Örjan Gustafsson,1*Martin Kruså,1Zdenek Zencak,1Rebecca J. Sheesley,1Lennart Granat,2Erik Engström,2P. S. Praveen,3P. S. P. Rao,4Caroline Leck,2Henning Rodhe2
Carbonaceous aerosols cause strong atmospheric heating and largesurface cooling that is as important to South Asian climateforcing as greenhouse gases, yet the aerosol sources are poorlyunderstood. Emission inventory models suggest that biofuel burningaccounts for 50 to 90% of emissions, whereas the elemental compositionof ambient aerosols points to fossil fuel combustion. We usedradiocarbon measurements of winter monsoon aerosols from westernIndia and the Indian Ocean to determine that biomass combustionproduced two-thirds of the bulk carbonaceous aerosols, as wellas one-half and two-thirds of two black carbon subfractions,respectively. These constraints show that both biomass combustion(such as residential cooking and agricultural burning) and fossilfuel combustion should be targeted to mitigate climate effectsand improve air quality.
1 Department of Applied Environmental Science, Stockholm University, 10691 Stockholm, Sweden. 2 Department of Meteorology, Stockholm University, 10691 Stockholm, Sweden. 3 Maldives Climate Observatory at Hanimaadhoo (MCOH), Republic of the Maldives. 4 Indian Institute of Tropical Meteorology, Dr. Homi Bhabha Road, Pashan, Pune 411 008, India.
The radiative effects of carbonaceous aerosols constitute oneof the largest uncertainties in climate modeling (16).Combustion-derived carbonaceous aerosols have traditionallybeen associated with pollution in urban areas, but researchover the past decade has revealed that the haze they cause mayenvelop entire subcontinents and ocean basins (3,79).The extensive atmospheric brown cloud (ABC) over South Asiaand the Indian Ocean persists during the winter season, andits cooling effect may regionally balance and even surpass thewarming effect of greenhouse gases (GHGs) (3,4,8), with predictedeffects including changed circulation and monsoon patterns withamplified droughts and floods (10,11), as well as increasedmelting of Himalayan glaciers (4). A conspicuous feature ofthe Asian ABC is its unusually high content of airborne blackcarbon (BC) particles (4,7,12,13). This highly condensedcarbonaceous residue of incomplete combustion is the “dark horse”in the current climate debate as substantial uncertainties existabout its atmospheric longevity (1), aerosol mixing state (14),measurement (1518), and sources (12,1925). Becauseit is becoming clear that BC represents a continuum of lightabsorbing carbon (LAC) forms (1618), exploiting differencesin BC analytical techniques may improve the characterizationof atmospheric BC. The primary motivation for the ABC BC Radiocarbon(14C) Campaign (ABC-BC14), the results of which are reportedhere, was the observational determination of the relative contributionof contemporary biomass versus fossil fuel combustion to boththe total carbonaceous aerosols and to two different atmosphericBC isolates.
Though it is now established that there is an unusually highmass fraction of BC in the Asian ABC (4,7,12,13,2325),there is a notable discrepancy in source apportionment of thisBC between top-down studies relying on measured ratios of BCto total carbon or other aerosol components (12,24,25) ascompared with bottom-up emission inventories based on fuel consumptionand laboratory-derived emission factors (1923) (Table 1).Several top-down studies conclude that 50 to 90% of the SouthAsian BC originates from fossil fuel combustion (12,24,25).However, employed end-member ratios were from other regionsand may not be representative of South Asian combustion processes(21,23). Further, the BC:organic carbon (OC) ratio is nonconservativeif there is substantial formation of secondary organic aerosols(13,26). In contrast, bottom-up approaches suggest that only10 to 30% of the BC loadings originate from fossil fuel combustion(1923) while recognizing that emission factors (particularlythose for biomass combustion) are difficult to constrain becauseof strong dependency on fuel type and efficiency of combustion(22,23). This current dichotomy is addressed in the ABC-BC14Campaign by using radiocarbon abundance (half life1/2 = 5730years) as a tool to distinguish between fossil (radiocarbon”dead”) and contemporary biomass (radiocarbon “alive”) combustionsources of the Asian ABC.
Table 1. Apportionment of Indian carbonaceous aerosols between fossil fuel and biomass combustion. The characteristics of different carbonaceous particle fractions (TOC, BC, EC, SC) are discussed in the text and SOM.StudyBC from fossil fuel combustionBC from biomass combustionMethods/comments
Emission inventories India inventory (19, 20) 29% 71% 0.1 Tg/year fossil fuel and 0.25 Tg/year biomass South Asia inventory (21) 12-45% 55-88% 0.059-0.37 Tg/year fossil fuel and 0.45 Tg/year biomass for India Global inventory (22) 30% 70% 0.18 Tg/year fossil fuel, 0.33 Tg/year biofuel, 0.087 Tg/year open burning for India South Asia biofuel study (23) 25% 75% India-specific emission factors and fuel usage Ambient measurements INDOEX* flights over the tropical Indian Ocean (24) 80% 20% EC:TC ratio for three flights INDOEX flights over the tropical Indian Ocean (12) 60-90% 10-40% EC:TC ratio for 13 flights ABC monitoring in the Maldives (25) 40-50% 30-40% Positive matrix factorization with EC and multiple elements Maldives + India (this study) 32 ± 5%68 ± 6% Radiocarbon analysis of ambient filter-based SC, range of 66-76% for biomass Maldives + India (this study) 54 ± 8% 46 ± 8% Radiocarbon analysis of ambient filter-based EC, range of 45-52% for biomass
* INDOEX is the Indian Ocean Experiment. Standard deviation of 9 samples. Ranges calculated from sensitivity analysis detailed in the SOM text.
The ABC-BC14 Campaign was conducted with identical samplingat two sites (Fig. 1) of the international ABC-Asia project.Aerosol samples for microscale 14C measurements were collectedat the Maldives Climate Observatory at Hanimaadhoo island (MCOH)(6.78°N, 73.18°E) from 31 January to 16 March 2006 andat the mountain top site of the Indian Institute of TropicalMeteorology located at Sinhagad, West India (SIN) (18.35°N,73.75°E, 1400 meters above sea level) from 27 March to 18April 2006 (27). The ABC-BC14 Campaign thus overlapped withthe previously reported Maldives Autonomous Unmanned AerialVehicle Campaign (4) that reported on the vertically resolvedaerosol solar heating, and the meteorological context is detailedtherein.
Fig. 1. Regional distribution of aerosol optical depth at 550 nm derived from a moderate resolution imaging spectro-radiometer (MODIS) instrument aboard the Terra satellite (average for March 2006). The black arrows denote dominant air mass transport patterns in the region during the winter monsoon. The two aerosol sampling sites are shown. [View Larger Version of this Image (101K GIF file)] Back-trajectory analyses illustrate the typical winter monsooncirculation, with most of the first half of the MCOH samples(31 January to 18 February 2006) reflecting a predominant low-levelair mass transport during the preceding 10 days from centralIndia (including the Gangetic Plain) flowing southward alongthe western Bay of Bengal 2 to 5 days before arrival (fig. S1A).During subsequent collections (19 February to 16 March 2006),most 10-day trajectories originated from the northern ArabianSea and adjacent land areas in northwest India and Pakistanwith transport along the Indian west coast margin (fig. S1B).Most of the surface air masses sampled at SIN were arrivingfrom a sector west and north, originating from Arabian Sea,Arabic peninsula, Pakistan, and northwest India (fig. S1C).Satellite-retrieved optical signals suggest that study locationswere influenced by aerosols, presumably brown clouds (Fig. 1).
Ground-based particle soot absorption photometer (PSAP) (550nm) measurements confirm high abundances of LAC at MCOH andSIN (Fig. 2A). Absorption coefficients of 1·105 m1 at the onset of the MCOH Campaign increased to a maximumabove 3·105 m1 toward the end of the GangeticPlain influenced period, followed by a consistent decrease to5·106 m1 as trajectories shifted towardthe western side of India. The lower-temporal-resolution PSAPdata for the SIN campaign varied from 0.3 to 3·105 m1.
Fig. 2. Concentration and radiocarbon-based source apportionment of carbonaceous aerosols over the Maldives and western India. The ABC-BC14 Campaign is divided into two periods with sampling at Hanimaadhoo (Maldives) 31 January to 16 March 2006 and at Sinhagad (India) 27 March to 18 April 2006. (A) Optical measurement (PSAP) of LAC. The data symbols are numbered 1 to 4, which corresponds to the four different source trajectory classes listed below the figure. The dashed symbols at the bottom the panel represent days with no data. (B) Carbon-mass based concentrations of total organic carbonaceous aerosols (TOC) (open circles) and EC (black squares). (C) Radiocarbon content of TOC, EC, and SC (gray inverted triangles). The radiocarbon end-member ranges are shown for both contemporary biomass/biofuel (upper green field) and for fossil fuel (lower black line) and are further detailed in the text. The predominant air mass source regions over the past 10 days are summarized at the bottom. Horizontal range bars in (B) and (C) represent the collection and integration period for each sample. [View Larger Version of this Image (19K GIF file)] The temporal evolutions of the mass-based carbonaceous aerosolconcentrations were broadly consistent with the PSAP data (Fig. 2B).The highest MCOH total organic carbon (TOC) values were associatedwith air from northern India (4 to 5 µg·m3)decreasing to 1.4 µg·m3 with Arabian Seaorigin. Similarly, the elemental carbon (EC), measured usinga thermo-optical technique (17,18,27,28), varied from 1.2to 0.2 µg·m3 (Fig. 2B) for MCOH. The sootcarbon (SC) fraction, measured by chemothermal-oxidation (18,2729) and representing a more recalcitrant portion ofthe BC spectrum (15,16,18), was lower but followed a similartemporal trend (table S1). The TOC, EC, and SC were all closelycoupled (r2 = 0.84 for EC versus SC) (fig. S2), indicating astrong contribution of combustion processes to the total carbonaceousaerosols.
Each of the three carbon isolates exhibited a marked temporaluniformity in radiocarbon signal and hence between the contributionsfrom fossil and contemporary biomass sources. The measured14Ccontent of TOC ranged from 239 ± 3 to 145± 2 per mil () in MCOH samples and from 235 ±2 to 187 ± 2 in SIN samples (Fig. 2C and tableS1). This consistency attests to the ability of the series ofweeklong samples to capture the broader-scale source contributions.Because the optical techniques used to quantify LAC-BC do notphysically isolate a carbon mass fraction, a prerequisite for 14C measurement, two techniques commonly used to quantify thecombustion-derived carbon mass were employed to isolate carbon(27).
The EC isolate was more fossil-rich than the TOC and rangedfrom 379 ± 4 to 319 ± 3 in Indiaand from 595 ± 12 to 430 ± 5 overthe Indian Ocean (Fig. 2C). The more recalcitrant SC fractionhad more modern14C values, indistinguishable between Indianand Maldivian sites, with averages of 227 ± 37versus 167 ± 70, respectively. Hence, there isa component included in the EC isolate but excluded from SCthatis, less recalcitrant but more 14C depleted. We hypothesizethat this is fossil “brown carbon” (17) from either domesticcoal combustion or fine coal dust released from pulverizationof coal for the many coal-fired power plants in India (21,22).Coal has been one key replacement of wood as domestic fuel (22),and it is conceivable that the BC produced by such small-scaleand inefficient coal burning is escaping detection as SC (16,18) but is included in EC (24). Further, uncombusted fine coaldust yields larger false positives for EC than SC (18), consistentwith differences in14C during the ABC-BC14 study. Sensitivitymodel calculations explored inclusion of up to 30% of the instrument-inherentpyrolyzed OC in the isolated EC and found that the potentialeffect would be within the uncertainty of the reported isotopevalues [table “” not found /]
To afford a detailed comparison with earlier bottom-up and top-downsource apportionment estimates of carbonaceous aerosols, a simpleisotopic mass balance equation (28,30), based on the14C data,was applied to apportion between the fractional contributionsof biomass (fbiomass) and fossil fuel (ffossil = 1 fbiomass)combustion sources to the carbonaceous aerosol components inthe investigated samples, as illustrated for EC
(1)where14CECis the measured radiocarbon content of the EC component and14Cfossil is 1000. The 14Cbiomass end member is between+70 and +225. The first14C value corresponds to contemporaryCO2 (31), and thus freshly produced biomass, whereas the second14C end member is for combustion particles emanating from thecombustion of wood (28,32,33). The latter14C value is higherbecause it is reflecting the14C of biomass that has accumulatedover the decadestocentury-long life span of trees,which includes the period after the atmospheric nuclear bombtesting that nearly doubled the14C value of CO2 by the early1960s. For India, there are several important contemporary biofueltypes, including wood fuel and cow dung; additionally, cropresidue burning is believed to be an important source of atmosphericBC. To regionally parameterize the contemporary14Cbiomass endmember, the relative contributions of fuel wood (83%) and dung+ crop waste (17%) provided by C. Venkataraman et al. were employed(23). Hence, an India-tailored14Cbiomass end member of +199 was used in the model calculations. The sensitivity of thesource apportionment results toward this end-member selectionis low (Table 1, table S3, and SOM text).
Application of this isotopic mass balance model to these Asianaerosol14C values revealed that bulk carbonaceous aerosols(TOC) were 67 ± 3% (1 SD) of contemporary origin. TheEC and SC isolates of the BC continuum were 46 ± 8 and68 ± 6% from biomass combustion (Table 1). Although thereare not yet any other reports of14C for EC isolates, the biomasscombustion fraction of SC was 63% for northwest African dustintercepted in the northeast Atlantic Ocean (34), 70 to 88%in wintertime Scandinavia (28), and an averaged 35% in a Swissalpine valley in the winter (33).
Our 14C-based constraint thus indicates a much larger role forbiomass and biofuel burning, compared with earlier top-downstudies, while attenuating the biofuel influence relative tobottom-up suggestions. In contrast to the two earlier approaches,the ABC-BC14 results also provide a much tighter source constraint.Dual isotopic probing, combining14C with13C (fig. S3), furtherunderscores biomass combustion. The13C suggests that wood fueland other C3 plants are complemented by C4 sources (such asfrom agricultural slash-and-burn practices) as substantial contributors.
This work demonstrates that both fossil and biomass combustionprocesses can be blamed for the extensive ABC over South Asia.Improved constraint on the sources is the first step towardenacting effective abatement strategies. The much shorter atmosphericlongevity for BC aerosols (approximately days to weeks) comparedwith GHGs raises the hope of a rapid response of the climatesystem. However, a consequence of thus decreasing the ABC “globaldimmer” would also be to remove its counterbalancing effecton anthropogenic GHGs (1).
References and Notes
1. M. O. Andreae, C. D. Jones, P. M. Cox, Nature435, 1187 (2005).
2. P. Forster et al., in Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, S. Solomonet al., Eds. (Cambridge Univ. Press, Cambridge, 2007).
3. V. Ramanathan, G. Carmichael, Nat. Geosci.1, 221 (2008). [CrossRef]
4. V. Ramanathan et al., Nature448, 575 (2007).
5. J. Seinfeld, Nat. Geosci.1, 15 (2008).
6. S. E. Schwartz, R. J. Charlson, H. Rodhe, Nat. Reports Clim. Change2, 23 (2007).
12. O. L. Mayol-Bracero et al., J. Geophys. Res. Atmos.107, 8030 (2002).
13. R. Rengarajan, M. M. Sarin, A. K. Sudheer, J. Geophys. Res. Atmos.112, D21307 (2007).
14. M. Z. Jacobson, Nature409, 695 (2001).
15. L. A. Currie et al., J. Res. Natl. Inst. Stand. Technol.107, 279 (2002).
16. M. Elmquist, G. Cornelissen, Z. Kukulska, Ö. Gustafsson, Global Biogeochem. Cycles20, GB2009 (2006).
17. M. O. Andreae, A. Gelencser, Atmos. Chem. Phys.6, 3131 (2006).
18. K. Hammes et al., Global Biogeochem. Cycles21, GB3016 (2007). [CrossRef]
19. M. S. Reddy, C. Venkataraman, Atmos. Environ.36, 677 (2002).
20. M. S. Reddy, C. Venkataraman, Atmos. Environ.36, 699 (2002).
21. R. R. Dickerson et al., J. Geophys. Res. Atmos.107, 8017 (2002).
22. T. C. Bond et al., J. Geophys. Res. Atmos.109, D14203 (2004).
23. C. Venkataraman, G. Habib, A. Eiguren-Fernandez, A. H. Miguel, S. K. Friedlander, Science307, 1454 (2005). [Medline]
24. T. Novakov et al., Geophys. Res. Lett.27, 4061 (2000).
25. E. A. Stone et al., J. Geophys. Res. Atmos.112, D22S23 (2007).
26. C. Neusüß, T. Gnauk, A. Plewka, H. Herrmann, P. K. Quinn, J. Geophys. Res. Atmos.107, 8031 (2002).
27. Aerosols were collected on pre-combusted microquartz filters using high-volume samplers. Mass concentrations of the total organic carbonaceous aerosol (TOC) and the thermo-optical transmission National Institute of Occupational Safety and Health (NIOSH) 5040 protocol EC and chemothermal-oxidation at 375°C SC subfractions of BC were quantified and isolated for offline carbon isotope analysis. Materials and methods details are available as supporting material on Science Online.
28. Z. Zencak, M. Elmquist, Ö. Gustafsson, Atmos. Environ.41, 7895 (2007).
29. Ö. Gustafsson et al., Global Biogeochem. Cycles15, 881 (2001).
30. M. Mandalakis et al., Environ. Sci. Technol.39, 2976 (2005). [Medline]
31. I. Levin, B. Kromer, M. Schmidt, H. Sartorius, Geophys. Res. Lett.30, 2194 (2003).
32. D. B. Klinedinst, L. A. Currie, Environ. Sci. Technol.33, 4146 (1999).
33. S. Szidat et al., Geophys. Res. Lett.34, L05820 (2007).
34. T. I. Eglinton et al., Geochem. Geophys. Geosyst.3, 1050 (2002).
35. This is a contribution of the Stockholm University Bert Bolin Centre for Climate Research. We gratefully acknowledge access to and support of the MCOH field site by the Maldives Meteorological Office and H. Nguyen of the international ABC program. We appreciate access to the thermo-optical analyzer at the Department of Applied Environmental Science, Stockholm University. We thank V. Ramanathan (University of California, San Diego, USA) and M. M. Sarin (Physical Research Laboratory, India) for useful discussions. This study was financed by the Swedish Strategic Environmental Research Foundation; the Swedish Research Council; the Swedish Research Council for Environment, Agricultural Sciences, and Spatial Planning; and the Swedish International Development Agency, Department for Research Co-Operation. Ö.G. also acknowledges support as an Academy Researcher from the Swedish Royal Academy of Science.