Environ. Sci. Technol. 2005, 39, 5293-5301
Influence of Agricultural Biomass Burning on Aerosol Size Distribution and Dry Deposition in Southeastern Brazil GISELE O. DA ROCHA,† A N D R E W G . A L L E N , * ,‡ A N D ARNALDO A. CARDOSO† Instituto de Quı´mica, Universidade Estadual Paulista (UNESP), CEP 14801-970, Araraquara, SP, Brazil, and School of Geography, Earth and Environmental Sciences, University of Birmingham, Birmingham B15 2TT, U.K.
The size distributed composition of ambient aerosols is used to explore seasonal differences in particle chemistry and to show that dry deposition fluxes of soluble species, including important plant nutrients, increase during periods of biomass (sugar cane trash) burning in Sa˜o Paulo State, Brazil. Measurements were made at a single site centrally located in the State’s sugar cane growing region but away from the immediate vicinity of burns, so that the air sampled was representative of the regional background. Calculation of ion equivalent balances showed that during burning periods smaller particles (Aitken and accumulation modes) were more acidic, containing higher concentrations of SO42-, oxalate, NO3-, HCOO-, CH3COO-, and Cl-, but insufficient NH4+ and K+ to achieve neutrality. Larger particles showed an anion deficit due to the presence of unmeasured ions and comprised resuspended dusts modified by accumulation of nitrate, chloride, and organic anions. Increases of resuspended particles during the burning season were attributed to release of earlier deposits from the surfaces of burning vegetation as well as increased vehicle movement on unsurfaced roads. During winter months the relative contribution of combined emissions from road transport and industry diminished due to increased emissions from biomass combustion and other activities specifically associated with the harvest period. Positive increments in annual particulate dry deposition fluxes due to higher fluxes during the sugar cane harvest were 44.3% (NH4+), 42.1% (K+), 31.8% (Mg2+), 30.4% (HCOO-), 12.8% (Cl-), 6.6% (CH3COO-), 5.2% (Ca2+), 3.8% (SO42-), and 2.3% (NO3-). Na+ and oxalate fluxes were seasonally invariant. Annual aerosol dry deposition fluxes (kg ha-1) were 0.5 (Na+), 0.25 (NH4+), 0.39 (K+), 0.51 (Mg2+), 3.19 (Ca2+), 1.34 (Cl-), 4.47 (NO3-), 3.59 (SO42-), 0.58 (oxalate), 0.71 (HCOO-), and 1.38 (CH3COO-). Contributions of this mechanism to combined aerosol dry deposition and precipitation scavenging (inorganic species, excluding gaseous dry deposition) were 31% (Na+), 8% (NH4+), 26% (K+), 63% (Mg2+), 66% (Ca2+), 32% (Cl-), 33% (NO3-), and 36% (SO42-). * Corresponding author phone: +44 (0) 121 414 7299; fax: +44 (0) 121 414 3078; e-mail:
[email protected]. † Universidade Estadual Paulista (UNESP). ‡ University of Birmingham. 10.1021/es048007u CCC: $30.25 Published on Web 06/14/2005
2005 American Chemical Society
Introduction Atmospheric aerosols range in size from small aggregates of molecules to dusts of ∼100 µm diameter. The size distribution typically includes coarse (larger than 2 µm), accumulation mode (0.1-2 µm), and nuclei or Aitken mode (smaller than 0.1 µm) particles (1). Knowledge of particle composition according to size can provide valuable insights into likely origins as well as subsequent behavior and impact on atmospheric processes, since coarse particles are usually formed by mechanical processes such as resuspension of particles from the Earth’s surface, emission from volcanoes, and sea spray, while fine particles (of concern due to their influences on climate and health) largely arise from combustion (of fossil fuels and biomass) or gas-to-particle conversion due to reaction, coagulation, or condensation of gaseous precursors. Once formed, particles provide an environment for a variety of secondary surface or aqueous phase processes acting to transform their physical and chemical properties. Water-soluble material, which can constitute much of the mass of tropospheric aerosols (2, 3), is a key component influencing gas/particle partitioning, rates of aqueous phase reactions, and ability of the particles to act as cloud condensation nuclei (4-6). Various studies of size segregated aerosol chemistry have been reported previously (7-16). However to our knowledge the influence of widespread rural biomass burning in the tropics on aerosol size distribution, dry deposition and transport has not been explored. The composition of the lower troposphere in rural Sa˜o Paulo State, Brazil, is representative of a tropical/subtropical region having an agriculture-based economy and low population density. Here sugar cane production constitutes the single largest agricultural enterprise (others are citrus and coffee production), with ∼75% of the area planted with sugar cane in Sa˜o Paulo State requiring in situ burning of the crop prior to harvest, with very large emissions to atmosphere of gases and aerosols (17-19). Sa˜o Paulo State is located within the South Atlantic convergence zone (SACZ), where high convective activity, particularly during the summer months, extends from the Amazon basin into southeastern Brazil and over the Atlantic Ocean (20). The wettest month is normally January; however, in the anomalous 2000/2001 summer, peak rainfall occurred in March. The winter of 2000 was notable for an unusually long dry season relative to the long term trend, while the 2001 dry season was comparatively short and total rainfall during the summer of 2000/2001 was the lowest recorded in the past 40 years. As pointed out by Liebmann et al. (20), lower mean rainfall is associated with an increased frequency of extreme events in this region. There is therefore a need to understand the impacts of large-scale anthropogenic processes which could influence climate. One such process, and the most important in our study region, is agricultural trash burning. More distant regions of central Brazil also experience widespread burning due to scrub (savannah) management or forest clearance (although these are unlikely to significantly affect the lower tropospheric composition of Sa˜o Paulo State), while Sa˜o Paulo city is the world’s second largest metropolitan area and a major urbanindustrial center possessing multiple sources of pollutant emissions to the atmosphere. The work reported here forms part of a research program to quantify the impacts of agricultural burning on the lower troposphere as well as on deposition of key nutrient and acidifying species to natural and agricultural systems. Within this context measurement of the atmospheric aerosol particle size distribution was undertaken to enable estimation of VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5293
aerosol deposition rates as well as the importance of regional or long-range transport of aerosols. In Sa˜o Paulo State (but not elsewhere in Brazil) burning as a land management practice will be gradually phased out over the next decades to be replaced by mechanized harvest, which does not require prior burning of trash leaves, topography permitting. Nonetheless wastes will continue to be burned, in electricity generation plants rather than in situ on the plantations, with the extent of any overall environmental gain largely dependent on the efficacy of emissions control technology.
Experimental Section Samples were collected on a 25 m agl rooftop of the Sa˜o Paulo State University (UNESP) campus near Araraquara (17). Araraquara is located centrally within the sugar cane growing region of Sa˜o Paulo State; however, there are no sugar cane plantations in the immediate vicinity of the site (within 5-10 km, depending on sector), so that air masses sampled were representative of the regional background atmosphere. A 12-stage Micro-Orifice Uniform Deposit Impactor (MOUDI, MSP Corp., NC) sampler system (21) was used. Impactor stage D50 cutoffs (at a flow rate of 30 L min-1) were (µm) 18, 10, 5.6, 3.2, 1.8, 1.0, 0.56, 0.32, 0.18, 0.10, and 0.056. Teflon membrane filters (47-mm diameter) were used as collection substrates on all stages, including a back-up filter. Inspection of the filters and impactor internal surfaces postsampling showed no evidence of particle bounce using this substrate material. Samples were collected over periods of 48 h, beginning in April 1999 and continuing through February 2001, on a total of 39 occasions. After sample collection, filters were sealed in sterile Eppendorff tubes in airtight plastic bags and immediately stored in a freezer (at -12 °C). Soluble material was later extracted at room temperature into 5 mL of distilled, deionized, water plus 0.5 mL of propanol for 30 min under mechanical agitation. Analyses of the water-soluble species Na+, NH4+, K+, Mg2+, Ca2+, HCOO-, CH3COO-, NO3-, Cl-, SO42-, and oxalate were performed using a Dionex Model DX-120 Ion Chromatograph (Dionex Corporation, Sunnyvale, CA) as previously described (17, 18). The form of the oxalate ion is variable according to pH; under alkaline conditions during analysis the C2O42- ion predominates; however, within aqueous atmospheric aerosols at lower pH oxalate may be present as HC2O4-. In ion balance calculations we have assumed the oxalate ion to possess a double charge. Overall detection limits, determined as 3σ of blank unexposed impactor substrates and considering a nominal sampled air volume of 86 m3, were in the range 0.16 ng m-3 (NH4+) to 2.3 ng m-3 (Ca2+). Meteorological data were collected at the same site using a micrometeorological station (Campbell Scientific Ltd., U.K.) equipped with temperature, wind speed, wind direction, and pyranometer sensors with outputs recorded every 10 min. Backward air mass trajectories on sampling days were obtained using the British Atmospheric Data Centre trajectory service, with data points output every 6 h for 5 days backward in time, for arrival time 00:00 UTC and pressure 950 mb. At this subtropical location solar intensity ranged from 500 to 900 W m-2, with daily temperature maxima between 20 and 30 °C, and typical wind speeds Na+ > Mg2+ > K+. Artaxo and Hansson (27) reported a mean calcium concentration of 2.29 ng m-3 in particles
