Volume 51, Issue 18 e2024GL110089
Research Letter
Open Access

Spatial Distribution in Surface Aerosol Light Absorption Across India

Taveen S. Kapoor

Taveen S. Kapoor

Department of Energy, Environmental and Chemical Engineering, Center for Aerosol Science and Engineering, Washington University in St. Louis, St. Louis, MO, USA

Centre for Climate Studies, Indian Institute of Technology Bombay, Mumbai, India

Contribution: Conceptualization, Formal analysis, Data curation, Writing - original draft, Visualization, Methodology

Search for more papers by this author
Chimurkar Navinya

Chimurkar Navinya

Centre for Climate Studies, Indian Institute of Technology Bombay, Mumbai, India

Contribution: Formal analysis, Data curation, Methodology, Writing - review & editing, Visualization

Search for more papers by this author
Adishree Apte

Adishree Apte

Department of Energy, Environmental and Chemical Engineering, Center for Aerosol Science and Engineering, Washington University in St. Louis, St. Louis, MO, USA

Department of Chemical Engineering, Indian Institute of Technology Bombay, Mumbai, India

Contribution: Formal analysis, Data curation, Writing - review & editing

Search for more papers by this author
Nishit J. Shetty

Nishit J. Shetty

Department of Energy, Environmental and Chemical Engineering, Center for Aerosol Science and Engineering, Washington University in St. Louis, St. Louis, MO, USA

Contribution: Formal analysis, Data curation, Writing - review & editing

Search for more papers by this author
Pradnya Lokhande

Pradnya Lokhande

Centre for Climate Studies, Indian Institute of Technology Bombay, Mumbai, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Sujit Singh

Sujit Singh

Department of Environmental Engineering, SJCE, JSS Science and Technology University, Mysuru, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Sadashiva Murthy B. M.

Sadashiva Murthy B. M.

Department of Environmental Engineering, SJCE, JSS Science and Technology University, Mysuru, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Meena Deswal

Meena Deswal

Department of Environmental Sciences, Maharshi Dayanand University Rohtak, Rohtak, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Jitender S. Laura

Jitender S. Laura

Department of Environmental Sciences, Maharshi Dayanand University Rohtak, Rohtak, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Akila Muthalagu

Akila Muthalagu

Department of Civil Engineering, Indian Institute of Technology Hyderabad, Kandi, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Asif Qureshi

Asif Qureshi

Department of Civil Engineering, Indian Institute of Technology Hyderabad, Kandi, India

Department of Climate Change, Indian Institute of Technology Hyderabad, Kandi, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Ankur Bhardwaj

Ankur Bhardwaj

Department of Earth and Environmental Sciences, Indian Institute of Science Education and Research Bhopal, Bhopal, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Ramya Sunder Raman

Ramya Sunder Raman

Department of Earth and Environmental Sciences, Indian Institute of Science Education and Research Bhopal, Bhopal, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Yang Lian

Yang Lian

Indian Institute of Tropical Meteorology, Ministry of Earth Sciences, Pune, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
G. Pandithurai

G. Pandithurai

Indian Institute of Tropical Meteorology, Ministry of Earth Sciences, Pune, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Pooja Chaudhary

Pooja Chaudhary

Department of Earth and Environmental Sciences, Indian Institute of Science Education and Research Mohali, Mohali, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Baerbel Sinha

Baerbel Sinha

Department of Earth and Environmental Sciences, Indian Institute of Science Education and Research Mohali, Mohali, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Shahadev Rabha

Shahadev Rabha

Coal & Energy Division, CSIR North-East Institute of Science & Technology, Jorhat, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Binoy K. Saikia

Binoy K. Saikia

Coal & Energy Division, CSIR North-East Institute of Science & Technology, Jorhat, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Tanveer Ahmad Najar

Tanveer Ahmad Najar

Department of Environmental Science, School of Earth and Environmental Science, University of Kashmir, Srinagar, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Arshid Jehangir

Arshid Jehangir

Department of Environmental Science, School of Earth and Environmental Science, University of Kashmir, Srinagar, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Sauryadeep Mukherjee

Sauryadeep Mukherjee

Department of Chemical Sciences, Bose Institute, Kolkata, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Abhijit Chatterjee

Abhijit Chatterjee

Department of Chemical Sciences, Bose Institute, Kolkata, India

Contribution: ​Investigation, Writing - review & editing

Search for more papers by this author
Harish C. Phuleria

Harish C. Phuleria

Centre for Climate Studies, Indian Institute of Technology Bombay, Mumbai, India

Environmental Science and Engineering Department, Indian Institute of Technology Bombay, Mumbai, India

Contribution: ​Investigation, Writing - review & editing, Supervision, Project administration

Search for more papers by this author
Rajan K. Chakrabarty

Corresponding Author

Rajan K. Chakrabarty

Department of Energy, Environmental and Chemical Engineering, Center for Aerosol Science and Engineering, Washington University in St. Louis, St. Louis, MO, USA

Correspondence to:

R. K. Chakrabarty and C. Venkataraman,

[email protected];

[email protected]

Contribution: Conceptualization, Methodology, Writing - review & editing, Supervision, Project administration

Search for more papers by this author
Chandra Venkataraman

Corresponding Author

Chandra Venkataraman

Centre for Climate Studies, Indian Institute of Technology Bombay, Mumbai, India

Department of Chemical Engineering, Indian Institute of Technology Bombay, Mumbai, India

Correspondence to:

R. K. Chakrabarty and C. Venkataraman,

[email protected];

[email protected]

Contribution: Conceptualization, Methodology, Writing - review & editing, Supervision, Project administration

Search for more papers by this author
First published: 19 September 2024
Citations: 1

Abstract

Light-absorbing carbonaceous aerosols that dominate atmospheric aerosol warming over India remain poorly characterized. Here, we delve into UV-visible-IR spectral aerosol absorption properties at nine PAN-India COALESCE network sites (Venkataraman et al., 2020, https://doi.org/10.1175/bams-d-19-0030.1). Absorption properties were estimated from aerosol-laden polytetrafluoroethylene filters using a well-constrained technique incorporating filter-to-particle correction factors. The measurements revealed spatiotemporal heterogeneity in spectral intrinsic and extrinsic absorption properties. Absorption analysis at near-UV wavelengths from carbonaceous aerosols at these regional sites revealed large near-ultraviolet brown carbon absorption contributions from 21% to 68%—emphasizing the need to include these particles in climate models. Further, satellite-retrieved column-integrated absorption was dominated by surface absorption, which opens possibilities of using satellite measurements to model surface-layer optical properties (limited to specific sites) at a higher spatial resolution. Both the satellite-modeled and direct in-situ absorption measurements can aid in validating and constraining climate modeling efforts that suffer from absorption underestimations and high uncertainties in radiative forcing estimates.

Key Points

  • Measurements at nine regional PAN-India sites reveal several regions with large aerosol absorption strength

  • Brown carbon contributes significantly (21%–68%) to near-ultraviolet absorption, indicating its importance in shortwave light absorption

  • Strong correlations observed between satellite data and surface absorption indicate future potential in modeling surface absorption

Plain Language Summary

Particulate pollution in the atmosphere scatter and absorb incoming solar energy, thus cooling or warming Earth's atmosphere. In developing countries and especially in India, one of the most polluted regions of the world, the extent to which particles can absorb solar energy and warm the atmosphere is not well understood. Here, for the first time, we measure particle absorption simultaneously at nine ground sites across India, in diverse geographical regions with different levels and types of particulate pollution. We find that organic carbon particles exert large absorption at near-ultraviolet wavelengths, which contain significant solar energy. These light absorbing organic carbon particles, called brown carbon, are emitted in large quantities from biomass burning (e.g., burning crop residue and cooking on wood-fired stoves). Comparing ground measurements of absorption with satellite-retrieved measurements that are representative of the entire atmospheric column, we find that near-surface atmospheric particles can exert significant warming. This study highlights the need to improve climate model simulations of particulate pollution's impact on the climate by incorporating spatiotemporal surface-level absorption measurements, including absorption by brown carbon particles.

1 Introduction

Climate models assign large relative uncertainties to radiative forcing due to light absorbing carbonaceous aerosol (Gliß et al., 2021; Szopa et al., 2022) because of our limited understanding of their abundance and particle optical properties (Sand et al., 2021). One reason may be the underestimation of absorption across several world regions (Cherian & Quaas, 2020; Gliß et al., 2021; Tegen et al., 2019), especially in the Indian subcontinent (Cherian & Quaas, 2020; Kapoor, Phuleria, et al., 2023; Sarkar et al., 2022), where concentrations of these species are large. The Indian region has a variety of carbonaceous aerosol sources in unorganized economic sectors, that is, those without emission guidelines (Pandey et al., 2014; Sadavarte & Venkataraman, 2014), leading to spatiotemporal heterogeneities of these species (Li et al., 2022). Moreover, measurements of aerosol absorption are scarce, leading to the poor spatial representation of ground data used in climate model validation (Samset et al., 2018; J. Wang et al., 2018). The COALESCE monitoring network (Venkataraman et al., 2020)—consisting of 11 regional background locations at diverse airsheds—was set up specifically to better understand the amounts and consequences of carbonaceous aerosol (black and brown carbon) in the region.

Black carbon (BC) particles absorb strongly in the near-UV-visible-infrared region of the electromagnetic spectrum, whereas organic carbon and dust particles absorb preferentially in the near-UV region. For all these particles, the chemical composition, physical dimensions, and, consequently, the absorption properties are dependent on the sources and formation pathways. For example, the large amounts of organic carbon particles emitted from biomass combustion have different optical properties, depending on the combustion conditions and volatilities of the chromophores (McClure et al., 2020; Saleh, 2020; Saleh et al., 2018). The dominant sources of organic carbon in India are biomass combustion from residential cooking, agricultural residue burning, and informal (or legally unregulated) industries (Pandey et al., 2014). Meanwhile, BC emissions arise primarily from biomass cooking in residential cookstoves, secondary lighting using kerosene lamps, informal industries, and diesel transportation (Lam et al., 2012; Pandey et al., 2014; Sadavarte & Venkataraman, 2014).

Several studies have made filter-based aethalometer measurements of ambient aerosol absorption at single locations in India (Ganguly et al., 2006; Hyvärinen et al., 2009; Jose et al., 2016; Niranjan et al., 2011; Pathak & Bhuyan, 2014; Rajesh & Ramachandran, 2019; Ram et al., 2012; Soni et al., 2010; Tiwari et al., 2015) and others have made measurements at multiple locations (Ganguly et al., 2005; Gogoi et al., 2017). The widely used aethalometer is easy to handle and deploy, and it has a good temporal resolution, but the instrument suffers from artifacts despite the development of correction algorithms (Bond et al., 1999; Virkkula et al., 2007; Weingartner et al., 2003) that partially improve accuracy (Kumar et al., 2022). Methods to measure absorption with greater accuracy include photoacoustic spectroscopy and cavity ring-down spectroscopy, but these methods are not often used in long-term ambient measurements. Pandey et al. (2019) developed a method to calculate spectrally resolved particle phase absorption using optically thin polytetrafluoroethylene (PTFE) filter substrates in a UV-visible spectrophotometer with an integrating sphere attachment. This non-destructive method makes it possible to measure aerosol absorption using PTFE filters, which are also extensively used to measure particulate matter (PM2.5) mass and chemical composition by measurement networks (Snider et al., 2015; Venkataraman et al., 2020).

In the present study, we aim to improve the understanding of spectral aerosol absorption across the Indian region, with the ultimate goal of improving radiative forcing estimates from absorbing aerosol species. To understand its spatiotemporal distribution, aerosol absorption is analyzed through ambient measurements at nine regional locations across India. Spectral aerosol absorption measurements can segregate absorption by aerosol constituents and help infer the emission sources contributing to absorption. Lastly, we study the correlations between surface-level absorption and satellite-retrieved columnar aerosol absorption properties. Through these calculations we explore the possibility of using satellite measurements to model surface absorption at a higher resolution and demonstrate it through preliminary multivariate regression models.

2 Materials and Methods

2.1 Sampling Locations

Measurements were conducted during the summer months (March–May) of 2020 at nine regionally representative sites (Lekinwala et al., 2020) of the COALESCE network (Venkataraman et al., 2020) located in diverse airsheds in India. Additional measurements at three locations—Rohtak, Hyderabad, and Mysuru—were also made during the monsoon (June-September), post-monsoon (October-November) and winter (December, January, and February); these were conducted to understand the north-south variation during these seasons. Table S1 in Supporting Information S1 summarizes the measurements. Most of the summer season measurements were made before COVID-19 related lockdowns in the country, and the monsoon and winter measurements were made after the lockdowns were lifted (details in Section S1 in Supporting Information S1) (Tibrewal & Venkataraman, 2022). A few site measurements were also carried out during the lockdown period (Table S2 and Table S3 in Supporting Information S1). When comparing the pre- and during-lockdown measurements, we observe changes in optical properties. These changes in absorption may be influenced by meteorological changes (Figure S1 in Supporting Information S1), in addition to the shut-down of some emission sources (Tibrewal & Venkataraman, 2022). Therefore, the presented measurements may not accurately represent atmospheric aerosols during normal (non-lockdown) years; however, the conclusions made from our analysis still stand.

2.2 Measurement Method

Aerosol particles were collected on PTFE filter substrates using a speciation air sampling system (SASS, Met One Instruments Inc., United States) for approximately 24 hr per the COALESCE network protocol (Venkataraman et al., 2020). Aerosol absorption (babs,λ) was estimated by first calculating the optical depths (OD) of the filters using a UV–Vis spectrophotometer (PerkinElmer LAMBDA 35) with an integrating sphere detector (M. Zhong and Jang, 2011). The optical depth was converted to particle phase absorption by using the correction method developed by Pandey et al. (2019) (Equation 1). Absorption by particles was calculated as a function of the optical depth of the aerosol-laden filters and aerosol sampling parameters. They used a one-dimensional two-stream radiative transfer model to simulate the particle-phase absorption measured by integrating photo-acoustic nephelometers. Equation 1 estimates the absorption coefficient (babs in Mm−1), where As is the filter area in m2, Q is the volume flow rate in L min−1, ts is the sampling duration in min, and 109 is a multiplication factor to convert absorption to Mm−1. The equation was specifically developed for highly absorbing aerosol with single-scatter albedo (SSA) <0.9, characteristic of the Indian region (Table S4 in Supporting Information S1) (Ganguly et al., 2005, 2006; Hyvärinen et al., 2009; Jose et al., 2016; Nandan et al., 2021; Niranjan et al., 2011; Rajesh & Ramachandran, 2019; Pathak & Bhuyan, 2014; Soni et al., 2010; Srivastava et al., 2012; Tiwari et al., 2015). We report the total uncertainty in absorption estimation as 18%, calculated by propagating the uncertainties in the terms in Equation 1. The uncertainty in filter optical depth (0.48 × ODλ1.32) is reported as 10% (Pandey et al., 2019), and we calculate the flow rate (Q) uncertainty as 15%. Filter deposit area (As) and sampling time (ts) have negligible uncertainties.
b abs , λ = 0.48 OD λ 1.32 10 9 × A s Q × t s ${b}_{\mathrm{abs},\lambda }=\left[0.48{\left({\mathrm{OD}}_{\lambda }\right)}^{1.32}\right]\frac{{10}^{9}\times {A}_{s}}{Q\times {t}_{s}}$ (1)
The measured absorption coefficients were used to calculate the absorption Angstrom exponents (AAEλ12), a measure of the wavelength dependence of the absorption (Equation 2). Using a microbalance (Sartorius Cubis MSU6.6S), the PTFE filter substrates were also used to calculate the total aerosol PM2.5. This mass was combined with the spectral absorption coefficients to calculate the mass absorption-cross section (MAC, Equation 3), a measure of the strength of aerosol absorption. We estimate MAC for the total aerosol mass (Titos et al., 2012; Martins et al., 2009), while other studies estimate MAC for the mass of the absorbing constituent (Bond et al., 2013; Laskin et al., 2015; Petzold et al., 2013). The formulation used here, throughout the manuscript, represents the absorption strength per unit PM2.5 mass concentration, with its magnitude representing the relative abundance of absorbing and scattering components.
AAE λ 1 / λ 2 = log b abs , λ 1 b abs , λ 2 log λ 1 λ 2 ${\mathrm{AAE}}_{\lambda 1/\lambda 2}=\frac{-\log \left(\frac{{b}_{\mathrm{abs},\lambda 1}}{{b}_{\mathrm{abs},\lambda 2}}\right)}{\log \left(\frac{\lambda 1}{\lambda 2}\right)\hspace*{.5em}}$ (2)
MAC λ = b abs , λ PM 2.5 ${\mathrm{MAC}}_{\lambda }=\frac{{b}_{\mathrm{abs},\lambda }}{{\mathrm{PM}}_{2.5}}$ (3)

The collective aerosol absorption in the near-UV-visible spectrum (babs,410) has contributions from both black carbon (BC) and brown carbon (BrC). Dust aerosol absorption is not considered, given the reported evidence about the large AAE675/870:AAE440/675 ratios and evidence from HYSPLIT back-trajectories (Figure S2 in Supporting Information S1). Bahadur et al. (2012) showed that dust absorption is negligible when AAE675/870:AAE440/675 > 0.8, ratios that are observed in the present study (Supplementary Section S2 in Supporting Information S1). Absorption by brown carbon is generally estimated by subtracting BC absorption from the measured total absorption at near-UV wavelengths. BC absorption at near-IR wavelengths is assumed to equal the total absorption and is extrapolated to near-UV wavelengths using the AAE. The assumed absence of near-IR BrC absorption likely leads to an underestimation in BrC absorption estimates since some BrC compounds, called dark-BrC, have recently been shown to have near-IR absorption similar to BC (Chakrabarty et al., 2023; Corbin et al., 2019; Hoffer et al., 2016; Saleh et al., 2018). Nevertheless, the AAE extrapolation methods provide a useful estimate using commonly deployed instrumentation. Methods to determine BC AAE include assuming that AAE = 1, using two-component methods (Chen et al., 2015; Izhar et al., 2020), and assuming the near-IR AAE is the same for the whole wavelength range (G. Zhang et al., 2018). These methods, however, are inaccurate for atmospheric BC particles (Bahadur et al., 2012; Bond et al., 2006; Gyawali et al., 2009; X. Wang et al., 2016) and often predict negative BrC absorption (Kapoor et al., 2022; J. Wang et al., 2018). Therefore, to improve upon the previous methods, we employed a Mie theory-based optimization framework using time-varying, observational, and theoretical constraints, to estimate BC AAE, following Kapoor et al. (2022).

We consider a core-shell aerosol particle structure with coating factors varying from 1 to 3, where the coating factor indicates the particle's diameter to the BC core's diameter. Inputs to Mie theory were a BC refractive index of 1.95 + 0.79ί (T. C. Bond & Bergstrom, 2006), a non-absorbing coating material refractive index of 1.55 + 10−6ί (Kopke et al., 1997), and a particle size distribution spanning count median core diameters from 20 to 300 nm, with a geometric standard deviation (GSD) of 1.5. We used 410, 800, and 880 nm wavelengths for this study, where any absorption between 800 and 880 nm (and above) wavelengths was considered to result purely from BC. The choices of 410 and 880 nm were based on the UV-visible spectrophotometer's working range, and signals beyond these wavelengths exhibited high noise. The 800–880 nm range was selected to ensure minimal interference from BrC. Mie theory (Bohren & Huffman, 1998; Mätzler, 2002) was used to determine BC absorption efficiencies at 410, 800 and 880 nm for various core diameters and coating factors, and we used these results to prepare look-up-tables of BC AAE800/880 and AAE410/880. Then, for each measurement data point, we compared the observed AAE800/880 with the modeled AAE800/880 to select an optically representative size distribution and coating factor for BC, and used the corresponding modeled AAEBC,410/880 to estimate BC absorption at 410 nm (Equation 4). To estimate the uncertainty in BrC absorption, we calculated the combined standard uncertainty associated with variations in the input parameters of BC's imaginary refractive index (Supplementary section S3 in Supporting Information S1), the coating's real refractive index, and the GSD of the BC distribution (Kapoor et al., 2022). The BC GSD led to the largest uncertainty (∼11%, Figure S3 in Supporting Information S1), with a total combined standard uncertainty of 23% (Supplementary section S3 in Supporting Information S1).
b abs , BrC , 410 = b abs , 410 b abs , BC , 880 × 410 880 AAE Modelled , BC , 410 / 880 ${b}_{\mathrm{abs},\mathrm{BrC},410}={b}_{\mathrm{abs},410}-{b}_{\mathrm{abs},\mathrm{BC},880}\times {\left(\frac{410}{880}\right)}^{-{\mathrm{AAE}}_{\mathrm{Modelled},\mathrm{BC},410/880}}$ (4)

2.3 Columnar Optical Properties From Remote Sensing

In comparison to the optical properties measured in the present study, satellite measurements provide a more continuous spatial coverage. Here, we explore the relationships between the two data sets. First, we use the aerosol optical depth at 550 nm (AOD550) from the MODIS instrument aboard the Terra satellite (MYD08_D3_v6.1 product) (Platnick et al., 2015). The AOD from the Ozone Monitoring Instrument (OMI, OMAEROe_v003 product) aboard the Aura satellite is also taken, with other parameters including the absorption AOD (AAOD) and the ultraviolet aerosol index (UVAI, OMAERUVd_v003 product) (O. O. Torres, 2008). The satellite-retrieved data sets are quality checked level 3 products downloaded from NASA's Giovanni portal (https://giovanni.gsfc.nasa.gov/).

3 Results

3.1 Spatial Variation of Absorption During Summer Months

The aerosol absorption properties measured during the summer months (March-May) of 2020 at the nine sites are summarized in Figure 1. Near-IR absorption, considered dominated by black carbon, ranges from 54 to 124 Mm−1. The highest near-IR absorption is observed at Shyamnagar (124 Mm−1), followed by Rohtak (71 Mm−1) and Jorhat (69 Mm−1). Meanwhile, the near-UV absorption (range: 107–691 Mm−1), which may have additional contributions from brown carbon and dust particles, is also the highest at Shyamnagar (691 Mm−1), followed by Kashmir (447 Mm−1). The substantial increase in absorption with wavelength indicates larger AAEs at Kashmir (2.6) and Shyamnagar (2.3) when compared to those at the other sites (0.75–1.2). Larger AAEs point toward contributions from brown carbon aerosols, which are discussed in Section 3.2. The MAC550 (range: 1.5–3.8 m2g−1), a measure of absorption strength, is highest at Kashmir (3.8 m2g−1) and Bhopal (3.4 m2g−1) and lowest at Hyderabad (1.6 m2g−1) and Shyamnagar (1.5 m2g−1). The smaller MAC550 at Shyamnagar, despite large absorption coefficients (large absorption accompanied by large PM2.5), is indicative of aerosol sources co-emitting non-absorbing species (in addition to absorbing species) that contribute to PM but not absorption. This absorption behavior is consistent with previous studies that show the influence of fossil fuel combustion—a source of sulfur dioxide precursor producing non-absorbing sulfate—at Shyamnagar during the summer (Maheshwarkar et al., 2022; Mukherjee et al., 2023). Moreover, at Shyamnagar, the increased AAE coupled with the decreased MAC may also indicate weakly absorbing brown carbon particles. The aerosols at Hyderabad, Mysuru, and Mahabaleshwar also have contributions from industrial emissions (Maheshwarkar et al., 2022; Yadav et al., 2022) and exhibit lower MAC values.

Details are in the caption following the image

Measured absorption at 880 nm (babs,880, Mm−1), absorption at 410 nm (babs,410, Mm−1), absorption Ångström exponent (AAE410/880), and mass absorption cross-section (MAC550, m2g−1) at the nine sampling locations. The absorption coefficients are plotted on the left y-axis, and AAE and MAC on the right y-axis. Bars indicate the mean values for the season at every site, and error-bars indicate the standard deviation of the estimate. Site name abbreviations: KMR, Kashmir; MOH, Mohali; RTK, Rohtak; JOR, Jorhat; SHY, Shyamnagar; BPL, Bhopal; MBL, Mahabaleshwar; HYD, Hyderabad; and MYS, Mysuru.

3.2 Seasonal Distribution of Aerosol Absorption

As expected, the absorption coefficients were lower during the summer and monsoon than in the winter and post-monsoon months (Figure 2a). Aerosol concentrations decrease in the summer because of increased boundary layer heights (Figure S4 in Supporting Information S1) (GMAO, 2015) that dilute aerosol particles in the boundary layer (Stull, 1988). Wet scavenging by the increased rainfall in the monsoon months (Figure S5 in Supporting Information S1) (Pai et al., 2014) causes increased deposition of aerosol particles and decreases aerosol concentrations (Prospero et al., 1983). The measured AAE (range: 1.10–1.14) at Rohtak is invariant across the seasons (Figure 2b), but the MAC is slightly lower during winter (Figure 2c). A recent study at the Rohtak site reported warming aerosol during the winter (SSA ∼ 0.7, and a large estimated imaginary refractive index of 0.1) (Kapoor, Phuleria, et al., 2023), and an abundance of strongly absorbing black carbon and low-volatility organic carbon species. The present study finds large measured MAC not only in the winter, but during other seasons as well, when MAC values are higher than those in the winter. Therefore, we conclude that aerosol absorption at Rohtak throughout all the seasons is likely dominated by black carbon particles and strongly absorbing (or dark) brown carbon particles (Chakrabarty et al., 2023) that approach BC-like behavior, that is, with a large and spectrally invariant imaginary refractive index (Saleh et al., 2018). At Hyderabad, the winter AAEs are higher (∼2) than those during the other months (1–1.3), indicating the contribution of brown carbon particles, while at Mysuru, the AAEs are also close to 1. Larger values of AAE are observed during post-monsoon and winter (Figure 2b); these seasons have increased emissions from biomass burning sources like water heating (Navinya et al., 2023) and brick production (Tibrewal et al., 2023), which may lead to AAE enhancement. The MAC at Hyderabad during the monsoon and post-monsoon (3.6–3.8 m2g−1) is twice the same during the summer and winter (1.6–1.7 m2g−1) but is comparable to that observed at Mysuru (2.8–8.3 m2g−1). The absorption coefficients at Mysuru (south) are lower than in Rohtak (north) across all the seasons, which is likely driven by the larger wind speeds in peninsular India and the larger planetary boundary layer heights (Figure S4 in Supporting Information S1). Absorption at Hyderabad is lower than that at Rohtak during the monsoon months but comparable during other seasons; absorption over Hyderabad is similar to that over Mysuru during all but the winter season. From north to south (Rohtak-Hyderabad-Mysuru), there is minimal change in AAE and an increase in MAC. Minor changes in the AAE and larger changes in MAC indicate changes in the non-absorbing portion of the aerosol composition.

Details are in the caption following the image

Seasonal variation of (a) the absorption coefficient (b) the absorption Angstrom exponent (AAE), and (c) the mass absorption cross section (MAC) at three sites. Bars show mean values across the season, and error bars show standard deviations. Winter months are December to February, summer months are March to May, monsoon months are June to September, and post-monsoon months are October and November.

3.3 Brown Carbon Absorption

Brown carbon absorption and its contribution at a near-UV (410 nm) wavelength were estimated using a Mie-theory based optimization framework. Kashmir (68%) and Shyamnagar (67%) have the highest brown carbon contributions (Figure 3), as also evidenced by the AAE. Likely sources of BrC are biomass-based space heating in Kashmir (Navinya et al., 2023) and industrial combustion and crop residue burning in Shyamnagar (Kapoor, Navinya, et al., 2023; Maheshwarkar et al., 2022). BrC contribution at the other sites varies from 18% to 42% (albeit with varying magnitudes), consistent with previous observations in the Indian region (Table S5 in Supporting Information S1). Previous measurements include those by Bhardwaj et al. (2023), who used filter-based attenuation measurements in a thermal/optical carbon analyzer to estimate a BrC absorption contribution of 37%, which is larger than the values estimated in the present study (18%).

Details are in the caption following the image

Site and season specific brown carbon absorption contribution (% babs,BrC,410/babs,410) (mean ± standard deviation) with colors representing the averaged brown carbon absorption (babs,BrC,41k0) and symbols representing the sampling season. Winter months are December to February, summer months are March to May, monsoon months are June to September, and post-monsoon months are October and November.

The contribution of brown carbon particles to near-UV absorption is relatively higher at Hyderabad (∼36%), increasing to 64% during winter. In this region, residential fuel combustion and crop waste burning are common (Kapoor, Navinya, et al., 2023; Maheshwarkar et al., 2022). Previous studies have also shown enhanced BrC contributions at background locations during biomass burning events (Kapoor et al., 2022; Q. Wang et al., 2019). BrC contributions at Mysuru were slightly lower during the summer (29%) compared to the other months (32%–39%). Meanwhile, at Rohtak, the contributions were consistent throughout the seasons (27%–31%), indicating similar emission source mixes were active across seasons. We note that the measurements at the northern Indian sites during the post-monsoon months were not made during the active post-monsoon crop residue burning periods.

3.4 Relationships Between Ground Measured and Columnar Optical Properties

While the ground-measured extrinsic properties correlate with each other (Section S4, Figure S6 in Supporting Information S1), we also explored the relationships between ground measurements and satellite-measured columnar properties. We used monthly averaged ground-measured properties across all sites (Table S1 in Supporting Information S1) and seasons (at three sites, Table S2 in Supporting Information S1) and compared them with the satellite-retrieved monthly average optical properties to explore spatiotemporally varying associations (using the Spearman's rank correlation coefficient). Ground-measured PM2.5 and columnar AOD correlate strongly (Figure 4a), and this relationship has been explored in depth (Hammer et al., 2020; van Donkelaar et al., 2015, 2019, 2021). However, the AOD does not correlate significantly with ground measured spectral absorption coefficients (Figure S6 in Supporting Information S1). Instead, the UVAI has significant correlations with total babs,550 (r = 0.58, Figure 4b) and MAC410 (r = −0.68, Figure 4c). UVAI and AAOD also exhibit relationships with other ground-measured parameters such as PM2.5, absorption coefficients, and MAC (Figure S6 in Supporting Information S1). The UVAI and AAOD specifically target aerosol absorption in the column, but the AOD also includes scattering aerosols. The UVAI is also significantly correlated with the ground-measured MAC (Figure 4c).

Details are in the caption following the image

Relationships between surface-level measurements and column-integrated remote sensing measurements from instruments aboard satellites. Within each plot, Spearman's rank correlation coefficients (statistical significance at p < 0.05) are shown between (a) PM2.5 and aerosol optical depth (AOD, from MODIS), (b) babs,550 and ultraviolet aerosol index (UVAI, from OMI), and (c) MAC410 and UVAI (from OMI). Winter months are December to February, summer months are March to May, monsoon months are June to September, and post-monsoon months are October and November.

Studies on the Indo-Gangetic plains have shown the dominance of aerosol absorption in the boundary layer (Mishra et al., 2014; Srivastava et al., 2011). The significant correlations between ground-measured absorption and satellite-retrieved absorption properties in the present study, using data from all the seasons, suggest the importance of boundary layer aerosol absorption to columnar absorption throughout the Indian region. We further examine the aerosol vertical extinction profiles from the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) instrument (NASA/LARC/SD/ASDC, 2018) and find substantial aerosol extinction in the boundary layer (Figure S7 in Supporting Information S1). Together the good correlations and large aerosol extinction in the boundary layer suggest that columnar aerosol absorption is dominant in the boundary layer. The significant correlations between UVAI, which is sensitive to biomass burning aerosols (Habib et al., 2006; Hammer et al., 2018), and surface measured absorption highlight the regional influence of biomass burning combustion. The UVAI is a measure of the wavelength dependence of absorption and is hence sensitive to aerosol particles with relatively large AAEs, such as organic carbon, dust, and volcanic ash (Herman et al., 1997; Torres et al., 2013). It is also more sensitive to particles at higher altitudes and has been used to study elevated dust aerosol layers (∼4 km) during the summer/pre-monsoon months in India (Brooks et al., 2019; Mishra et al., 2014). We further investigated using the satellite-retrieved absorption properties (higher spatial resolution) to model ground measured optical properties (point data sets) using multi-variate regression models. The developed regression models resulted in promising equations for babs,410 (babs,410 = 133.4 + 471.5 × AODMODIS − 4,248.6 × AAODOMI; R2adj = 0.69), AAE410/880 (AAE410/880 = 1.3 + 0.9 × AODOMI − 0.36 × UVAIOMI; R2adj = 0.57), and near-UV BrC absorption contribution (%babs,410,BrC = 40 + 25 × AODOMI − 16 × UVAIOMI; R2adj = 0.52) (Table S6 in Supporting Information S1). These optimized multi-variate regression models that involve absorption at near-UV wavelengths (410 nm) include the UVAI (for AAEs and BrC absorption) and AAOD (for babs,410). The good performance of these simple models indicates the possibility of using satellite measured optical absorption properties to predict surface aerosol absorption properties, akin to using satellite measured AOD for surface PM2.5 concentrations (van Donkelaar et al., 2015). Future studies may target more robust models, using location-specific conversion factors and larger data sets to extract accurate and valuable information to validate and constrain climate modeling simulations.

4 Discussion

The ground measurements from the present study provide insights into the spatiotemporal variation of carbonaceous aerosol absorption in the Indian region. Such information is scarce in developing countries and useful for validating and constraining climate models that suffer from absorption underestimations. One reason for the aerosol absorption underestimation by climate models may be the absence of BrC absorption. This BrC absorption is shown to be significant even at regional background locations in the present study. Previous observations of ambient brown carbon absorption contributions report a range from 5% to 50% (summarized in Table S5 in Supporting Information S1) (Bhardwaj et al., 2023; Chen et al., 2019; Cho et al., 2019; de Sá et al., 2019; Devi et al., 2016; Kapoor et al., 2022, 2023b; Kaskaoutis et al., 2021; Kasthuriarachchi et al., 2020; Kim et al., 2021; Kirillova et al., 2016; Li et al., 2018, 2019; Liakakou et al., 2020; Pani et al., 2021; Park & Yu, 2018; Peng et al., 2020; Qin et al., 2018; Qiu et al., 2019; Rathod & Sahu, 2022; Shamjad et al., 2016; A. Soni & Gupta, 2022; X. Wang et al., 2016; J. Wang et al., 2018; Q. Wang et al., 2019; Xie et al., 2019; Yuan et al., 2016; G. Zhang et al., 2018; A. Zhang et al., 2020; Zhu et al., 2017, 2021), with the higher values generally associated with periods or locations influenced by biomass burning emissions (Kapoor et al., 2022; Kaskaoutis et al., 2021; Q. Wang et al., 2019; Y. Zhang et al., 2020). From the present measurements, we see daily average brown carbon absorption contributions at most regional locations ranging from 18% to 42%, which is comparable to previous reports in the region (Bhardwaj et al., 2023; Kapoor et al., 2022, 2023b; Rathod & Sahu, 2022; Shamjad et al., 2016; A. Soni and Gupta, 2022). We also observe contributions as large as 68% at some locations. From these observations, we find a significant contribution by BrC particles in the atmosphere that cannot be ignored when making direct radiative forcing estimates in climate modeling simulations. Measurement-based calculations have demonstrated BrC's non-trivial contribution to radiative forcing (Chakrabarty et al., 2016; Feng et al., 2013; Kirchstetter & Thatcher, 2012; Pandey et al., 2020; Shamjad et al., 2018; Zhu et al., 2021). However, most weather and climate modeling simulations consider organic carbon particles only as scattering and do not include their absorption (Sand et al., 2021). Despite their significant absorption contributions, only limited studies have attempted accurate representation of BrC particles in climate models (Brown et al., 2021; DeLessio et al., 2023; Jo et al., 2016; Neyestani & Saleh, 2022; Sand et al., 2021; X. Wang et al., 2018; A. Zhang et al., 2020). Their absence in modeling studies is due to reports of their photo-bleaching or absorption loss soon after being emitted (Sumlin et al., 2017; Wong et al., 2017; Zhong & Jang, 2014). However, recent measurements have shown that highly-absorbing, lower-volatility, BrC fractions exhibit longer atmospheric absorption lifetimes (Chakrabarty et al., 2023; Saleh, 2020), likely leading to significant atmospheric BrC absorption at the regional background sites in the present study. These BrC contributions to absorption need careful representation in climate models. Improved radiative forcing estimates require accurate spatial, temporal, and vertical distribution information of these absorbing aerosols–an important area for future research.

In the present study, we find that spatiotemporal variations in ground-measured spectral absorption properties are captured by the columnar AAOD and UVAI, which focus on aerosol absorption only, as opposed to AOD, which encompasses both absorption and scattering. These relationships indicate that aerosol absorption in the boundary layer dominates columnar absorption. Moreover, the strong correlation between ground-measured absorption (an extrinsic property) and UVAI, a columnar intensive property sensitive to biomass burning aerosols, indicates the importance of biomass combustion to aerosol absorption in the Indian region. Although the data set from the present study is limited, initial attempts reveal promising multivariate linear regression models to predict ground measured properties, including the AAEs and BrC absorption. Climate models would benefit from such information, akin to efforts to model BC and PM2.5 (Bao et al., 2020; van Donkelaar et al., 2015), helping reduce absorption underestimations in the region.

Acknowledgments

This work was supported by the Indian Ministry of Environment Forest and Climate Change under the NCAP-COALESCE project {Grant 14/10/2014-CC(Vol.II)}. The authors thank the internal review committee of the NCAP-COALESCE project for their comments and suggestions on this paper. The views expressed in this document are solely those of the authors and do not necessarily reflect those of the Ministry. The Ministry does not endorse any products or commercial services mentioned in this publication. TSK, RKC, AA, and NJS thank the WashU-IITB Joint Research and Education Initiative for supporting this study.

    Data Availability Statement

    The measured spectral absorption coefficients and calculated brown carbon absorption data set from the present study are available in Kapoor et al. (2024). Remote sensing data from satellites was downloaded from NASA's Giovanni portal (https://giovanni.gsfc.nasa.gov/giovanni/) and is available at Platnick et al. (2015), O. O. Torres (2008), and NASA/LARC/SD/ASDC (2018). Reanalysis data on boundary layer heights and wind speed were taken from GMAO (2015) and downloaded from the Giovanni portal. The rainfall data developed by Pai et al. (2014) was downloaded from https://imdpune.gov.in/cmpg/Griddata/Rainfall_25_NetCDF.html.