SEARCH

SEARCH BY CITATION

Keywords:

  • biomass burning;
  • West Africa;
  • emission ratio

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

[1] This paper investigates the physical and chemical characteristics of biomass burning aerosol over West Africa using data from the UK Facility for Airborne Atmospheric Measurements aircraft. Measurements of biomass burning aerosol were made during the Dust and Biomass-burning Experiment (DABEX) and Dust Outflow and Deposition to the Ocean (DODO) field experiments in January and February 2006. Layers of aged biomass burning aerosols were observed throughout the region, and fresh biomass burning aerosols were encountered during the penetration of smoke plumes at low altitudes. Vertical profiles of aerosol properties across the region are shown. Measurements from an Aerodyne Quadrupole Aerosol Mass Spectrometer (Q-AMS) show changes in chemical composition between fresh and aged biomass burning aerosols, over a region spanning thousands of kilometers. These data represent the first time that continental-scale variability in biomass burning aerosol composition has been observed. However, an almost linear relationship between organic aerosol mass concentration and carbon monoxide concentration was observed across the region. A net carbon loss occurs over the aerosol lifetime in the region owing to a combination of chemical processing and repartitioning of organic mass to the gas phase. Evolution of the number size distribution was observed, with coagulation concluded to be the dominant process involved, a finding supported by coagulation box modeling. Regional-scale emission ratios for organic (0.041) and black carbon (0.0072) with respect to CO have been estimated over West Africa, one of the largest sources of biomass burning globally. Biomass burning emissions from the West African Sahel are poorly represented in the literature, and these results represent important continental-scale emissions. They are in good agreement with literature values from other regions.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

[2] Biomass burning aerosols have an important effect on the Earth's radiation budget through scattering and absorption of solar radiation and usually act to cool the climate, though warming can also result, depending on the amount of absorption that occurs. Filter measurements indicate that the composition of biomass burning aerosols is dominated by carbonaceous particles, with lesser contributions from inorganic species such as K, Ca and Fe [Formenti et al., 2003; Reid et al., 1998]. The global focus of in situ measurements has been in South America, where there has been considerable research into the physical and optical properties and chemical composition of biomass burning aerosols [e.g., Andreae et al., 2004; Fuzzi et al., 2007; Reid et al., 1998]. Number size distributions for biomass burning aerosols are dominated by the accumulation mode and the count mean radius consistently increases as the aerosol ages [Dubovik et al., 2002]; a process which seems to be strongly affected by both coagulation and condensation [Reid et al., 2005, and references therein]. In southern Africa and in Brazil, the optical properties of biomass burning aerosols have been shown to evolve over time; the single scattering albedo increases downwind from source [Abel et al., 2003; Reid et al., 1998]. Chemical composition, in particular the black carbon to organic carbon (BC/OC) ratio is thought to be the major factor influencing the optical properties of aging biomass burning aerosols. A key issue in modeling biomass burning effects has been the extent to which measurements from the literature can be applied at a regional scale [Reid et al., 2005]; observations of the chemical and physical aging processes on a continental scale are presented here (i.e., over thousands of kilometers). To our knowledge, this is the first time such data have been presented.

[3] There have been few in situ measurements over West Africa, even though this region represents a globally significant source of biomass burning aerosols. The impact of biomass burning aerosols in this region is particularly uncertain because they can mix with the large quantities of mineral dust aerosols produced to the north which advect into the region. The Dust and Biomass Experiment (DABEX) went some way to addressing this paucity of measurements, forming part of the African Monsoon Multidisciplinary Analyses (AMMA) dry season Special Observation Period (SOP-0). The FAAM aircraft made 13 research flights from Niamey, Niger (13.3°N 2.6°E), between 13 January 2006 and 3 February 2006 during DABEX, and 6 research flights from Dakar, Senegal, between 4 and 17 February 2006 during the Dust Outflow and Deposition to the Ocean (DODO-1) project. Biomass burning aerosols originating from the band of intense agricultural fires in sub-Sahelian West Africa were observed in thick layers stretching across much of the continent at altitudes up to 4000 m above sea level (ASL). Surface winds blowing from the north of the region gave rise to dust storms at lower altitudes, which tracked south and westward and a degree of mixing was observed with the biomass burning aerosols. Both aerosol types were subject to westward advection across the continent and over the Atlantic [Haywood et al., 2008]. This study investigates the chemical and physical transformation of those biomass burning aerosols, and examines the contributions of primary (POA) and secondary organic aerosol (SOA) to the regional mass loading.

[4] A more in-depth overview of DABEX is given by Haywood et al. [2008].

2. Aircraft Instrumentation

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

[5] Data was collected on the UK Facility for Airborne Atmospheric Measurements (FAAM), a modified BAe-146 aircraft. The FAAM contains a range of instruments to measure aerosol number, size distribution, and composition, trace gas concentrations (carbon monoxide and ozone are considered in this study), cloud microphysics, solar and terrestrial radiative fluxes and standard meteorological variables.

[6] Aerosol size distributions were measured by a wing mounted Passive Cavity Aerosol Spectrometer Probe 100X (PCASP), an optical particle counter for measurement of aerosol number size distribution between radii of 0.05–1.5 μm. Further information regarding calibration of the instrument is provided in [Johnson et al., 2008]. Total aerosol number concentration was measured with a condensation nuclei optical particle counter (CN). This counts particles greater than 1.5 nm in radius by nucleation with supersaturated butane vapor [Kesten et al., 1991].

[7] An Aerodyne Research Inc. Quadrupole Aerosol Mass Spectrometer (Q-AMS) was used to provide near real time mass loadings and size-resolved chemical composition of the nonrefractory components of submicron aerosols. The instrument samples aerosol particles into vacuum through an aerodynamic lens, which focuses the particles at a heated vaporizer, where they volatilize. The gas plume is ionized using electron impact at 70 eV and the ion fragments are analyzed using a quadrupole mass spectrometer. This yields a mass spectrum of unit mass resolution where the individual peaks at a given mass to charge ratio (m/z) can be directly related to the fragment's mass. The Q-AMS has been described thoroughly in previous publications [Allan et al., 2003b; Jayne et al., 2000; Jimenez et al., 2003] and the particular aircraft installation of the Q-AMS on the FAAM is described in detail by Crosier et al. [2007]; submicron particle losses are considered negligible for the operating altitudes in this study [Zhang et al., 2002]. The thermal vaporizer is held at ∼600°C, so the Q-AMS is insensitive to refractory components such as black carbon and mineral dust, which do not vaporize at this temperature. The Q-AMS was calibrated for mass quantification (Ionisation Efficiency, IE) pre- and post-flight using the method of Jimenez et al. [2003].

[8] Previous studies [e.g., Canagaratna et al., 2007, and references therein] have shown that particle bounce off the heater is significant when particles are solid, reducing the collection efficiency (CE) below unity. The CE is often evaluated by comparison with other instruments such as PILS-IC or filter measurements and has been shown to be around 0.42 for sulphate aerosol [Drewnick et al., 2005]. Previous comparison of AMS data with filter and TEOM measurements indicated a CE of 0.7 for wood burning dominated aerosol [Alfarra et al., 2007]. However, this study was for fresh wood smoke from household fires in Switzerland. During DABEX and DODO, comparison with volume convolved PCASP measurements using a density of 1.35 g m−3 from previous biomass burning measurements [Reid and Hobbs, 1998] yielded a CE estimate of 0.34. The PCASP measures aerosol particles irrespective of composition whereas the Q-AMS measures only nonrefractory aerosol material and this will introduce an uncertainty in our CE estimate. The aerosol mass estimated by the Q-AMS is likely to be lower than the aerosol mass inferred from the PCASP, especially when a significant proportion of the aerosol is composed of black carbon or mineral dust. Johnson et al. [2008] estimate that approximately 12.5% of the fine aerosol mass was composed of black carbon, on the basis of absorption measurements with the PSAP. This therefore cannot explain the low CE used. Furthermore, typical BC/OC ratios are between 10 and 15% meaning that most of the PCASP submicron mass can be explained by biomass burning, and mineral dust mass is likely to be minimal. The derived Q-AMS CE is at the low end of that observed over a range of environments, but the assumed density used to calculate mass from the PCASP data is unlikely the factor of 2 too high required for agreement with the CE of Alfarra et al. [2007]. Hoffer et al. [2006] have investigated the volatility of organic carbon using evolved gas analysis and show that over 50% of the organic carbon may be refractory in nature. It is possible that a function of this material is not flash vaporized on the heater of the Q-AMS. Our derived CE may therefore be reduced. The absorption properties of the sampled aerosols were measured using a Radiance Research Particle Soot Absorption Photometer (PSAP). The PSAP reports a reduction in transmittance across a filter using a 567 nm LED, due to absorption by particles. PSAP data was corrected for errors in filter exposure, airflow rate, scattering by aerosols collected on the filter and multiple scattering within the filter, as outlined by Bond et al. [1999]. An estimate of black carbon (BC) mass concentration can be derived from the absorption measurement using an appropriate specific absorption coefficient (12 g m−2 was used in this study). The PSAP and the Q-AMS both sampled through Rosemount inlets, which have a particle transmission efficiency close to unity for particles smaller than 1 μm radius [Haywood et al., 2003a].

3. Regional Overview

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

3.1. Vertical Distribution of Aerosols

[9] Data was collected continuously throughout each flight on the research aircraft, and flight patterns included a mixture of vertical profiles to investigate the vertical distribution of aerosols, and straight and level runs (SLRs) to sample aerosol layers at different altitudes.

[10] The operating region during the field projects is shown in Figure 1 and has been divided into four characteristic regional zones, each representing distinctly different mean vertical aerosol distributions; see Figure 2 for example profiles typical of each zone. Zone 1, to the south of Niamey in Nigeria and Benin, represents the source region of most intense burning and was dominated by biomass burning aerosol at all altitudes. Zone 2, centered on the Niamey region, was characterized by presence of mineral dust at low altitudes, with aged, elevated biomass burning layers at altitudes between 1000 and 3000 m. To the NE of Niamey in central Niger, Saharan dust events yielded high mineral dust concentrations at low level, and transported biomass burning layers were observed at higher altitudes around 3000–4000 m (zone 3). During DODO, transported dust plumes from the western Sahara were sampled over Mauritania at altitudes below 1000 m, and transported biomass burning aerosols were encountered in layers around 3000 m off the coast of Guinea (zone 4). It can be seen that the biomass burning aerosol gets progressively more lofted with transport outside of zone 1.

image

Figure 1. Map of the operating region showing the flight tracks as white lines and the location of straight and level runs in biomass burning plumes as red crosses. The region has been geographically divided into four different characteristic zones as numbered on the map and described in the text.

Download figure to PowerPoint

image

Figure 2. Typical vertical profiles from each zone, illustrating aerosol number concentration, organic mass loading, and CO concentration.

Download figure to PowerPoint

[11] Figure 2 shows that there is a clear correlation between the measured organic mass concentration, CO and the total accumulation mode number concentration during each of the profiles. This was a consistent feature in all biomass burning plumes that were sampled. In zone 1, spikes associated with individual smoke plumes can be seen superimposed on the background biomass burning pollution. CO, OM and accumulation mode number also correlated closely during these individual plumes.

3.2. Composition of Aerosol Mass

[12] The data discussed in this paper have been screened to identify SLRs where the aerosol chemistry and PCASP number size distributions are dominated by biomass burning aerosols. Data are presented only from runs where mean CO concentration is at least 20 ppb higher than the background value, and mean CN count is greater than 1000 cm−3. From a total of 213 SLRs over the two field campaigns, 106 runs met these criteria and were categorized as “biomass burning runs.” The source region (zone 1) consisted of a large number of small fires (tens of meters in diameter) covering a large area: such fires are widespread in Africa at this time of year and cover much of the continent in a band between 4°N and 14°N. Satellite images from the MODIS Rapid Response Project at NASA/GSFC (Terra and Aqua true color subsets) indicate that there were a greater number of fires in the east of the operating region over Nigeria and Benin than farther west in Guinea-Bissau and Guinea.

[13] From visual observations, most of the fires appeared to be in the smoldering stage, emitting light-gray colored smoke. It should be noted that most of the region is quite rural and there are no major anthropogenic fossil fuel pollution sources, so the measurements of organic aerosol can be attributed almost exclusively to biomass burning. This is supported by measurements of light hydrocarbons, which show no evidence of fossil fuel combustion and indicate highly aged air (J. McQuaid, unpublished data, 2006). Later flights after biomass burning had ceased show that the organic mass loading was less than 1 μg m−3 across the whole region. These data represented the input from natural sources and will be discussed in a future publication. The chemically speciated average mass loadings for the biomass burning runs in each region are shown in Figure 3. Organic mass, sulphate (SO4), nitrate (NO3) and ammonium (NH4) loadings were measured by the Q-AMS. Black carbon estimates were derived from the PSAP, assuming a mass absorption efficiency of 12 m2 g−1, an average value for biomass burning aerosol in Brazil [Martins et al., 1998].

image

Figure 3. Chemical composition of aerosol mass measured in the four geographical areas described in Figure 1.

Download figure to PowerPoint

[14] The majority of the mass (up to 85%) measured by the Q-AMS throughout the measurement period was of organic composition, but inorganic species were also present in varying smaller quantities (discussed further in section 5). Black carbon is the second largest contributor to aerosol mass, making up 10–16% of the submicron mass loading. This result is consistent with previous measurements of biomass burning aerosol [e.g., Reid et al., 1998; Haywood et al., 2003b], which suggest black carbon mass fractions of 5–10%.

[15] The relationship between organic mass concentration measured by the Q-AMS over West Africa and CO is shown in Figure 4a, and a strong correlation (r = 0.93) is observed. All the runs shown as a red cross in Figure 1 correspond to a data point in Figure 4, so the latter represents the whole range of ages observed during the field campaign: low-level penetrations of individual fire plumes in the near field through to aged regional hazes. Most of the fires sampled were concentrated in Nigeria and Benin, indicating that likely transport times from the source region to zone 4 would be of the order of 5 days [Haywood et al., 2008]. CO is conserved on these transport timescales, having a lifetime due to oxidation by the hydroxyl radical of around 2 months [Wang and Prinn, 1999] and can therefore be used as an inert tracer for the biomass burning. None of the air masses had been subjected to wet or dry deposition, and so only aerosol chemical and microphysical processes had affected the aerosol mass. No significant change in OM to CO ratio was observed over all times and locations, and hence atmospheric ages, sampled throughout the West African plume. This is in contrast to measurements of anthropogenic pollution in continentally influenced northern hemisphere mid latitudes, where the majority of the organic aerosol mass is secondary in origin [Zhang et al., 2007] and the SOA fraction appears to increase over time [Crosier et al., 2007; de Gouw et al., 2005; Volkamer et al., 2006]. While the organic mass concentration to CO ratio appears constant, as discussed in section 5 the composition of the aerosol changes and so this is not simply a straightforward dilution process.

image

Figure 4. (a) Organic aerosol mass concentration versus excess CO (i.e., ΔCO = CO – background CO). Data are shown from both the DABEX and DODO campaigns, for all biomass burning runs which meet the selection criteria outlined in section 3. The error quoted on the fit coefficients is ± 1σ. (b) Residuals to the fit of OM and CO. Example error bars are shown separately for clarity in both Figures 4a and 4b.

Download figure to PowerPoint

[16] The average OM/CO ratio for the region is 0.055. For comparison, this lies between the primary emission ratio for urban environments of ∼0.004 [Zhang et al., 2005a] and that measured in aged air polluted by industrial emissions of 0.08 to 0.1 observed in the NE United States [de Gouw et al., 2005; Kleinman et al., 2007; Peltier et al., 2007] and the Po Valley, Italy [Crosier et al., 2007].

4. Physical Aging of Biomass Burning Aerosols

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

4.1. Number Size Distribution

[17] A consistent transformation in the shape of the size distribution was manifest in the PCASP data gathered across the region. The number distributions of fresh aerosol (the red traces in Figure 5) were dominated by particles with radii less than 0.1 μm, whereas the aged aerosol (the blue traces in Figure 5) were characterized by a peak in the number distribution at radii between 0.1 and 0.2 μm. The freshest aerosol size distributions in Figure 5 correspond to low-level runs where the aircraft flew through smoke plumes directly above visually observed fires see (Johnson et al. [2008] for details). The more aged aerosol size distributions are typically from runs at high altitudes (2–4 km), far from source regions (zones 2–4 in Figure 1). The fresh aerosol samples had very high CN concentrations (∼104 cm−3), whereas the aged biomass burning layers had much lower CN concentrations (∼103 cm−3). The transition in the shape of the normalized size distribution is quite consistent and is defined very well by the ratio of concentrations at r = 200 nm to r = 55 nm (C200nm/C55nm; see color scale in Figure 5). Variability in the number of intermediate size particles (radii > 0.35 μm) is most likely due to variability of dust concentrations [Johnson et al., 2008].

image

Figure 5. Size distribution from the PCASP, colored to the ratio of concentrations at r = 200 nm/r = 55 nm.

Download figure to PowerPoint

[18] The evolution of the size distribution is accompanied by a decrease in the total number concentration compared to the CO concentration (Figure 6a). This is indicative of coagulation since CO is conserved on short timescales of a few days, whereas particle numbers are not conserved. The coagulation seems to be related to the changes in size distribution shape shown in Figure 5, since the excess CO/CN ratio is positively correlated to the PCASP C200nm/C55nm ratio. There is a linear correlation between PCASP number concentration and organic mass concentration, though there is clearly also some scatter and evidence that fresh plumes show an enhanced number concentration compared to the aged plumes (Figure 6b). It seems probable then that the coagulation is dominated by particles smaller than those detectable by the PCASP (<55 nm radius). Aitken mode particles appear to be coagulating with the dominant 55 nm radius particles and growing them into the accumulation mode around 150 nm. Were a significant quantity of particles within the size range of the PCASP to be coagulating with each other this would act to reduce the PCASP number concentration faster at source than in the far field (where dilution is dominant), giving a nonlinear relation, which is not observed. The ratio between PCASP number concentration and OM does however appear to be lower when the size distribution indicates more aged aerosols (higher C200nm/C55nm); see color scale in Figure 6b. This suggests the coagulation is not exclusively with particles smaller than r = 55 nm, but also between particles within the size range of the PCASP.

image

Figure 6. (a) Shape of size distribution from PCASP versus CO/CN and (b) number concentration from the PCASP versus AMS organic mass concentration.

Download figure to PowerPoint

[19] Emissions from individual fire plumes during DABEX tended not to mix with clean background air, but instead into a boundary layer already polluted to varying degrees by biomass burning. For this reason, no trend between size distribution and CO concentration was observed and hence it was not possible to estimate the timescale for coagulation on the basis of the CO dilution rate.

[20] The change in the size distribution would have the tendency to increase the single scattering albedo of the aerosol because the particles are moving toward sizes where scattering is more efficient. Changes in particle composition and shape associated with condensation and internal mixing were also shown to increase the single scattering albedo of biomass burning aerosol during past experiments in southern Africa [Abel et al., 2003]. The evolution of single scattering albedo with aerosol age was difficult to observe during DABEX because the dust concentration was too high during flights when fresh biomass burning aerosol was observed, and therefore strongly biased aerosol optical properties [Johnson et al., 2008].

4.2. Model Simulation of Coagulation

[21] The coagulation process was simulated using a coagulation box model (including no dilution term because this is unknown). The coagulation model is based on the semi-implicit coagulation scheme described by Jacobson [1999] and uses a moving bin center on a fixed grid. The intervals of the 128 size bins are fixed and spaced according to a geometric progression in volume space (Volbin_n+1 = k*Volbin_n, where k is a constant) resulting in a logarithmic spacing in radius space. Particles in each bin are subject to growth through random collisions with particles of all sizes (i.e., Brownian coagulation). When two particles coagulate, they are assigned to a bin on the basis of their new combined volume, and this may act to increase or decrease the mean radius in that bin, as well as the bins the particles originated from. At the end of each time step, all particles in each bin are assigned the average size of the new bin population. This process is summed over all bins for each time step to yield the modified size distribution. The size distribution used to initialize the model was derived using a lognormal fit to a typical fresh size distribution from the PCASP convolved with SMPS measurements from a site at Djougou in Benin [Mallet et al., 2008], which was directly impacted by near field fire plumes during this time, in order to capture the fine mode distribution. A total CN concentration of 10,000 cm−3 was used, which is typical of concentrations sampled by the aircraft during the fire plume penetrations.

[22] The gradient of the measured size distribution at 55 nm < r < 100 nm changes from negative to positive with aging (Figure 5), a feature reproduced by the coagulation box model over an atmospherically realistic timescale (Figure 7). During the model simulation this change in the size distribution shape occurred concurrently with a decrease in total number concentration of 30- to 40-nm-sized particles over a time period of ∼1 day (Figure 8). This is consistent with the time frame of the measurements and supports the hypothesis that fine mode particles dominate the coagulation. Coagulation involves a rapid decrease in total particle number with time. In the field, this process can be inferred by an increase in the ratio of ΔCO to particle number concentration (CN) (Figure 6a), since CO is approximately conserved while particle number drops rapidly during coagulation. According to the model, the drop in total number concentration occurs simultaneously with the growth of a peak in number concentration at radii of around 100 nm (changing the distribution gradient from negative to positive for 55 nm < r < 100 nm). This helps to explain the correlation in Figure 6a, where CN/ΔCO is plotted against C200nm/C55nm (C200nm/C55nm gives a measure of the gradient of the distribution for 55 nm < r < 100 nm).

image

Figure 7. Number size distributions from the model simulation and PCASP measurements.

Download figure to PowerPoint

image

Figure 8. Contour plot describing the modeled number size distribution. At t = 0 s the distribution peaks around 0.03 μm, which evolves through coagulation to 0.1 μm by t = 80,000 s.

Download figure to PowerPoint

[23] The coagulation model does not develop the particle size distribution to the extent that is shown in the PCASP aged aerosol data. In particular, the model does not have as many particles in the size range of 100−200 nm radius. From back trajectory analyses [Chazette et al., 2007; Heese and Wiegner, 2008; Johnson et al., 2008] it seems likely that the aged aerosol would have included particles of up to 5 days in age, however a longer simulation yields no significant changes to distribution shape in this case, hence the simulation is shown only for one day. Furthermore, the model does not include dilution as the atmosphere contained multiple plumes diluting into one another. The model suggests that coagulation alone cannot simulate the growth of the observed size distribution between 100 and 200 nm over the timescale of several days, although the entrainment of multiple plumes into the same air parcel could affect the size distribution in this way. Size-dependent condensation and reevaporation processes may also have played a role in the size distribution evolution and these were not represented in the model.

5. Chemical Aging of Biomass Burning Aerosol

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

5.1. Mass Spectral Markers of Biomass Burning Aerosol

[24] While the OM/CO relation was observed to be constant for all plumes, the chemical functionality of the aerosols was observed to change with time. Mass spectra from the Q-AMS are shown in Figure 9, which illustrate the chemical transformation observed as the aerosols aged during the campaign in terms of the changing signal strength at key mass to charge ratios (m/z). The mass spectra shown in Figures 9b and 9c are averaged over complete SLRs (∼10 min) and are examples typical of the near and midfield. The mass spectrum shown in Figure 9c has been averaged over all SLRs in biomass burning layers in zone 4 in order to improve the poor signal-to-noise ratio from individual SLRs and represents the far field. There is a large degree of similarity between the near source mass spectrum (Figure 9b) and emissions from domestic wood burning in Europe (Figure 9a) [Alfarra et al., 2007]. Both spectra have larger signal at mass to charge ratio (m/z) 43 than m/z 44 and show clear peaks at m/z 57, 60 and 73 (the significance of these m/z values is detailed in the following paragraphs). The transformation from the freshest (near field) to the most aged (far field) spectra is best characterized by the increase in m/z 44 concurrent with a decrease in m/z 43, 57, 60 and 73 (representing 7.3%, 7.8%, 2.5%, 1.3% and 1.2% of OM respectively in the fresh spectrum – in the far field data m/z 57 was below the detection limit, but m/z 44, 43, 60 and 73 represent 14%, 3.8%, 0.31% and 0.69% of OM respectively) and is a clear demonstration that the chemical functionality of the aerosol alters over time. Note that the bottom of the sticks in the spectra in Figure 9 represent the calculated random uncertainty or error, which is due to the presence of a finite number of randomly generated ions that introduce an intrinsic variability in the number of ions detected by the mass spectrometer [Allan et al., 2003a].

image

Figure 9. Typical mass spectra of organic aerosols illustrating chemical changes as the aerosols age: (a) ambient spectrum from domestic wood burning emissions in Switzerland [Alfarra et al., 2007], (b) data from zone 1, representative of fresh biomass burning aerosol, (c) more aged aerosol from zone 2, and (d) very aged aerosol from zone 4. The tops of the vertical bars represent the signal at a particular m/z. The bottoms of the bars indicate the corresponding error, i.e., the calculated random uncertainty (1 sigma).

Download figure to PowerPoint

[25] Levoglucosan has widely been reported as a marker for biomass burning aerosols [Jordan et al., 2006; Simoneit et al., 1999] and has been shown to be one of the most abundant speciated compounds in Amazonian biomass burning periods [Decesari et al., 2006]. This is because levoglucosan and other anhydrosugars are emitted at much higher levels during the smoldering phase of fires [Schkolnik et al., 2005], a phase typical throughout the west African burning period during this study. Levoglucosan fragments at m/z 60 and m/z 73 are notably present in the spectra from fresh emissions throughout the measurement period. A good marker for fresh aerosols is m/z 57, which is a typical fragment of saturated hydrocarbon compounds or of long alkyl chains (C4H9+) [Alfarra et al., 2004], a fragment which was pronounced in the fresh spectra. Long-chain fatty acids and large sugars have previously been shown to be present in the organic fraction of biomass burning aerosols [Decesari et al., 2006; Falkovich et al., 2005; Fuzzi et al., 2007].

[26] The more aged spectra show an increase in the dominance of the m/z 44 fragment, indicating the presence of oxygenated organic compounds (signal at m/z 44 from the CO2+ ion). This results from thermal break down of van der Waals' coupled carboxylic acid groups in di- and multi-functional acids and peroxides on the AMS vaporizer, and is generally small for primary aerosols [Alfarra et al., 2004]. However, m/z 44 is also present albeit with a lesser significance in even the freshest spectra, and has previously been associated with wood burning emissions [Alfarra et al., 2007]. Runs with high m/z 57 and 60 also tend to have high m/z 43, and for those runs, m/z 43 often represents the greatest contribution to total organic signal. The m/z 43 is a characteristic fragment of both saturated hydrocarbon compounds or long alkyl chains (C3H7+) and oxidized functionalities found in organic compounds such as aldehydes and ketones (CH2CHO+ or CH3CO+). Concurrent presence of high m/z 43 with high 57 and 60, and low 44 suggests that the former may be more likely for these runs.

[27] The near source spectra provide a very distinct source fingerprint mass spectrum of biomass burning aerosols and are similar to those of Alfarra et al. [2007] from household wood fires in Switzerland (r = 0.869) and Schneider et al. [2006] from laboratory burns of savanna grass (r = 0.852). The aged spectra, on the other hand are similar to those observed in aged anthropogenic pollution plumes [Alfarra et al., 2004; Zhang et al., 2005a] and are characteristic of multifunctional high molecular weight products of secondary organic material [Alfarra et al., 2006]. This is also consistent with the observations of Hoffer et al. [2006] who used evolved gas analysis to demonstrate the increase in low-volatility, high molecular weight organic material in aged biomass burning aerosol.

5.2. Repartitioning of Organic Aerosols

[28] Donahue et al. [2006] maintain that the conventional treatment of primary emissions as nonvolatile is incorrect, and that all organic aerosol emissions should be treated using partitioning theory. Dilution with progressively cleaner background air can hence present the possibility for repartitioning into the gas phase. Oxidation of condensed organic aerosol species usually yields less volatile oxidation products; however oxidation can also lead to the volatilization of organic species, particularly in the latter stages, and so can act to decrease aerosol mass through evaporation.

[29] There is a clear, and consistent transformation in all the mass as a function of distance from the source region; the addition of oxygenated mass fragments at the relative expense of long-chain aliphatic mass fragments indicates oxidation is important and can be expressed as an increase in O/C. When combined with the fixed OM/CO ratio this shows that net carbon loss occurs throughout the lifetime of the aerosol in the region, and must mean that both chemical processing and evaporation are happening. The processing is an undetermined combination of: (1) condensation of additional secondary material and (2) a chemical transformation of the primary organic aerosol. Ozone concentrations during SLRs within biomass burning layers were on average 10 ppb higher than background values, suggesting photochemical activity within the plumes. Condensable products could result from this, in line with hypothesis 1. Either way the increase in O/C must be balanced by evaporation in order for the mass to remain constant.

[30] The residual to the regional OM/CO relationship (Figure 4b) shows considerable scatter, although there is some indication of a systematic reduction in OM at low CO; this may be an indication that in the later stages of dilution, repartitioning to the gas phase is larger than condensation. These data points at low CO also represent the most aged, oxygenated aerosol (farthest from source), which is consistent with the work from Donahue et al. [2006]. Despite the scatter, the signal to noise for individual data points is good and for clarity is shown for just a few example data points in Figure 4.

[31] There is some evidence that the inorganic fraction of the aerosol increases away from source (outside of zone 1; see Figure 2) and this may be indicative of secondary processing of NOx and SO2 from biomass burning sources. The uncertainty of the peak areas of the inorganic masses in the far field may be up to 50% of the signal. The surrounding organic mass peaks are similarly uncertain. Given that organic interferences on inorganic peaks are estimated on the basis of a combination of peak ratios and similarities to adjacent mass peaks, there is insufficient signal to estimate the magnitude of these systematic uncertainties.

[32] Organic mass concentration correlated well with absorption measurements over the whole region (Figure 10). The residual to the fit exhibits no significant skew, indicating that the changing organic composition observed at lower ΔCO (and hence lower OM) has little bearing on aerosol optical properties in this instance. Dinar et al. [2008] suggest that oxidation of wood burning aerosols can increase their absorption – so-called “brown carbon” [Andreae and Gelencser, 2006]; however during DABEX the BC/OM ratio seemed to dominate the optical properties.

image

Figure 10. Organic mass loading plotted against absorption coefficient.

Download figure to PowerPoint

[33] During DABEX, SLRs with low CO/CN tended to also have low m44/m57 and vice versa; there was some correlation between the physical and chemical aging measures, but that is not to suggest a causal link between the two.

6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

[34] Emission of aerosol particles and trace gases from fire sources can be expressed relative to that of a reference species such as CO2 or CO (known as an emission ratio), or relative to the mass of fuel burned (an emission factor). In this section, emission ratios have been calculated using the method outlined by Andreae and Merlet [2001], by dividing excess aerosol mass concentration (concentration measured in plume – background concentration) by the corresponding excess CO concentration. Data from zone 1 are shown as triangle markers in Figure 4a and represent unprocessed aerosols in the source region. A linear regression to these data points can hence be used to estimate an emission ratio for OM/CO of 0.057 ± 0.003 for biomass burning in West Africa during the 2006 dry season.

[35] Emission factors for trace gases and aerosols from a range of biomass burning sources have been compiled from the literature by Andreae and Merlet [2001]. Using these emission factors to derive an emission ratio, gives an organic carbon to CO ratio of 0.052. A conversion factor of 1.4 has been used to convert between OC and OM [White and Roberts, 1977] which yields an OM/CO ratio of 0.073, somewhat higher than the emission ratio from the DABEX data, but still broadly in agreement. However, one might not expect close agreement for these emission estimates because the estimates are in part generated using different methods. Also the DABEX data represent one specific region and time period (compared to the generalized estimates of Andreae and Merlet [2001], based on numerous sources from several different regions). However, unlike the estimates of Andreae and Merlet [2001], which were typically from single fires, the estimate presented here represents a wide geographical scale.

[36] An estimate for black carbon emission (BC/CO) from PSAP measurements of 0.0072 was derived from a linear regression to the zone 1 data. This is in good agreement with the estimate from Andreae and Merlet [2001] (0.0074). These two emission ratios combine to give a BC/OC estimate of 0.18, also in reasonable agreement with Andreae and Merlet [2001]. Table 1 summarizes the emission estimates from DABEX and comparison with literature values.

Table 1. Emission Ratios From DABEX Compared to Literature Values Compiled by Andreae and Merlet [2001]a
Emission RatioDABEX ValueAndreae and Merlet [2001]
  • a

    DABEX uncertainties reflect variability in the data, i.e., ±standard error.

OC/CO0.041 ± 0.0020.052
BC/CO0.0072 ± 0.00090.0074
BC/OC0.18 ± 0.030.14

[37] The emission ratios presented here have been estimated on the basis of a number of assumptions, which introduce uncertainties beyond the variability indicated in Table 1. The appropriate choice of OC to OM conversion is composition dependent, with 1.4 being typical of fresh, less oxidized organic matter and 2.2 and higher have been suggested for very aged aerosol [Turpin and Lim, 2001]. The emission ratio presented was determined from the near fire regions sampled in zone 1 and corresponds to the freshest aerosols which are those with the lowest O/C ratio and therefore lowest molecular weight per carbon weight hence the choice to use 1.4. We used a value of 12 m2 g−1 for BC mass absorption efficiency, the average from SCAR-B [Martins et al., 1998]. While this gives excellent agreement with BC/CO emission ratios in the literature, the values of mass absorption efficiency from that study range from 5 to 19 m2 g−1, consistent with the variability in literature values [e.g., Formenti et al., 2003; Liousse et al., 1993], and this introduces a considerable uncertainty into the BC emission estimate (0.0045 < BC/CO < 0.017). BC is associated with biomass burning emissions, and so we assume that the BC mass is largely attributable to particles of this size (diameter < 700 nm; see Figure 5); this is supported by the correlation between absorption and accumulation mode particles reported by Johnson et al. [2008]. However, it is not possible to determine the exact particle size range measured by the PSAP, and hence the relative contribution to absorption from coarse (e.g., dust) particles, which is another source of uncertainty in the BC estimate. It was not possibly to verify the PCASP derived AMS CE with another independent instrument. However, the fact that there is good agreement between the OC/CO emission ratios derived here and those available in the literature supports this approach.

[38] These emission estimates represent a large number of fires spread over a wide geographical area and though limited to a single season, are to our knowledge the first regionally averaged assessment of the emission ratios of BC and OC over such a large scale.

7. Conclusion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

[39] In this paper we have described the physical and chemical properties of biomass burning aerosols over a broad region of West Africa. Organic aerosol mass (OM) and CO were found to be linearly correlated during the 2006 dry season, which points to a lack of net secondary organic aerosol formation (SOA) production. Ozone production was observed within biomass burning plumes indicating photochemical activity, which suggests the atmosphere is capable of processing VOCs to form low-volatility products that can form SOA. The constant OM/CO over time but increasing oxygen to carbon ration in the mass spectrum demonstrate that some carbon loss is taking place, probably through reevaporation of organic mass during dilution and processing in line with Donahue et al. [2006]. The OM becomes more oxygenated over time, which is a clear indicator of atmospheric chemical processing. The chemical processing of biomass burning aerosol has previously been inferred from ground based measurements in the Amazon basin. Diel variation in organic composition was used to show that a transition from long-chain monoacids and sugars to high molecular weight polyacidic compounds occurs [Fuzzi et al., 2007; Hoffer et al., 2006]. The changes in the mass spectra shown across west Africa are direct evidence for such transformation and are consistent with the chemical speciation changes identified in Amazonia. What is less clear is the extent to which these changes arise from the processing of the primary OM only or whether there is active production of significant secondary organic aerosol that is balanced by evaporation of some fraction of the OM.

[40] This behavior is in stark contrast to continentally influenced northern hemisphere mid latitude environments, where previous measurements using the AMS have consistently shown a substantial increase in organic mass concentration (OM) compared to CO away from source, which represents secondary organic aerosol formation [e.g., Crosier et al., 2007; Heald et al., 2005]. These measurements show OM loadings of urban pollution an order of magnitude greater than the corresponding OM/CO emission ratio of approximately 4 μg m−3 ppm−1 [Allan, 2004; Zhang et al., 2005b], indicating that most of the observed mass is secondary in origin. In these polluted environments, modeling estimates consistently underpredict SOA by an order of magnitude, and the discrepancy between models and measurements increases over time [Volkamer et al., 2006]. The OM/CO ratio from West African biomass burning is around half that observed in polluted northern hemisphere midlatitude environments [Crosier et al., 2007]; the key difference is that the ratio doesn't change with time for biomass burning aerosol during DABEX. This indicates markedly different behavior in the transformation of organic material in biomass burning aerosol plumes sampled in West Africa compared to that in plumes from industrial sources that have been more widely studied. An improved knowledge of these processes is necessary to model the contrasting transformations of secondary organic material in both of these environments.

[41] Despite the total organic mass concentration only changing owing to dilution away from the source region, the aerosol size distribution has been shown to evolve over time. Near the source of emissions the number concentration peaked at radii of 55 nm, whereas in the far field the peak was at 150 nm. This evolution of size distribution was reasonably well captured by a simple coagulation box model, although the model was not able to fully capture the peak in particle concentration at 150 nm. Coagulation of Aitken-mode particles was primarily responsible for the evolution of the size distribution; it was not necessary to represent condensation to simulate the observed growth of aerosols, a result that is consistent with the chemical analyses of OM over time.

[42] Emission ratio estimates for BC and OC compared to CO (0.0072 and 0.041 respectively) were found to be in good agreement with literature values and to our knowledge represent the first regionally averaged assessment of the emission ratios for BC and OC over such a large scale. These represent the first estimates of organic and black carbon from biomass burning emissions over West Africa, one of the world's largest sources of black carbon.

[43] What is clear is that biomass burning aerosol behavior over West Africa in the 2006 dry season is quite distinct from that of organic aerosols in northern hemisphere midlatitude environments polluted by industrial emissions. Implications for understanding the aging of organic aerosols (in particular SOA) and the link between aerosol composition and their optical properties have been highlighted.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

[44] On the basis of a French initiative, AMMA was built by an international scientific group and is currently funded by a large number of agencies, especially from France, UK, United States, and Africa. It has been the beneficiary of a major financial contribution from the European Community's Sixth Framework Research Programme. Detailed information on scientific coordination and funding is available on the AMMA International web site (http://www.amma-international.org). The authors wish to acknowledge all the efforts of FAAM, the Met Office, and the BAe-146 air and ground crews. G. Capes was supported by a NERC studentship NER/S/J/2004/13127.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Aircraft Instrumentation
  5. 3. Regional Overview
  6. 4. Physical Aging of Biomass Burning Aerosols
  7. 5. Chemical Aging of Biomass Burning Aerosol
  8. 6. Emission Ratios of Organic and Black Carbon Estimated Over the West African Sahel
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  12. Supporting Information
FilenameFormatSizeDescription
jgrd14707-sup-0001-t01.txtplain text document0KTab-delimited Table 1.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.