occur in winter or spring, whilst the smallest seasonal averages are observed in summer without exception. Strong seasonal means are observed during 2016–2018, and weak values during 2005–2015 are seen. Absorbable aerosol contributes more in cold seasons, and the seasonal variation trends of the AAOD are in line with the results obtained by other studies [38,62]. Since 2005, the total consumption of coal and diesel display an overall uptrend, increasing by 91.51 and 7.42 million tons each year, respectively [29]. The uncompleted combustion of fossil fuels is a significant source of BC and BrC [63], and the high concentrations partially illustrate the upward trend of the AAOD in most parts of China during cold winters. The increasing energy consumption (e.g., coal and other fuels) for heating in winter, accompanied by the frequent occurrence of atmospheric inversion and lower boundary layer height, leads to the unfavorable dilution diffusion of BC, thus exacerbating pollution. The winter AAOD values in eastern (NS), northern (the YRB, BT), and central (the SB) China have significantly increased. The northwest (the TB) is sparsely populated and emits less BC throughout the year. On the other hand, spring is prone to sandstorms, and its AAOD values are higher than those of winter.

Nevertheless, the AAE_388–500 for six typical regions shows maximum mean values in summer, followed by spring and autumn, whereas winter shows weaker AAEs, being different from the seasonal variation trend of the AAOD. The discrepancy in the AAE is related to two factors: the dominant absorptive aerosol types and emission sources. The AAE over the TB with its desert source obtained in our study (3.088, 3.232, 3.029, and 2.825 for spring, summer, autumn, and winter) is significantly larger than those seen in the other five areas, whereas the values of the AAE_388–500 are found to be near 2.750.

The monthly average variations of the AAOD_388nm, AAOD_500nm, and AAE_388–500 over the six sampling areas from 2005 to 2018 are illustrated in Figure 11. Further analysis found that the monthly averaged AAOD₅₀₀ in each region is consistently about half of the AAOD₃₈₈ (Figure 11a,b). The monthly patterns of the AAOD and AAE_388–500 over the six regions represent almost identical trends and fluctuations, in line with their seasonal patterns. The monthly statistics of the AAOD indicate that the monthly averages of the PRD are lower than those of the other five regions except in March and April. The monthly average AAOD₃₈₈ in the TB, SB, YRB, BT, PRD, and NS regions shows the minimal values of 0.072, 0.024, 0.026, and 0.027 in July, 0.024 in June, and 0.025 in September, respectively. Absorptive aerosols can be effectively removed by summer wet precipitation. Moreover, more aerosols from ambient air can be captured by leaves of flourishing trees. The peak events of AAOD₃₈₈ occur in January for NS, the YRB and BT with the average values of 0.055, 0.077 and 0.067, respectively, whereas the peaks are in April for the TB, February for the SB, and March for the PRD with 0.123, 0.073, and 0.075, respectively. The monthly mean AAOD values vary two- or three-fold between the lowest and maximum months, and high alert levels of AAOD are preserved from November to the following March. The meteorological conditions in winter and early spring associated with more energy burning facilitate the formation of high BC aerosol contamination. A previous study also declares a significant upward trend in BC and OC emissions after September [64]. For the TB, sandstorm weather is more frequent from March to May, and can explain why the TB can have a high AAOD in spring, rather than in winter. The monthly mean AAE_388–500 is in a range of 2.759–3.254 for the TB, with the largest value in June and the smallest in February. Compared to the monthly variations for the TB, the fluctuation range between the maximum and minimum values in the other five regions (SB: 2.748–2.887, PRD: 2.789–2.906, NS: 2.719–2.887, YRB: 2.674–2.916, BT: 2.661–2.891) is weaker. All typical regions show a low AAE in cold months and high AAE in June.

3.4. AAE on Episode and Non-Episode Days

The AAE plays an important indicator role in air pollution, and it can be used to distinguish aerosol particle types. The theoretical AAE of BC is close to 1.00 while a higher AAE between 1.00 and 2.00 indicates a significant increase in urban pollution by organic carbons because the atmospheric pollutants are usually mixed with strongly absorbing particles. Dust aerosols are likely to increase when the AAE increases to about 3.00 [65]. The Grade II of Chinese Ambient Air Quality Standard (CAAQS) for a daily mean of PM_2.5 is 75 μg/m³; thus, the daily average PM_2.5 > 75 μg/m³ is defined as PM_2.5 episode days [66,67]. According to the previous statistical research, average PM_2.5 concentrations during the episode days are 2–3 times higher as compared to PM_2.5 concentrations during the non-episode days [66]. The air pollutants monitoring data were provided by the China Air Quality Online Monitoring and Analysis Platform [68].

We selected 13 monitoring cities from six regions (the TB: Aksu; the SB: Chengdu, Deyang, Chongqing; the PRD: Guangzhou, Shenzhen, Zhuhai; NS: Nanjing, Shanghai; the YRB: Puyang, Jinan; BT: Beijing, Tianjin) to study the diversities in the AAE between PM_2.5 episode days and non-episode days (Figure 12). As compared to other cities, more episode days and higher AAE values occurred in Aksu over the Tarim Basin during 2017 and 2018. As discussed above, the AAE values were significantly affected by dust aerosols from the Taklimakan Desert. Guangzhou, Shenzhen, and Zhuhai, belonging to the YRD, had the minimum number of episode days. Furthermore, the AAE values on these days were regularly low, especially in 2018 with the most obvious difference. Beijing and Tianjin in the BT area displayed no significant differences of AAEs between episode days and non-episode days. In 2018, the AAE_388–500 over industrial-emission-dominant regions like Guangzhou, Shanghai, and Beijing was 2.64 ± 0.044, 2.71 ± 0.18, and 2.72 ± 0.13 for episode days, and 2.85 ± 0.20, 2.83 ± 0.13, and 2.72 ± 0.12 for non-episode days. The AAEs obtained for episode days were generally lower than those in non-episode days.

3.5. Comparison

The AAOD and AAE_388–500 exhibit regular annual, seasonal, and monthly variations. The optical properties of aerosols obtained by different instruments, testing methods, and measuring wavelengths are inconsistent. The AAODs in the six regions except the TB observed in our study are approximately close to northeast China, where the maximum AAOD occurred in Liaoning Province in the range of 0.02 to 0.03 based on MERRA-2 data [5]. The hot summer produces a relatively lower AAOD, while the cold season reveals the highest absorptive aerosol loading. The seasonal trend changes of the AAOD agree quite well with the results of Kang et al.’s analysis over hyper-arid regions [29]. The winter AAOD₃₈₈ over the desert region is higher than those seen in urban areas of China, e.g., Hangzhou (AAOD₄₄₀: 0.06) [69], and Hefei (AAOD_550nm: 0.05 [70], but similar to that obtained in Shenyang (AAOD₄₄₀: ~0.1) [71] on the basis of CE-318 sun photometer measurements. Similarly, Zhang et al. observed the AAOD variation trend in the United States from 2005 to 2015, and pointed out that the increase in AAOD was mainly attributed to an increase in dust concentration rather than BC [24]. In contrast, the AAOD in our study presents an obvious upward trend in winter and spring, which has a bearing on the concentrations of carbonaceous aerosols and dust.

As for the AAE_388–500, the values in the TB in the four seasons herein are significantly larger than those for Lhasa (AAE_370–950: 1.04), Jiaozuo (AAE_440–870: 1.09), and Nanjing (AAE_470–660: 1.56) with the seven-channel Aethalometer model AE33 [72], ground-based sun photometer measurement [73], and seven-channel Aethalometer model AE31 [74], respectively. Bahadur et al. estimated that the average AAE_440–675 is 2.2 with the observational analysis method [75]. The monthly average AAE shows its lowest value in cold months and the maximum in June; Adam et al. investigated and found that Singapore had a lower AAE_365–700 in February and a higher value in June [76], while their AAEs were lower than those seen in our study. The AAE_388–500 values acquired on the OMI usually surpass other approaches, and one reason may be that the AAE was obtained at shorter wavelengths in our study. The conclusion can be confirmed that the OMI AAE_388–500 (Xianghe: 2.766, Taihu: 2.826, Hongkong: 2.850, Sacol: 2.793) with short wavelengths is almost twice that of the AERONET AAE_440–870 (Xianghe: 1.288, Taihu: 1.310, Hongkong: 1.272, Sacol: 1.553) with a longer wavelength range (Table 1). This study provides a brief discussion, but more comprehensive and detailed research is needed to offer a detailed explanation, which is beyond our research scope.

4. Conclusions

The detailed characteristics of the atmospheric OMI AAOD as well as the AAE in China over 2005 to 2018 were investigated based on the OMI dataset. The OMI AAOD₅₀₀ generally portrayed a good correlation (~0.55) with ground-based AERONET AAOD₅₀₀ at four typical stations around China. During the study period, the annual mean OMI AAOD bespeaks multifarious spatial distributions across China with a mean value of 0.046 for AAOD₃₈₈ and 0.022 for AAOD₅₀₀ and it shows a slight fluctuation during 2005–2015, and a relatively fast increase after that. The high-value center of the OMI AAOD is predominantly scattered over the northwestern TB region, the BTH region and the surrounding northern part of Henan province, and the SB in China.

The annual average AAOD₅₀₀ during 2005–2018 follows the order of the TB (0.041) > the YRB (0.023) > BT (0.026) > the SB (0.023) > NS (0.021) > the PRD (0.017). Our results indicated that the AAOD_388nm and AAOD_500nm increases by a factor of approximately 1.5, especially in the northwest by 2 times. The situation of absorptive aerosol pollution became severe in the recent years of 2016–2018. The AAE_388–500 depicts the opposite trend except for the TB with a high value of 3.058. The winter or spring seasons have a higher mean AAOD, followed by autumn, whilst the summer displays a relatively weaker value. The monthly average AAOD₃₈₈ in the TB, SB, YRB, BT, PRD, and NS regions shows the smallest values of 0.072, 0.024, 0.026, 0.027 in July, 0.024 in June, and 0.025 in September, respectively, whereas the peak events of AAOD₃₈₈ occur in January for NS (0.055), the YRB (0.077) and BT (0.067), April for the TB (0.123), February for the SB (0.073), and March for the PRD (0.075), respectively. The monthly average AAEs exhibit the smallest value in December and January and largest in June. The daily AAE for episode days is generally lower than those in non-episode days. Sandstorm weather is more frequent in March, April, and May, causing high concentrations of dust aerosols and AAOD over the TB in spring. In addition, the meteorological conditions in winter and early spring associated with more energy consumption are conducive to the accumulation of high BC aerosol pollution, causing high alert levels for the AAOD from November to the following March. Future research will be conducted to further gain a more detailed understanding of the relationship between chemical composition, sources of aerosol absorption and AAOD values, as well as the need for more data to analyze the relationship.