Aerosol Components Derived from Global AERONET Measurements by GRASP: A New Value-Added Aerosol Component Global Dataset and Its Application

Xindan Zhang State Key Laboratory of Severe Weather and Key Laboratory of Atmospheric Chemistry, Chinese Academy of Meteorological Sciences, CMA, Beijing, China;
Department of Atmospheric and Oceanic Sciences, Institute of Atmospheric Sciences, Fudan University, Shanghai, China;

Search for other papers by Xindan Zhang in
Current site
Google Scholar
PubMed
Close
,
Lei Li State Key Laboratory of Severe Weather and Key Laboratory of Atmospheric Chemistry, Chinese Academy of Meteorological Sciences, CMA, Beijing, China;

Search for other papers by Lei Li in
Current site
Google Scholar
PubMed
Close
,
Huizheng Che State Key Laboratory of Severe Weather and Key Laboratory of Atmospheric Chemistry, Chinese Academy of Meteorological Sciences, CMA, Beijing, China;

Search for other papers by Huizheng Che in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0002-9458-3387
,
Oleg Dubovik Université de Lille, CNRS, UMR 8518 – Laboratoire d’Optique Atmosphérique, Lille, France;

Search for other papers by Oleg Dubovik in
Current site
Google Scholar
PubMed
Close
,
Yevgeny Derimian Université de Lille, CNRS, UMR 8518 – Laboratoire d’Optique Atmosphérique, Lille, France;

Search for other papers by Yevgeny Derimian in
Current site
Google Scholar
PubMed
Close
,
Brent Holben NASA Goddard Space Flight Center, Greenbelt, Maryland;

Search for other papers by Brent Holben in
Current site
Google Scholar
PubMed
Close
,
Pawan Gupta STI, Universities Space Research Association, Huntsville, Alabama;
NASA Marshall Space Flight Center, Huntsville, Alabama;

Search for other papers by Pawan Gupta in
Current site
Google Scholar
PubMed
Close
,
Thomas F. Eck NASA Goddard Space Flight Center, University of Maryland, Baltimore County, Baltimore, Maryland;

Search for other papers by Thomas F. Eck in
Current site
Google Scholar
PubMed
Close
,
Elena S. Lind Department of Electrical and Computer Engineering, Virginia Tech, Blacksburg, Virginia;

Search for other papers by Elena S. Lind in
Current site
Google Scholar
PubMed
Close
,
Carlos Toledano Grupo de Optica Atmosférica, Universidad de Valladolid, Paseo Prado de la Magdalena, Valladolid, Spain;

Search for other papers by Carlos Toledano in
Current site
Google Scholar
PubMed
Close
,
Xiangao Xia Key Laboratory of Middle Atmosphere and Global Environment Observation, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China

Search for other papers by Xiangao Xia in
Current site
Google Scholar
PubMed
Close
,
Yu Zheng State Key Laboratory of Severe Weather and Key Laboratory of Atmospheric Chemistry, Chinese Academy of Meteorological Sciences, CMA, Beijing, China;

Search for other papers by Yu Zheng in
Current site
Google Scholar
PubMed
Close
,
Ke Gui State Key Laboratory of Severe Weather and Key Laboratory of Atmospheric Chemistry, Chinese Academy of Meteorological Sciences, CMA, Beijing, China;

Search for other papers by Ke Gui in
Current site
Google Scholar
PubMed
Close
, and
Xiaoye Zhang State Key Laboratory of Severe Weather and Key Laboratory of Atmospheric Chemistry, Chinese Academy of Meteorological Sciences, CMA, Beijing, China;

Search for other papers by Xiaoye Zhang in
Current site
Google Scholar
PubMed
Close
Open access

Abstract

Aerosols affect Earth’s climate both directly and indirectly, which is the largest uncertainty in the assessment of radiative forcings affecting anthropogenic climate change. The standard Aerosol Robotic Network (AERONET) aerosol products have been widely used for more than 30 years. Currently, there is strong community interest in the possibility of determining aerosol composition directly from remote sensing observations. This work presents the results of applying such a recently developed approach by Li et al. to extended datasets of the directional sky radiances and spectral aerosol optical depth (AOD) measured by AERONET for the retrievals of aerosol components. First, the validation of aerosol optical properties retrieved by this component approach with AERONET standard products shows good agreement. Then, spatiotemporal variations of the obtained aerosol component concentration are characterized globally, especially the absorbing aerosol species (black carbon, brown carbon, and iron oxides) and scattering aerosol species (organic carbon, quartz, and inorganic salts). Finally, we compared the black carbon (BC) and dust column concentration retrievals to the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2), products in several regions of interest (Amazon zone, Indo-China Peninsula, North India, southern Africa, sub-Sahel, Gobi Desert, Middle East, Sahara Desert, and Taklamakan Desert) for new insights on the quantitative assessment of MERRA-2 aerosol composition products (R = 0.60–0.85 for BC; R = 0.75–0.90 for dust). The new value-added and long-term aerosol composition product globally is available online (https://doi.org/10.6084/m9.figshare.25415239.v1), which provides important measurements for the improvement and optimization of aerosol modeling to enhance estimation of the aerosol radiative forcing.

© 2024 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding authors: Huizheng Che, chehz@cma.gov.cn; Oleg Dubovik, oleg.dubovik@univ-lille.fr

Lind’s current affiliation: NASA Goddard Space Flight Center, Greenbelt, Maryland.

Xindan Zhang and Lei Li contributed equally to this work.

Abstract

Aerosols affect Earth’s climate both directly and indirectly, which is the largest uncertainty in the assessment of radiative forcings affecting anthropogenic climate change. The standard Aerosol Robotic Network (AERONET) aerosol products have been widely used for more than 30 years. Currently, there is strong community interest in the possibility of determining aerosol composition directly from remote sensing observations. This work presents the results of applying such a recently developed approach by Li et al. to extended datasets of the directional sky radiances and spectral aerosol optical depth (AOD) measured by AERONET for the retrievals of aerosol components. First, the validation of aerosol optical properties retrieved by this component approach with AERONET standard products shows good agreement. Then, spatiotemporal variations of the obtained aerosol component concentration are characterized globally, especially the absorbing aerosol species (black carbon, brown carbon, and iron oxides) and scattering aerosol species (organic carbon, quartz, and inorganic salts). Finally, we compared the black carbon (BC) and dust column concentration retrievals to the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2), products in several regions of interest (Amazon zone, Indo-China Peninsula, North India, southern Africa, sub-Sahel, Gobi Desert, Middle East, Sahara Desert, and Taklamakan Desert) for new insights on the quantitative assessment of MERRA-2 aerosol composition products (R = 0.60–0.85 for BC; R = 0.75–0.90 for dust). The new value-added and long-term aerosol composition product globally is available online (https://doi.org/10.6084/m9.figshare.25415239.v1), which provides important measurements for the improvement and optimization of aerosol modeling to enhance estimation of the aerosol radiative forcing.

© 2024 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding authors: Huizheng Che, chehz@cma.gov.cn; Oleg Dubovik, oleg.dubovik@univ-lille.fr

Lind’s current affiliation: NASA Goddard Space Flight Center, Greenbelt, Maryland.

Xindan Zhang and Lei Li contributed equally to this work.

1. Introduction

Atmospheric aerosols are considered to be an important climate forcing factor that plays a key role in the global climate system (Charlson et al. 1992; IPCC 2021). For example, aerosols impact the radiation balance and atmospheric stability by perturbing the vertical temperature profile through direct scattering and absorption (Balkanski et al. 2010; Chen et al. 2022). Aerosols can further indirectly change the reflectivity and lifetime of clouds by acting as cloud condensation nuclei or ice-nucleating particles (Li et al. 2022; Jia et al. 2021). Large differences in the estimation of aerosol forcings attributed to different aerosol compositions could lead to large discrepancies in climate sensitivity (Zelinka et al. 2020), and thus, aerosols are considered to be the largest contributor of uncertainty in quantifying present-day climate change. The characterization of aerosol properties in details including optical property, size, and chemical composition, has always been a significant basic research direction in the field of atmospheric science (Che et al. 2024; Ghan and Schwartz 2007; Kolb and Worsnop 2012). Over the past several decades, the development of satellite remote sensing has provided important materials for the study of aerosol optical properties. It also serves as critical data for model assimilation to improve the accuracy of model simulations and also to promote the development of widely used reanalysis data [e.g., Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2)]. Although aerosol optical products from model simulations and reanalysis data have been widely applied for ground-based validation and climate change assessment (Mortier et al. 2020; Isaza et al. 2021; Gliß et al. 2021), it is noteworthy that aerosol composition products from model simulations and reanalysis were validated only by some near-surface observations. For example, the mass concentration of carbonaceous aerosols in the MERRA-2 dataset was evaluated using the ground-based measurements at several sites over the Indo-Gangetic Plain (Soni et al. 2021). Ma et al. (2021) and Prabhu et al. (2020) reported the association of ground-based measured components with MERRA-2 located in China and the foothills of western Himalayas, respectively. A comprehensive assessment and extensive validation of the aerosol optical component products in the MERRA-2, especially for the whole atmospheric column, has not been well addressed. The greatest challenge is the lack of global, long time-scale, whole-column observations of aerosol component concentrations.

Ground-based remote sensing networks have been rapidly expanded worldwide in the past two decades throughout the world. The Aerosol Robotic Network (AERONET) has been widely considered to be a standard monitoring platform (Holben et al. 1998, 2001). AERONET provides global measurements using standard and well-calibrated instruments for over three decades. Many studies have been conducted to investigate aerosol properties on a site-specific or global scale using AERONET products (e.g., Dubovik et al. 2002a; Eck et al. 2005; Sena et al. 2013; Cordero et al. 2023). Several studies employed the differences in the spectral aerosol optical properties provided by AERONET to identify the aerosol types or to estimate the concentrations of different aerosol components (Dubovik et al. 2002b; Logothetis 2020, 2021; Choi and Ghim 2016; Choi et al. 2020; Schuster et al. 2005, 2016). Specifically, the classifications of aerosol types are based on single scattering albedo (SSA), absorbing aerosol optical depth (AAOD), absorbing Ångström exponent (AAE) for distinguishing aerosol absorbing properties, and Ångström exponent (AE), fine-mode fraction (FMF) for distinguishing the particle size of aerosols (Dubovik et al. 2002b; Giles et al. 2012; Russell et al. 2010; Lee 2010; Ozdemir 2020; Shin et al. 2019). As for the estimation of aerosol component concentrations, an additional analysis of AERONET retrieved aerosol complex refractive index was employed (Schuster et al. 2005, 2016; Dey et al. 2006; Koven and Fung 2006; Sato et al. 2003; Arola et al. 2011). This analysis allowed the derivations of aerosol component concentrations using the differences in spectral variations of absorbing and scattering properties for each aerosol component. These previous aerosol component retrievals mentioned above could have large uncertainties because the estimation of aerosol component concentration was based on the aerosol optical retrievals (e.g., the complex refractive index, SSA) provided by AERONET (Schuster et al. 2016; Sinyuk et al. 2020). Recently, component inversion approach (Li et al. 2019) has been developed in the framework of the Generalized Retrieval of Atmosphere and Surface Properties (GRASP) algorithm (Dubovik et al. 2011, 2014, 2021), which can directly retrieve aerosol component concentration from AERONET radiance and spectral AOD measurements. The results obtained with that approach were analyzed by Li et al. (2022a,b) and Zhang et al. (2022) and showed to provide robust and valuable additional aerosol information based on remote sensing measurements.

The goal of this study is to process all the global AERONET directional sky radiance and spectral AOD measurements at 918 sites by the GRASP/component approach for generating the new value-added and long-term aerosol composition products globally with high accuracy. We also validate the retrievals of aerosol optical property inverted by GRASP/component approach with all corresponding AERONET standard products. The spatiotemporal variations of aerosol composition concentration globally are characterized using the newly retrieved aerosol component dataset. The new insight quantitative comparisons of MERRA-2 black carbon (MERRA-2 BC) and dust column concentration to measurements are conducted in several regions of interest globally.

2. Data and methods

a. AERONET measurements and products.

AERONET can provide long-term sun photometer measurements and aerosol optical property products globally (Holben et al. 1998). Giles et al. (2019) describes the AERONET algorithms for cloud screening and quality control of the AERONET version 3 AOD datasets, and Sinyuk et al. (2020) describe the retrievals and uncertainties of the retrieval products for the version 3 dataset. The estimated uncertainty in AERONET measured AOD, due primarily to calibration uncertainty, is about 0.01–0.02 at optical air mass of one for network field instruments (with the highest errors in the UV; Eck et al. 1999). The AERONET inversion fits the measured spectral AOD to within 0.01 consistent with the measurement accuracy of the direct Sun measured AOD. In the long-term trend study, the level 1.5 measurements with some screening at AERONET sites can be chosen due to the available and extensive data record (Li et al. 2014). For example, the limiting of level 1.5 retrievals to AOD values higher than 0.2 at AERONET La Paz site can fulfill the level 2.0 sky conditions (Pérez-Ramírez et al. 2017). Therefore, for a preliminary examination of aerosol absorbing and scattering component content worldwide, which is still limited in the literature, we make use of level 1.5 measurements with some screening. Here, in this work, AERONET total optical depth (TOD) and filtered almucantar observations (440, 675, 870, and 1020 nm) in level 1.5 were used as the input for GRASP/component approach to ensure cover the majority of the stations, especially those located in North America (NA), Europe (EU), and the Southern Hemisphere, where AOD is usually low (Li et al. 2014). We have to point out that AERONET level 1.5 measurements could have some poorly constrained and/or biased measurement inputs. For example, the almucantar scattering angle range is small in some small solar zenith angle measurements and the sky radiances have much reduced information content, which might result in high sky radiance residuals. Therefore, we then use quality-assured AERONET level 2.0 standard products as a reference to validate corresponding aerosol optical retrievals inverted by GRASP/component approach. Additionally, we use the total residuals of retrievals < 10% for the GRASP/AERONET quality-assured products. The aerosol absorbing and scattering component products in our distributing data are labeled as high uncertainty when AOD_440nm ≤ 0.2; medium uncertainty when 0.2 < AOD_440nm ≤ 0.4; and low uncertainty when AOD_440nm ≥ 0.4. The spatial distribution of all inverted AERONET sites is shown in Fig. 1 including 918 sites in total for retrievals and 899 sites for quality control. To ensure temporal continuity, we selected the sites that can provide measurements at least 6 years with no more than a 2-yr break in a 10-yr record. Thus, there are 191 sites selected with quality control during the period 2012–21 for the analysis of spatiotemporal variations of aerosol component concentration and 70 sites chosen for the 20-yr (2002–21) continuous observations.

Fig. 1.
Fig. 1.

The information of AERONET measurements: (a) spatial distribution of sites and (b) the total number of inversion month for all sites.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

b. MERRA-2 products.

The MERRA-2 reanalysis dataset generated by NASA’s Global Modeling and Assimilation Office (GMAO) based on the Goddard Earth Observing System Model, version 5 (GEOS-5), is currently one of the most significant long-term global databases (Gelaro et al. 2017; Randles et al. 2017). Its 1-hourly, 3-hourly, and monthly products are available at the website (https://disc.gsfc.nasa.gov). In this study, we downloaded hourly BC, and dust column mass concentrations, total and speciated extinction, and scattering AOD (SAOD) data at 550 nm from the MERRA-2 tavg1_2d_aer_Nx, with a grid resolution of 0.5° × 0.625°. The hourly MERRA-2 products have been chosen when the AERONET site is in the corresponding grid. And then, a total of 713 sites can be identified with at least 30 matched data at each site. It is noted that MERRA-2 utilizes the AERONET spectral AOD data as input to a neural net retrieval (NNR) algorithm, which improves the agreement between MERRA-2 AOD with the sun photometer measured AOD for the assimilated MODIS cloud-free radiances (Randles et al. 2017). The quantitative comparisons of AOD, AAOD, SAOD, and BC and dust components between the MERRA-2 and GRASP/component retrievals were conducted globally and in several regions of interest using the matched data. The definition of the relative difference is described in Eq. (1), which is used for the statistics of comparisons between GRASP/AERONET retrievals and MERRA-2 products
relative difference=2N|i=1NMERRAii=1NGRASPii=1NMERRAi+i=1NGRASPi|.

c. Component inversion approach.

In the GRASP/component approach, it is assumed that the aerosol mixture is internally mixed with multiple aerosol components of known complex refractive indices (Table 1). The GRASP/component approach imposed additional physical constraints on the spectral dependence of complex refractive indices as compared to other retrieval algorithms (without fixed refractive index of each component), because the complex refractive index of each assumed component is fixed and the fraction of each component is used to drive the calculation of aerosol mixture refractive indices in Eqs. (2)(4), as
εMG=εm[1+3(f1ε1εmε1+2εm+f2ε2εmε2+2εm)1f1ε1εmε1+2εmf2ε2εmε2+2εm],
mr=εr2+εi2+εr2,
mi=εr2+εi2εr2,
where em, e1, and e2 indicate the complex dielectric functions of the host solution matrix and employed components, f1 and f2 indicate the volume fractions of the employed components, mr and mi indicates the real and imaginary parts of the mixture refractive index, respectively, and er and ei indicate the real and imaginary components of the mixture dielectric function, eMG, respectively. Therefore, GRASP/component approach directly retrieves aerosol components from radiance and spectral AOD measurements, without the intermediate retrieval of the complex refractive indices.
Table 1.

Description of aerosol components employed in the GRASP/component approach. “AN” denotes ammonium nitrate, which can be used to create a host in aerosols. “AS” denotes ammonium sulfate, which is an alternative species for host estimation in aerosols.

Table 1.

In the GRASP/component approach, the Maxwell–Garnet (MG) mixing rule [Eqs. (2)(4)] was employed to calculate the mixture refractive indices based on the fraction of components and their fixed refractive index, which has been widely employed in many aerosol component algorithms (Schuster et al. 2005, 2016; Dey et al. 2006; Arola et al. 2011). In the MG, several insoluble particles are assumed suspended in a host solution (Lesins et al. 2002). The complex refractive index of the host solution can be estimated by the quantitative description of hygroscopic properties of a certain soluble inorganic salt with water. The complex refractive indices of host solutions calculated by ammonium sulfate, ammonium nitrate, and sea salt are very similar, even though their dry refractive indices are different (Schuster et al. 2009). In this work, we selected the ammonium nitrate for the estimation of host solution refractive index (Tang 1996). GRASP/component approach has been successfully applied to different satellite observations for aerosol component retrievals (Li et al. 2019, 2020, 2022a,b).

The major aerosol components assumed in the GRASP/component approach are listed in Table 1. There are two kinds of absorbing components commonly found in the atmosphere: absorbing carbon and mineral dusts containing iron oxides (Sokolik and Toon 1999). Absorbing carbon includes BC and brown carbon (BrC), which are distinguished by the difference of spectral absorption (Bond and Bergstrom 2006; Chen and Bond 2010). Because carbonaceous aerosols are mainly dominated in the fine mode, BC and BrC are classified as fine-mode absorbing insoluble components. Although BrC and iron oxides contained in dust particles exhibit similar spectral absorption, they are differentiated based on particle size. The coarse-mode absorbing insoluble (CAI) species, which are assumed to be the only absorbing insoluble species in the coarse mode, mainly represents the iron oxides contained in the mineral dust particles (Di Biagio et al. 2019). Coarse-mode nonabsorbing components include insoluble (CNAI) species representing mainly scattering dust and nonabsorbing organic carbon, soluble (CNAS) species representing inorganic salts. Similarly, the fine-mode nonabsorbing components include insoluble species [fine-mode nonabsorbing insoluble (FNAI) and fine-mode nonabsorbing soluble species (FNAS)].

3. Results and discussion

a. Comparisons of GRASP/component and MERRA-2 aerosol optical properties to AERONET standard products.

Figure 2 shows the comparisons of GRASP/component and MERRA-2 aerosol optical properties (AOD, AAOD, SSA, and AE) to AERONET standard products. The statistical metrics such as correlation coefficient R, bias, mean absolute error (MAE), and root-mean-square error (RMSE) demonstrate that agreements between GRASP/component and AERONET products are much better than that for MERRA-2 products (Figs. 2a–d vs Figs. 2e–h). In particular, the R = 0.965, RMSE = 0.006, and MAE = 0.005 are obtained for the GRASP AAOD validation (Fig. 2b) and R = 0.951, RMSE = 0.015, and MAE = 0.011 are obtained for GRASP SSA validation (Fig. 2c). However, for the MERRA-2 AAOD and SSA validation, the parameters of R = 0.549, RMSE = 0.026, MAE = 0.018, and R = 0.485, RMSE = 0.038, MAE = 0.029 are obtained (Figs. 2f,g), respectively. AAOD and SSA products are strongly related to the aerosol absorption and scattering properties. The statistics of comparisons between GRASP/component and AERONET products for the other wavelengths are listed in Table 2, which also indicate very good consistency. Overall, the retrievals of aerosol optical properties derived by the GRASP/component approach from AERONET sun photometer measurements present much better agreement with AERONET standard products than that of MERRA-2 aerosol optical properties. It is noted that both the algorithmic and scientific basis for the retrievals of aerosol optical properties from sun photometer measurements in both AERONET and GRASP algorithms are similar.

Fig. 2.
Fig. 2.

Comparisons of aerosol optical products derived by the GRASP/component approach from sun photometer measurements and the MERRA-2 aerosol optical products to the corresponding AERONET standard products at all sites for (a),(e) AOD, (b),(f) AAOD, (c),(g) SSA, and (d),(h) AE.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

Table 2.

The statistics for the intercomparison of aerosol optical properties derived by the GRASP/component approach from sun photometer measurements to the corresponding AERONET products.

Table 2.

b. The distributions of absorbing aerosol component concentration.

Figure 3 illustrates the spatial distribution of BC column concentrations retrieved by the GRASP/component approach from AERONET observations at 191 sites during 2012–21. We can see that high BC concentrations in Southeast Asia (AS), South Asia, southern Africa (AF), and South America (SA) are identified, which is consistent with Wang et al. (2023). In these regions, motor vehicle emissions, power generation, and industrial emissions, along with forest, peatland, and agricultural fires, can make significant contributions to BC concentration, especially in Southeast Asia. In Thailand and Laos, the annual BC concentration at the Silpakorn_Univ site has the maximum of 18.7 mg m−2, which may be attributed to its frequent forest fires. Due to factors such as slash-and-burn agriculture, industrial pollution, forest fires, and unfavorable meteorological conditions, a significant portion of Southeast Asia experiences severe pollution with elevated BC concentrations during the dry season. BC emissions in South Asia primarily originate from various sources such as transportation, industry, biomass burning, and coal combustion (Gargava et al. 2014; Chowdhury et al. 2007). India is one of the most air-polluted countries in the world with rapid industrialization, economic development, and population growth, especially in the densely populated northern areas of India. The peak of BC concentration (14.5 mg m−2 at Gandhi_College site) is observed during the winter/postmonsoon periods because of the unfavorable dispersion conditions and increased seasonal emission sources (Kumar et al. 2020). The highest BC in SON was observed at the Mongu_Inn site (17.6 mg m−2) in Zambia among all AERONET sites. In South America, BC concentrations in the Amazon region are higher than that in other regions, with the maximum in the SON season (10.1 mg m−2 at the Medellin site) followed by the MAM season (4.7 mg m−2 at the Sao_Paulo site). The strong biomass burning events occurring in northern Australia (AU) (Mitchell et al. 2006; Yang et al. 2021) during DJF and SON seasons generate many BC particles (5.8 and 4.0 mg m−2 in DJF and SON seasons, respectively, at the Lake_Argyle site).

Fig. 3.
Fig. 3.

Spatial distribution of BC column mass concentration (mg m−2) derived from AERONET observations at 191 sites using the GRASP/component approach for the period 2012–21.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

Generally, BC and BrC have similar spatial and temporal distribution (Fig. 4 for BrC). In Southeast Asia, elevated BrC concentrations are observed notably with a peak during the dry season (e.g., 4.8 mg m−2 at the Chiang_Mai_Met_Sta site). This phenomenon is primarily influenced by the factors such as biomass burning agricultural activities, forest fires, and prevailing meteorological conditions (Phairuang et al. 2019; Duc et al. 2021). BrC exhibit relatively higher values in South Asia during both DJF and JJA seasons, which can be attributed to seasonal pollution and intense dust influence, which might be related to iron oxides in fine particles and therefore a false identification of BrC. BrC concentrations are relatively high in South America (e.g., 3.3 mg m−2 at the CUIABA_MIRANDA site, 3.2 mg m−2 at the Alta_Floresta site for the annual concentration, and 12.6 mg m−2 in SON at the Medellin site). In North America, BrC concentrations are relatively low overall with relatively higher concentrations in the SON season and similarly at the northern and eastern margins of the United States. In the Sahel region, BrC exhibits its highest annual concentrations, which could be attributed to the biomass burning events and the combustion in cooking. We should point that in the Sahel region, a high CAI concentration was observed and the similarity in spectral absorption of BrC and CAI may lead to compositional misjudgment in the current retrievals (Li et al. 2019). Future endeavors will incorporate iron-containing dust into fine-mode aerosol fractions.

Fig. 4.
Fig. 4.

As in Fig. 3, but for BrC.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

The CAI is primarily distributed in regions such as the Sahara Desert, the Middle East, and South Asia (shown in Fig. 5). The Sahara region exhibits the highest CAI concentrations, surpassing other areas throughout the year, with peak concentrations during the JJA season (more than 50 mg m−2 at Banizoumbou and IER_Cinzana sites with the mass percentage of 2.0% and 1.8% in the dust, respectively; more than 30 mg m−2 at Dakar, Tamanrasset_INM, and Capo_Verde sites with the mass percentage of 2.4%, 2.5%, and 2.3% in the dust, respectively) and lowest values in the DJF season (lower than 16 mg m−2). The CAI mass fraction in our component retrievals present good agreement with the measurements in field campaigns that iron oxides represented between 2.4% and 4.5% of the total dust in the western Africa (Formenti et al. 2008). Similarly, the Middle East witnesses a JJA peak (e.g., 21.7 mg m−2 at the KAUST_Campus site with the mass percentage of 1.3% in the dust) and DJF low (e.g., 1.1 mg m−2 at the SEDE_BOKER site) in CAI concentrations. In South Asia, the highest CAI concentrations occur during the JJA and SON seasons. It is noted that CAI concentrations in Europe exhibit relatively higher levels during the JJA and SON seasons, aligning with the seasonal distribution observed in the American continent. The distributions of the CAI concentration have typical seasonal cycles in different regions.

Fig. 5.
Fig. 5.

As in Fig. 3, but for CAI component.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

c. The distributions of scattering insoluble aerosol component concentration.

Figure 6 shows the distribution of CNAI component column concentration. We can see that high CNAI concentrations are retrieved in the Sahara Desert region (e.g., more than 1580 mg m−2 at IER_Cinzana, Banizoumbou, Dakar sites for the annual concentration). A pronounced seasonal cycle that the maximum values of CNAI concentration are observed during the MAM and JJA seasons over North Africa, which is consistent with Prospero et al. (2002) and Ginoux et al. (2012). High CNAI concentrations such as 1756 mg m−2 at the Pune site and 1513 mg m−2 at the Mezaira site for the annual concentration are also found in South Asia and Middle East. Because of the intense winds and mesoscale convective activities (Choobari et al. 2014), CNAI concentrations observed in the Indian region during the JJA season can be up to 2530 mg m−2 at the Kanpur site. Elevated CNAI concentrations mostly occurred in the alternating MAM and JJA seasons in East Asia (e.g., 1600 mg m−2 at the Beijing site). Relatively low CNAI concentrations are observed at most of sites in Europe, where the variations of dust concentrations are impacted by the seasonal dependence of the intensity of dust sources in Africa and seasonal variations in atmospheric circulation (Barkan et al. 2004; Barkan and Alpert 2008). It is the same situation for the sites in North America that some CNAI particles retrieved in the Caribbean, Central America, and southern United States during the JJA season could be transported by tropical easterly winds from North Africa (Griffin et al. 2002). African dust (e.g., from the emission in the Bodélé depression over West Africa) is mainly transported toward the Amazon region in South America during DJF and MAM seasons (Gläser et al. 2015; Ben-Ami et al. 2010) with an averaged transport duration of 7–10 days, which provides essential minerals to the Amazon rain forest (Ben-Ami et al. 2010; Gläser et al. 2015; Huang et al. 2010), even for the JJA season (Bi et al. 2024). The comprehensive characterization and analysis of coarse-mode aerosol loading and composition in the Amazonian region have demonstrated that the frequent intrusion of African long-range transport aerosols including Saharan dust and African biomass burning smoke, sea salts from the Atlantic Ocean, bioaerosols released by the rain forest ecosystem, and coarse smoke particles from local deforestation fires altogether consist of the remarkably constant concentration of coarse-mode aerosols with rather weak seasonality (Moran-Zuloaga et al. 2018). For most of the time, bioaerosol particles emitted from the forest biome are dominated in the coarse-mode background quantity (Moran-Zuloaga et al. 2018). It is probably in the same situation for the Southeast Asian region that exhibits similar levels of MODIS coarse-mode AOD (MODIS CAOD) to the Amazonian region (Luo et al. 2024).

Fig. 6.
Fig. 6.

As in Fig. 3, but for CNAI component.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

The distributions of the FNAI concentration (in Fig. 7) are quite different from the CNAI concentration, mainly associated with mineral dust particles. The levels of the FNAI concentration are dependent on the industrial emissions, anthropogenic activity, and biomass burning events, as well as some transported fine-mode dust. Therefore, notably high FNAI concentrations are observed at the Asian sites covering Gandhi_College (212.6 mg m−2), Kanpur (204.8 mg m−2), Chiang_Mai_Met_Sta (188.7 mg m−2), and Beijing (131 mg m−2) sites for the annual average. In addition to fossil fuel combustion and biomass burning, the open burning of plastics in metropolitan areas also contributes to organic aerosols in South Asia (Fu et al. 2010). Specifically, the Kanpur and Gandhi_College sites in India present very high FNAI concentrations during the DJF (322.1 mg m−2 at the Kanpur site), JJA (182.9 mg m−2 at the Gandhi_College site), and MAM (181.3 mg m−2 at the Kanpur site) seasons. The high FNAI concentrations in DJF and MAM can be attributed to the fact that colder seasons in India are associated with higher levels of pollution, including high emissions from biomass burning with large amounts of organic aerosols (Bhandari et al. 2020). However, FNAI concentrations in the JJA season can be characterized by the fine-mode aerosols generated from human activities and industrial emissions, together with frequently transported fine dust particles. We also should point out that in the SON season, very high FNAI concentrations are observed at the Mongu_Inn site (316 mg m−2) in southern Africa and the Alta_Floresta site (211.8 mg m−2) and CUIABA_MIRANDA site (153.1 mg m−2) in South America, which are emitted from strong biomass burning events.

Fig. 7.
Fig. 7.

As in Fig. 3, but for FNAI component.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

d. Characterizing the long-term patterns of aerosol component concentration.

Figure 8 shows the retrievals of (panel a) total, (panel b) absorbing, and (panel c) scattering aerosol mass concentrations at AERONET sites for 20 years of continuous observations. High aerosol concentration sites are predominantly located in the Sahara Desert region and densely populated industrial areas of Asia (Fig. 8a). Among these sites, the highest aerosol mass concentration (more than 2800 mg m−2) was observed at the IER_Cinzana site in the Sahara Desert region, primarily attributed to the presence of large-sized and high-density dust particles. Urban sites such as Kanpur, Pune, and Beijing, influenced by human activities and industrial development together with the transported dust within specific seasons, also display very high aerosol concentrations (around 2000 mg m−2). Because of large emissions from strong biomass burning events, the AERONET sites (e.g., Alta_Floresta and CUIABA_MIRANDA) in the regions dominated by biomass burning aerosol also present high aerosol mass concentrations. On the other hand, in most European sites, aerosol component concentrations are relatively close and slightly higher compared to many North American sites. The Mauna_Loa site located on the northern slope of the Mauna Loa volcano exhibits the lowest aerosol component concentration (31.7 mg m−2) because of its high-altitude position and being far away from the influences of human activities.

Fig. 8.
Fig. 8.

Distributions of (a) total aerosol mass concentrations (mg m−2) with error bars indicating the standard deviations of the component retrievals; (b) different absorbing aerosol concentrations (mg m−2); and (c) different scattering aerosol concentrations (mg m−2) for AERONET sites with continuous observations more than 20 years.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

High concentrations of absorbing aerosol components are found at AERONET sites in Africa, Asia, and South America (Fig. 8b). Specifically, important contributions of CAI to high-absorbing component concentrations in northern Africa (e.g., 29.2 mg m−2 CAI at the IER_Cinzana site, 24.3 mg m−2 CAI at the Banizoumbou site, and 23.2 mg m−2 CAI at the Dakar site) are influenced by strong dust storm with high emissions. The situation in India and China is different. For example, high-absorbing component concentrations are observed at Kanpur and Beijing sites, where aerosols are impacted by both seasonal dust and strong anthropogenic pollution, comparable contributions of BC and CAI to the absorbing component concentrations (11.5 mg m−2 BC, 10.4 mg m−2 CAI for Kanpur; 9.6 mg m−2 BC, 6.3 mg m−2 CAI for Beijing). A significant predominance of BC and BrC generated from biomass burning events was obtained, such as 11.2 mg m−2 BC and BrC retrieved at the CUIABA_MIRANDA site in South America, where wildfires occurred, with the savanna vegetation making important contributions to absorbing component concentrations. It is noted that BC mass concentrations are higher than BrC mass concentrations at some sites. The main seasons for this situation can be 1) black and organic carbons are the major components in biomass burning aerosol; BrC as a part of the organic carbon representing a species with significant absorption at 440 nm could have low concentrations (supplemental Fig. 1 in the online supplemental material) (Li et al. 2010; Moosmüller et al. 2009); 2) previous studies demonstrated that the fresh biomass burning aerosols have a relatively high BC proportion and the aged biomass burning aerosols have a relatively high BrC proportion (Li et al. 2020; Abel et al. 2003; Haywood et al. 2003; Reid et al. 1998). Aging processes have important effects on the chemical and optical properties, especially in the regions where atmospheric gases (O3, NO2, and SO2) have an elevated content (Pósfai et al. 2003; Zhang et al. 2008; Yokelson et al. 2009; Adachi and Buseck 2013; China et al. 2013). The Mexico_City site located in North America presents slightly high absorbing component concentrations associated with human activities, meteorological conditions, and geographic factors (Raga et al. 2001). Furthermore, it is evident that several sites in the eastern United States, aligning with the distribution of densely populated areas, consistently present some BC retrievals not to be ignored. The very low-absorbing component concentrations are retrieved at Mauna_Loa (0.2 mg m−2 in total) and Izana (1.2 mg m−2 in total) sites, which are background sites in the ocean or at elevated altitude with less influence of anthropogenic activity. We should point out that the retrievals of absorbing components with very low aerosol loading will have extremely high uncertainty (Sinyuk et al. 2020; Li et al. 2019).

The characteristics of scattering aerosol component concentration are shown in Fig. 8c. We should point out that the CNAI concentration is substantially higher than the other scattering component concentrations because of the significant contributions of mineral dust with large particle size and density to the coarse-mode aerosol particles. However, since fine-mode particles have much greater scattering efficiency as compared to coarse-mode particles, the contributions of coarse-mode dust particles to scattering AOD are as not prominent as to mass concentrations. Specifically, the IER_Cinzana, Banizoumbou, and Dakar sites in Africa as the typical dust aerosols present the maximum CNAI concentrations accounting for more than 85% of the total scattering aerosol concentration at these sites. In addition, the Capo_Verde site located in the Atlantic Ocean, which is influenced significantly by the dust aerosol emitted from the Sahara Desert region, also shows a high concentration of scattering component with the CNAI proportion of approximately 80%. Among the sites in Asia, remarkably high concentrations of scattering aerosol components are observed at some sites (e.g., 2148.3 mg m−2 at the Kanpur site, 1726.1 mg m−2 at the Pune site, and 1600.6 mg m−2 at the Beijing site). Various aerosol emissions (e.g., dust emissions from natural sources and anthropogenic sulfate and organic carbon emissions) make significant contributions to these scattering aerosol components (Hu et al. 2021; Mhawish et al. 2017, 2019; Zhou et al. 2020). Low concentrations of scattering aerosol components are retrieved at the Osaka and Shirahama sites in Japan for their locations near the coastline and far from strong anthropogenic emissions. The differences of the scattering aerosol component concentration among the European sites are unremarkable, except several sites (e.g., Thessaloniki and Moscow_MSU_MO), which are under the influences of dust aerosol, industrial aerosol, and biomass burning aerosol, could present a bit high concentration. All sites in North America present relatively low concentrations of the scattering aerosol component except the Mexico City site. Mexico City is one of the world’s megacities, where topography, meteorology, population growth, and its activities have combined to severely impact the local air quality, with levels of particulate matter still higher than the respective Mexican air quality standards (Mensink et al. 2020).

e. A new insight into the evaluations of MERRA-2 aerosol component products.

The new value-added and long-term aerosol composition products globally have been inverted by the GRASP/component approach from the AERONET sun photometer measurements. Therefore, it provides the possibility of a new insight into the assessment of MERRA-2 BC and dust column concentration on a global scale, which was not possible in the past for the lack of observational data. To ensure that the comparisons are representative with less uncertainty as possible, the available data selected are more than 10 days per month for the analysis. We should point out that the available aerosol component retrievals at AERONET sites in regions of interest may have different time intervals, and the number of available sites in these regions is limited for the long-term measurements. Therefore, we also provide the daily retrievals of aerosol components from the AERONET sun photometer measurements at 899 sites (https://doi.org/10.6084/m9.figshare.25415239.v1) for further in-depth analysis in future. Here, in this section, the comparisons of MERRA-2 products to GRASP/AERONET retrievals for the monthly averages including BC and dust column concentration, AOD, AAOD, and SAOD are conducted in several BC- or dust-dominated regions of interest globally (Table 3). Because aerosols in North India can be impacted by the combustion of fossil fuel and biomass fuels, together with the transported dust, therefore we selected the sites in North India as the cases in both BC- and dust-dominated regions. The sites in each region of interest together with the statistics of the comparisons for the monthly averages are listed in Table 4.

Table 3.

Comparisons of the BC and mineral dust column concentration as well as AOD, AAOD, and SAOD between MERRA-2 data and GRASP component retrievals over several key BC-dominated and dust-dominated regions for the monthly averages. The R* denotes the correlation coefficients of the BC and dust mass concentrations in the BC-dominated and dust-dominated regions, respectively.

Table 3.
Table 4.

Comparisons of the BC and mineral dust column concentration as well as AOD, AAOD, and SAOD between MERRA-2 data and GRASP/component retrievals at each site in the BC-dominated and dust-dominated regions. The “N” denotes the monthly average valid count. The R* denotes the correlation coefficients of the BC and dust mass concentrations in the BC-dominated and dust-dominated regions, respectively.

Table 4.

Figures 9 and 10 present the comparisons of MERRA-2 AOD, AAOD, and SAOD products to corresponding GRASP/AERONET retrievals in the BC-dominated and dust-dominated regions for the monthly averages, respectively. Generally, MERRA-2 and GRASP/AERONET aerosol optical property products present similar temporal variations with good consistencies (R > 0.75 for AAOD, R > 0.9 for AOD and SAOD in the BC-dominated regions; R > 0.6 for AAOD, R > 0.9 for AOD and SAOD in the dust-dominated regions). Specifically, as shown in Fig. 9 for the BC-dominated regions, we can see that the maximum differences of AOD, and SAOD are obtained in the Indo-China Peninsula (ICP) region (relative difference = 27%, 26% for AOD, SAOD, respectively). The maximum difference of AAOD is obtained in the sub-Sahel (SS) region (relative difference = 38%), which can be attributed to large differences of the strong absorbing BC component concentration in such a region between GRASP retrievals and MERRA-2 products (Fig. 11). For the dust-dominated regions (Fig. 10), the maximum differences of AOD and SAOD are obtained in the North Indian (NI) region (relative difference = 14% for AOD; relative difference = 15% for SAOD), and the Gobi Desert (GD) region shows the maximum difference of AAOD (relative difference = 19%) in the dust-dominated regions. MERRA-2 AOD and SAOD were found to be underestimated in most parts of the Northern Hemisphere except the United States (Che et al. 2019), which may be related to the lack of nitrate aerosols in the GOCART model (Buchard et al. 2017). It is noted that the correlation coefficients R of AAOD are better in the BC-dominated regions than that in the dust-dominated regions. The detailed statistics for the comparisons of monthly averages at all sites are presented in Table 4.

Fig. 9.
Fig. 9.

Statistical correlation and the comparisons of MERRA-2 products to GRASP/component retrievals based on AERONET measurements for the monthly averages of (a) AOD, (b) AAOD, and (c) SAOD in the BC-dominated regions.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

Fig. 10.
Fig. 10.

Statistical correlation and the comparisons of MERRA-2 products to GRASP/component retrievals based on AERONET measurements for the monthly averages of (a) AOD, (b) AAOD, and (c) SAOD in the dust-dominated regions.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

Fig. 11.
Fig. 11.

Statistical correlation and the comparisons of MERRA-2 component products to GRASP/component retrievals based on AERONET measurements for the monthly averages of (a),(c) BC and (b),(d) dust column concentration in the BC-dominated and dust-dominated regions, respectively. Darker colors in the same region correspond to MERRA-2 products, and lighter colors correspond to GRASP retrievals.

Citation: Bulletin of the American Meteorological Society 105, 10; 10.1175/BAMS-D-23-0260.1

Figure 11 presents the comparisons of MERRA-2 and GRASP/AERONET components for the monthly averages of BC and dust column concentration. The correlation coefficients of BC obtained in the BC-dominated regions vary from 0.60 to 0.85. The highest correlation was found in the southern African (SA) region and the lowest in the NI region (Table 3). It is noted that very low correlations are found at several sites in the NI region (e.g., R = 0.03 for the EVK2-CNR site at 5079 m, R = 0.05 for the Kyanjin_Gompa site at 3862 m, and R = −0.12 for the Kathmandu–Bode site in Table 4), which can be attributed to the very low BC concentration at the background site and the constrained emissions used in the aerosol chemistry and transport model (Li et al. 2024; Soni et al. 2021). It is noted that the GRASP/AERONET BC retrievals are about two times or more than the MERRA-2 BC products. The maximum differences of the BC concentration are observed in the SS region (relative difference = 142%). We should point out that the large differences mainly could be attributed to the assumed value of the BC density (1.8 g cm−3 for GRASP BC retrieval; 1.0 g cm−3 in the Optical Properties of Aerosols and Clouds (OPAC) for MERRA-2 BC product) (Hess et al. 1998). In OPAC data, BC has a density of 1 g cm−3, which is often employed by the transport models, but it is difficult to find such low BC density by in situ measurements (Schuster et al. 2005). Therefore, in order to better investigate the differences between GRASP BC and MERRA-2 BC, we converted the MERRA-2 BC concentration by a parameter of 1.8 (based on the BC density) (see converted MERRA-2 in supplemental Table 1). We also reprocessed the component retrievals replacing of the original BC complex refractive index in GRASP/component (see GRASP/Com. case 1 in supplemental Table 1) by the MERRA-2 BC refractive index (see GRASP/Com. case-2 in supplemental Table 1). The concentration differences based on the long-term measurements at several typical BC-dominated AERONET sites (Alta_Floresta in the Amazon zone, Chiang_Mai_Met_Sta in the Indo-China Peninsula, Kanpur in North India, and Mongu_Inn in South Africa) are shown in supplemental Fig. 2. We can see that generally the GRASP BC retrievals in both cases are still larger than the converted MERRA-2 BC concentration, especially when BC loading is high (e.g., strong biomass burning occurs in South Africa and Indo-China Peninsula). It is noted that the differences between GRASP/Com. case-2 and converted MERRA-2 become even larger when the same MERRA-2 BC complex refractive index is employed in the GRASP/component retrievals. The corresponding comparisons of AOD, AAOD, and SAOD products for GRASP/component versus AERONET and MERRA-2 versus AERONET in the BC-dominated regions are shown in supplemental Figs. 3 and 4 and supplemental Table 2. We can see that the agreement between GRASP/component and AERONET is much better than that of MERRA-2, especially the AAOD and SAOD products (e.g., the AAOD comparison in the Indo-China Peninsula in supplemental Figs. 3 and 4), which, in principle, can indicate GRASP absorbing component retrievals with high accuracy.

The variation and magnitude of the monthly BC concentration retrieved at Alta_Floresta in the Amazon zone and Mongu_Inn in South Africa (supplemental Fig. 2) are in line with the study of Schuster et al. (2005) where the range of BC concentrations varies from nearly 0 mg m−2 to more than 30 mg m−2 in different months. The monthly BC concentration retrieved at Kanpur in North India also presents good agreement with the study of Li et al. (2015). At the same time, the underestimation of the BC concentration in the MERRA-2 was reported in India, Europe, United States, and China (Prabhu et al. 2020; Buchard et al. 2016; Provençal et al. 2017; Cao et al. 2021). The main reasons for the underestimation can be as follows: 1) the aerosol absorption underestimated grossly by models (Shindell et al. 2013; Ramachandran et al. 2020); 2) poorly constrained in estimating anthropogenic emission inventories (Puliafito et al. 2017; Ali et al. 2022); and 3) large uncertainties in the simulation of meteorological parameter, geographic information, and aerosol aging processes (Ma et al. 2021; Soni et al. 2021; Wang and Zeng 2012).

The correlation coefficients of dust obtained in the dust-dominated regions, varying from 0.75 to 0.90, are better than that for BC. The highest correlation of dust was found in the NI region and the lowest in the Sahara Desert (SD) region. MERRA-2 dust also presents the underestimation of dust in dust-dominated regions. GRASP/AERONET dust retrievals are approximately three times higher than MERRA-2 dust, with the relative differences varying from 111% to 143%. The corresponding comparisons of AOD, AAOD, and SAOD products for GRASP/component versus AERONET and MERRA-2 versus AERONET in the dust-dominated regions are shown in supplemental Figs. 5 and 6 and supplemental Table 3. We can see that the agreement between GRASP/component and AERONET is much better than that of MERRA-2, especially the AAOD and SAOD products, which in principle can indicate GRASP absorbing and scattering component (dust) retrievals with high accuracy. To better investigate the differences between GRASP dust and MERRA-2 dust, we reprocessed the component retrieval replacing of the original dust complex refractive index in GRASP/component (see GRASP/Com. case 1 in supplemental Table 4) by the MERRA-2 dust refractive index (see GRASP/Com. case 2 in supplemental Table 4). We should point out that the assumed density of dust in GRASP/component retrievals (2.5 g cm−3) and MERRA-2 is almost the same (2.5 g cm−3 for dust bin 1; 2.65 g cm−3 for dust bins 2–5). The concentration differences based on the long-term measurements at several typical dust-dominated AERONET sites (Dalanzadgad in the Gobi Desert, Tamanrasset_INM in the Sahara Desert, Solar_Village in the Middle East, and Issyk-Kul in the Taklamakan Desert) are shown in supplemental Fig. 7. We can see that although the differences become smaller when the MERRA-2 dust refractive index was used in the GRASP/component retrievals, the GRASP dust retrievals in both cases generally are larger than the converted MERRA-2 dust concentration. The probable reasons for the underestimation of dust in MERRA-2 can be attributed to the different ranges of particle size (10 μm for MERRA-2 dust; 15.6 μm for GRASP/component dust) (Ukhov et al. 2021). Serious aerosol-polluted events including strong dust storms and widespread biomass burning can be incorrectly eliminated by the cloud screening in the satellite aerosol remote sensing algorithm (Song et al. 2018), which can lead to the obvious aerosol underestimation by models for the lack of satellite measurements in the assimilation. For the lack of local dust source emission, satellite- and ground-based measurements, and dust-transported processes parameterized poorly by models, MERRA-2 could have large uncertainty in reproducing the spatial locations of dust hotspots, dust concentration, and transport processes (Tao et al. 2022; Zamora et al. 2022).

To know the influence of assumed component complex refractive indices on BC and dust retrievals, we also investigated the uncertainty of BC and dust retrievals based on long-term real measurements at several AERONET sites (Alta_Floresta in the Amazon zone, Chiang_Mai_Met_Sta in the Indo-China Peninsula, Kanpur in North India, and Mongu_Inn in South Africa for the BC-dominated regions; Dalanzadgad in the Gobi Desert, Tamanrasset_INM in the Sahara Desert, Solar_Village in the Middle East, and Issyk-Kul in the Taklamakan Desert for the dust-dominated regions; see supplemental Figs. 8 and 9). We replace the original BC and dust complex refractive indices assumed in our GRASP/component retrievals (see GRASP/Com. ST-1 in supplemental Tables 5 and 6, respectively) by the BC and dust complex refractive indices used in MERRA-2 (see GRASP/Com. ST-2 in supplemental Tables 5 and 6, respectively). We can see that the monthly BC differences associated with the assumed component refractive index are less than 50%, even for the elevated concentration (higher than 20 mg m−2) in supplemental Fig. 8, and the monthly dust differences vary from ∼20% to ∼90% in supplemental Fig. 9. Therefore, the differences in BC and dust retrievals are also dependent on their own loadings.

4. Summary

The AERONET aerosol optical property products have been widely used for more than 30 years since the 1990s, which have been helpful to improve the assessment of radiative forcing, among several other applications such as satellite validation and retrieval algorithm development. In this study, we have developed an optimal approach to retrieve aerosol components directly from AERONET sun photometer measurements of directional sky radiances and spectral AOD. The long-term AERONET measurements were processed by the GRASP/component approach not only for the retrievals of aerosol optical properties but also for aerosol component inversion. The main conclusions are as follows.

The retrievals of aerosol optical properties inverted by the GRASP/component from sun photometer measurements show good agreement with the AERONET standard products (R = 1.0, RMSE = 0.003 for AOD; R = 0.96, RMSE = 0.006 for AAOD; R = 0.95, RMSE = 0.015 for SSA; R = 0.98, RMSE = 0.1 for AE). At the same time, GRASP/AERONET AOD retrievals present consistency with MERRA-2 AOD products on a global scale (R = 0.85, RMSE = 0.12). The consistency of AAOD in the BC-dominated regions (R > 0.76 for AAOD; R > 0.9 for AOD and SAOD) is better than that in the dust-dominated regions (R > 0.6 for AAOD; R > 0.9 for AOD and SAOD).

Spatiotemporal variations of the aerosol composition column concentration are characterized by the GRASP/AERONET long-term component retrievals, especially the absorbing aerosol species (black carbon, brown carbon, and iron oxides contained in dust) and scattering aerosol species (organic carbon, quartz contained in dust) in several regions. For example, high BC concentrations are mainly distributed in Southeast Asia, South Asia, southern Africa, and South America. The absorbing (CAI) and scattering (CNAI) coarse-mode dust are both concentrated mostly in the Sahara Desert, the Middle East, and South Asia.

A new value-added and long-term aerosol component retrieval database based on AERONET observations has been established, which provides the possibility of new insights into the assessment of MERRA-2 BC and dust column concentration on a global scale. Similar temporal variations of BC and dust column concentration are described by MERRA-2 and GRASP/AERONET retrievals with R of 0.60–0.85 for BC and R of 0.75–0.90 for dust in the regions of interest. Large bias differences are found in the sub-Sahel regions for BC (relative difference = 142%) and in the North Indian regions for dust (relative difference = 143%). Thus, we should pay more caution on the use of MERRA-2 component products in these regions and suggests a need for more effort on improvements in future.

The new value-added and long-term aerosol composition product globally is available online (https://doi.org/10.6084/m9.figshare.25415239.v1). This potentially valuable aerosol component dataset is expected to provide more information for the improvement and optimization of aerosol component estimation in models, which might reduce uncertainties in the assessment of radiative forcing in the future.

Acknowledgments.

This work was supported by the National Natural Science Foundation of China (42030608 and 42275195) and Youth Innovation Team of China Meteorological Administration (CMA2024QN13). The component algorithm was developed as part of the Labex CaPPA project, which is funded by the Agence Nationale de la Recherche (Grant ANR-II-LABX-0005-01). The authors appreciate the AERONET team and all PI at each site for their contributions to the measurements and product processing. The authors are grateful to the MERRA-2 product team.

Data availability statement.

The value-added and long-term aerosol composition dataset at 899 AERONET sites from 1993 to 2021 is available at https://doi.org/10.6084/m9.figshare.25415239.v1.

References

  • Abel, S. J., J. M. Haywood, E. J. Highwood, J. Li, and P. R. Buseck, 2003: Evolution of biomass burning aerosol properties from an agricultural fire in southern Africa. Geophys. Res. Lett., 30, 1783, https://doi.org/10.1029/2003GL017342.

    • Search Google Scholar
    • Export Citation
  • Adachi, K., and P. R. Buseck, 2013: Changes of ns-soot mixing states and shapes in an urban area during CalNex. J. Geophys. Res. Atmos., 118, 37233730, https://doi.org/10.1002/jgrd.50321.

    • Search Google Scholar
    • Export Citation
  • Ali, M. A., and Coauthors, 2022: Accuracy assessment of CAMS and MERRA-2 reanalysis PM2.5 and PM10 concentrations over China. Atmos. Environ., 288, 119297, https://doi.org/10.1016/j.atmosenv.2022.119297.

    • Search Google Scholar
    • Export Citation
  • Arola, A., G. Schuster, G. Myhre, S. Kazadzis, S. Dey, and S. N. Tripathi, 2011: Inferring absorbing organic carbon content from AERONET data. Atmos. Chem. Phys., 11, 215225, https://doi.org/10.5194/acp-11-215-2011.

    • Search Google Scholar
    • Export Citation
  • Balkanski, Y., G. Myhre, M. Gauss, G. Rädel, E. J. Highwood, and K. P. Shine, 2010: Direct radiative effect of aerosols emitted by transport: From road, shipping and aviation. Atmos. Chem. Phys., 10, 44774489, https://doi.org/10.5194/acp-10-4477-2010.

    • Search Google Scholar
    • Export Citation
  • Barkan, J., and P. Alpert, 2008: Synoptic patterns associated with dusty and non-dusty seasons in the Sahara. Theor. Appl. Climatol., 94, 153162, https://doi.org/10.1007/s00704-007-0354-9.

    • Search Google Scholar
    • Export Citation
  • Barkan, J., H. Kutiel, P. Alpert, and P. Kishcha, 2004: Synoptics of dust intrusion days from the African continent into the Atlantic Ocean. J. Geophys. Res., 109, D08201, https://doi.org/10.1029/2003JD004416.

    • Search Google Scholar
    • Export Citation
  • Ben-Ami, Y., I. Koren, Y. Rudich, P. Artaxo, S. T. Martin, and M. O. Andreae, 2010: Transport of North African dust from the Bodélé depression to the Amazon Basin: A case study. Atmos. Chem. Phys., 10, 75337544, https://doi.org/10.5194/acp-10-7533-2010.

    • Search Google Scholar
    • Export Citation
  • Bhandari, S., and Coauthors, 2020: Sources and atmospheric dynamics of organic aerosol in New Delhi, India: Insights from receptor modeling. Atmos. Chem. Phys., 20, 735752, https://doi.org/10.5194/acp-20-735-2020.

    • Search Google Scholar
    • Export Citation
  • Bi, H., and Coauthors, 2024: The circumglobal transport of massive African dust and its impacts on the regional circulation in remote atmosphere. Bull. Amer. Meteor. Soc., 105, E605E622, https://doi.org/10.1175/bams-d-23-0072.1.

    • Search Google Scholar
    • Export Citation
  • Bond, T. C., and R. W. Bergstrom, 2006: Light absorption by carbonaceous particles: An investigative review. Aerosol Sci. Technol., 40, 2767, https://doi.org/10.1080/02786820500421521.

    • Search Google Scholar
    • Export Citation
  • Buchard, V., and Coauthors, 2016: Evaluation of the surface PM2.5 in version 1 of the NASA MERRA aerosol reanalysis over the United States. Atmos. Environ., 125, 100111, https://doi.org/10.1016/j.atmosenv.2015.11.004.

    • Search Google Scholar
    • Export Citation
  • Buchard, V., and Coauthors, 2017: The MERRA-2 aerosol reanalysis, 1980 onward. Part II: Evaluation and case studies. J. Climate, 30, 68516872, https://doi.org/10.1175/JCLI-D-16-0613.1.

    • Search Google Scholar
    • Export Citation
  • Cao, S., and Coauthors, 2021: A long-term analysis of atmospheric black carbon MERRA-2 concentration over China during 1980–2019. Atmos. Environ., 264, 118662, https://doi.org/10.1016/j.atmosenv.2021.118662.

    • Search Google Scholar
    • Export Citation
  • Charlson, R. J., S. E. Schwartz, J. M. Hales, M. Wessinger, L. Thomas, and W. Schneider, 1992: Climate forcing by anthropogenic aerosols. Science, 255, 423430, https://doi.org/10.1126/science.255.5043.423.

    • Search Google Scholar
    • Export Citation
  • Che, H., and Coauthors, 2019: Large contribution of meteorological factors to inter-decadal changes in regional aerosol optical depth. Atmos. Chem. Phys., 19, 10 49710 523, https://doi.org/10.5194/acp-19-10497-2019.

    • Search Google Scholar
    • Export Citation
  • Che, H., and Coauthors, 2024: Aerosol optical and radiative properties and their environmental effects in China: A review. Earth-Sci. Rev., 248, 104634, https://doi.org/10.1016/j.earscirev.2023.104634.

    • Search Google Scholar
    • Export Citation
  • Chen, C., and Coauthors, 2022: Multi-angular polarimetric remote sensing to pinpoint global aerosol absorption and direct radiative forcing. Nat. Commun., 13, 7459, https://doi.org/10.1038/s41467-022-35147-y.

    • Search Google Scholar
    • Export Citation
  • Chen, Y., and T. C. Bond, 2010: Light absorption by organic carbon from wood combustion. Atmos. Chem. Phys., 10, 17731787, https://doi.org/10.5194/acp-10-1773-2010.

    • Search Google Scholar
    • Export Citation
  • China, S., C. Mazzoleni, K. Gorkowski, A. C. Aiken, and M. K. Dubey, 2013: Morphology and mixing state of individual freshly emitted wildfire carbonaceous particles. Nat. Commun., 4, 2122, https://doi.org/10.1038/ncomms3122.

    • Search Google Scholar
    • Export Citation
  • Choi, Y., and Y. S. Ghim, 2016: Estimation of columnar concentrations of absorbing and scattering fine mode aerosol components using AERONET data. J. Geophys. Res. Atmos., 121, 13 62813 640, https://doi.org/10.1002/2016JD025080.

    • Search Google Scholar
    • Export Citation
  • Choi, Y., Y. S. Ghim, Y. Zhang, S.-M. Park, and I.-h. Song, 2020: Estimation of surface concentrations of black carbon from long-term measurements at AERONET sites over Korea. Remote Sens., 12, 3904, https://doi.org/10.3390/rs12233904.

    • Search Google Scholar
    • Export Citation
  • Choobari, O. A., P. Zawar-Reza, and A. Sturman, 2014: The global distribution of mineral dust and its impacts on the climate system: A review. Atmos. Res., 138, 152165, https://doi.org/10.1016/j.atmosres.2013.11.007.

    • Search Google Scholar
    • Export Citation
  • Chowdhury, Z., M. Zheng, J. J. Schauer, R. J. Sheesley, L. G. Salmon, G. R. Cass, and A. G. Russell, 2007: Speciation of ambient fine organic carbon particles and source apportionment of PM2.5 in Indian cities. J. Geophys. Res., 112, D15303, https://doi.org/10.1029/2007JD008386.

    • Search Google Scholar
    • Export Citation
  • Cordero, R. R., and Coauthors, 2023: Surface solar extremes in the most irradiated region on Earth, Altiplano. Bull. Amer. Meteor. Soc., 104, E1206E1221, https://doi.org/10.1175/BAMS-D-22-0215.1.

    • Search Google Scholar
    • Export Citation
  • Dey, S., S. N. Tripathi, R. P. Singh, and B. N. Holben, 2006: Retrieval of black carbon and specific absorption over Kanpur city, northern India during 2001–2003 using AERONET data. Atmos. Environ., 40, 445456, https://doi.org/10.1016/j.atmosenv.2005.09.053.

    • Search Google Scholar
    • Export Citation
  • Di Biagio, C., and Coauthors, 2019: Complex refractive indices and single-scattering albedo of global dust aerosols in the shortwave spectrum and relationship to size and iron content. Atmos. Chem. Phys., 19, 15 50315 531, https://doi.org/10.5194/acp-19-15503-2019.

    • Search Google Scholar
    • Export Citation
  • Dubovik, O., B. Holben, T. F. Eck, A. Smirnov, Y. J. Kaufman, M. D. King, D. Tanré, and I. Slutsker, 2002a: Variability of absorption and optical properties of key aerosol types observed in worldwide locations. J. Atmos. Sci., 59, 590608, https://doi.org/10.1175/1520-0469(2002)059<0590:VOAAOP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Dubovik, O., B. N. Holben, T. Lapyonok, A. Sinyuk, M. I. Mishchenko, P. Yang, and I. Slutsker, 2002b: Non-spherical aerosol retrieval method employing light scattering by spheroids. Geophys. Res. Lett., 29, 1415, https://doi.org/10.1029/2001GL014506.

    • Search Google Scholar
    • Export Citation
  • Dubovik, O., and Coauthors, 2011: Statistically optimized inversion algorithm for enhanced retrieval of aerosol properties from spectral multi-angle polarimetric satellite observations. Atmos. Meas. Tech., 4, 9751018, https://doi.org/10.5194/amt-4-975-2011.

    • Search Google Scholar
    • Export Citation
  • Dubovik, O., and Coauthors, 2014: GRASP: A versatile algorithm for characterizing the atmosphere. SPIE Newsroom, 19 September, https://doi.org/10.1117/2.1201408.005558.

  • Dubovik, O., and Coauthors, 2021: A comprehensive description of multi-term LSM for applying multiple a priori constraints in problems of atmospheric remote sensing: GRASP algorithm, concept, and applications. Front. Remote Sens., 2, 706851, https://doi.org/10.3389/frsen.2021.706851.

    • Search Google Scholar
    • Export Citation
  • Duc, H. N., H. Q. Bang, N. H. Quan, and N. X. Quang, 2021: Impact of biomass burnings in Southeast Asia on air quality and pollutant transport during the end of the 2019 dry season. Environ. Monit. Assess., 193, 565, https://doi.org/10.1007/s10661-021-09259-9.

    • Search Google Scholar
    • Export Citation
  • Eck, T. F., B. N. Holben, J. S. Reid, O. Dubovik, A. Smirnov, N. T. O’Neill, I. Slutsker, and S. Kinne, 1999: Wavelength dependence of the optical depth of biomass burning, urban, and desert dust aerosols. J. Geophys. Res., 104, 31 33331 349, https://doi.org/10.1029/1999JD900923.

    • Search Google Scholar
    • Export Citation
  • Eck, T. F., and Coauthors, 2005: Columnar aerosol optical properties at AERONET sites in central eastern Asia and aerosol transport to the tropical mid-Pacific. J. Geophys. Res., 110, D06202, https://doi.org/10.1029/2004JD005274.

    • Search Google Scholar
    • Export Citation
  • Formenti, P., and Coauthors, 2008: Regional variability of the composition of mineral dust from western Africa: Results from the AMMA SOP0/DABEX and DODO field campaigns. J. Geophys. Res., 113, D00C13, https://doi.org/10.1029/2008JD009903.

    • Search Google Scholar
    • Export Citation
  • Fu, P. Q., K. Kawamura, C. M. Pavuluri, T. Swaminathan, and J. Chen, 2010: Molecular characterization of urban organic aerosol in tropical India: Contributions of primary emissions and secondary photooxidation. Atmos. Chem. Phys., 10, 26632689, https://doi.org/10.5194/acp-10-2663-2010.

    • Search Google Scholar
    • Export Citation
  • Gargava, P., J. C. Chow, J. G. Watson, and D. H. Lowenthal, 2014: Speciated PM10 emission inventory for Delhi, India. Aerosol Air Qual. Res., 14, 15151526, https://doi.org/10.4209/aaqr.2013.02.0047.

    • Search Google Scholar
    • Export Citation
  • Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 54195454, https://doi.org/10.1175/JCLI-D-16-0758.1.

    • Search Google Scholar
    • Export Citation
  • Ghan, S. J., and S. E. Schwartz, 2007: Aerosol properties and processes: A path from field and laboratory measurements to global climate models. Bull. Amer. Meteor. Soc., 88, 10591084, https://doi.org/10.1175/BAMS-88-7-1059.

    • Search Google Scholar
    • Export Citation
  • Ghosh, G., 1999: Dispersion-equation coefficients for the refractive index and birefringence of calcite and quartz crystals. Opt. Commun., 163, 95102, https://doi.org/10.1016/S0030-4018(99)00091-7.

    • Search Google Scholar
    • Export Citation
  • Giles, D. M., and Coauthors, 2012: An analysis of AERONET aerosol absorption properties and classifications representative of aerosol source regions. J. Geophys. Res., 117, D17203, https://doi.org/10.1029/2012JD018127.

    • Search Google Scholar
    • Export Citation
  • Giles, D. M., and Coauthors, 2019: Advancements in the Aerosol Robotic Network (AERONET) version 3 database – Automated near-real-time quality control algorithm with improved cloud screening for Sun photometer aerosol optical depth (AOD) measurements. Atmos. Meas. Tech., 12, 169209, https://doi.org/10.5194/amt-12-169-2019.

    • Search Google Scholar
    • Export Citation
  • Ginoux, P., J. M. Prospero, T. E. Gill, N. C. Hsu, and M. Zhao, 2012: Global-scale attribution of anthropogenic and natural dust sources and their emission rates based on MODIS Deep Blue aerosol products. Rev. Geophys., 50, RG3005, https://doi.org/10.1029/2012RG000388.

    • Search Google Scholar
    • Export Citation
  • Gläser, G., H. Wernli, A. Kerkweg, and F. Teubler, 2015: The transatlantic dust transport from North Africa to the Americas—Its characteristics and source regions. J. Geophys. Res. Atmos., 120, 11 23111 252, https://doi.org/10.1002/2015JD023792.

    • Search Google Scholar
    • Export Citation
  • Gliß, J., and Coauthors, 2021: AeroCom phase III multi-model evaluation of the aerosol life cycle and optical properties using ground- and space-based remote sensing as well as surface in situ observations. Atmos. Chem. Phys., 21, 87128, https://doi.org/10.5194/acp-21-87-2021.

    • Search Google Scholar
    • Export Citation
  • Gosse, S. F., M. Wang, D. Labrie, and P. Chylek, 1997: Imaginary part of the refractive index of sulfates and nitrates in the 0.7–2.6-μm spectral region. Appl. Opt., 36, 36223634, https://doi.org/10.1364/AO.36.003622.

    • Search Google Scholar
    • Export Citation
  • Griffin, D., C. Kellogg, V. Garrison, and E. Shinn, 2002: The global transport of dust: An intercontinental river of dust, microorganisms and toxic chemicals flows through the Earth’s atmosphere. Amer. Sci., 90, 228235, https://doi.org/10.1511/2002.9.228.

    • Search Google Scholar
    • Export Citation
  • Haywood, J. M., S. R. Osborne, P. N. Francis, A. Keil, P. Formenti, M. O. Andreae, and P. H. Kaye, 2003: The mean physical and optical properties of regional haze dominated by biomass burning aerosol measured from the C-130 aircraft during SAFARI 2000. J. Geophys. Res., 108, 8473, https://doi.org/10.1029/2002JD002226.

    • Search Google Scholar
    • Export Citation
  • Hess, M., P. Koepke, and I. Schult, 1998: Optical properties of aerosols and clouds: The software package OPAC. Bull. Amer. Meteor. Soc., 79, 831844, https://doi.org/10.1175/1520-0477(1998)079<0831:OPOAAC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Holben, B. N., and Coauthors, 1998: AERONET—A federated instrument network and data archive for aerosol characterization. Remote Sens. Environ., 66, 116, https://doi.org/10.1016/S0034-4257(98)00031-5.

    • Search Google Scholar
    • Export Citation
  • Holben, B. N., and Coauthors, 2001: An emerging ground-based aerosol climatology: Aerosol optical depth from AERONET. J. Geophys. Res., 106, 12 06712 097, https://doi.org/10.1029/2001JD900014.

    • Search Google Scholar
    • Export Citation
  • Hu, Z., Q. Jin, Y. Ma, B. Pu, Z. Ji, Y. Wang, and W. Dong, 2021: Temporal evolution of aerosols and their extreme events in polluted Asian regions during Terra’s 20-year observations. Remote Sens. Environ., 263, 112541, https://doi.org/10.1016/j.rse.2021.112541.

    • Search Google Scholar
    • Export Citation
  • Huang, J., C. Zhang, and J. M. Prospero, 2010: African dust outbreaks: A satellite perspective of temporal and spatial variability over the tropical Atlantic Ocean. J. Geophys. Res., 115, D05202, https://doi.org/10.1029/2009JD012516.

    • Search Google Scholar
    • Export Citation
  • IPCC, 2021: Climate Change 2021: The Physical Science Basis.