Environ. Sci. Technol. 2010, 44, 4577–4582
Biomass Burning Contributions to Ambient VOCs Species at a Receptor Site in the Pearl River Delta (PRD), China B I N Y U A N , † Y I N G L I U , †,‡ M I N S H A O , * ,† SIHUA LU,† AND DAVID G. STREETS§ State Joint Key Laboratory of Environmental Simulation and Pollution Control, College of Environmental Sciences and Engineering, Peking University, Beijing, 100871, China, Chinese Research Academy of Environmental Sciences, Beijing, 100012, China, and Decision and Information Sciences Division, Argonne National Laboratory, Argonne, Illinois 60439
Received January 30, 2010. Revised manuscript received May 14, 2010. Accepted May 18, 2010.
Ambient VOCs were measured by a proton transfer reaction-mass spectrometer (PTR-MS) at a receptor site in the Pearl River Delta (PRD) during October 19-November 18, 2008. Biomass burning plumes are identified by using acetonitrile as tracer, and enhancement ratios (ERs) of nine VOCs species relative to acetonitrile are obtained from linear regression analysis and the source-tracer-ratio method. Enhancement ratios determined by the two different methods show good agreement for most VOCs species. Biomass burning contributions are investigated by using the source-tracerratio method. Biomass burning contributed 9.5%-17.7% to mixing ratios of the nine VOCs. The estimated biomass burning contributions are compared with local emission inventories. Large discrepancies are observed between our results and the estimates in two emission inventories. Though biomass burning emissions in TRACE-P inventory agree well with our results, the VOCs speciation for aromatic compounds may be not appropriate for Guangdong.
1. Introduction Biomass burning, as a significant source of volatile organic compounds (VOCs), is the burning of living and dead vegetation, including biofuel use for cooking and open biomass fires (1). VOCs participate in the formation of tropospheric ozone and other oxidants, as well as secondary organic aerosol (SOA) in the atmosphere (2). The resulting photochemical oxidants and fine particles can cause severe regional air pollution problems and contribute to climate change (2, 3). Though China experienced rapid economic growth since the 1970s, biofuel still provides more than half of energy consumption in rural areas of China (4). In addition, significant amounts of crop residues are burned in the open field after harvesting, with the fractions ranging from 6.6% to 25.6%, as investigated by different researchers (5). VOCs * Corresponding author e-mail:
[email protected]; phone: 8610-62757973. † Peking University. ‡ Chinese Research Academy of Environmental Sciences. § Argonne National Laboratory. 10.1021/es1003389
2010 American Chemical Society
Published on Web 05/27/2010
from biomass burning account for 23.3% of the total VOCs emissions according to a bottom-up emission inventory of China (6). However, uncertainties associated with the treatment of biomass burning in emission inventories are large, due to difficulties with obtaining detailed statistical data on the amount of burning that occurs and in measuring emission factors in the field (7). Laboratory simulations and field measurements of VOCs provide useful information to assess the importance of biomass burning. Recent studies have reported source profiles of nonmethane hydrocarbons (NMHCs) emitted from crop residues burned in traditional Chinese cookstoves (8); and receptor models have been applied to apportion biomass burning contributions to ambient NMHCs (9). However, very little is known about the emission characteristics of biomass burning and its contributions to oxygenated VOCs (OVOCs) in China, which appear to dominate the emissions of VOCs from biomass burning (10). In this paper we present the results from a measurement campaign using a proton transfer reaction-mass spectrometer (PTR-MS) technique at a receptor site in the Pearl River Delta (PRD), which is a central part of Guangdong Province in China. Several biomass burning plumes, as indicated by enhancements of acetonitrile mixing ratios were observed during the campaign. The emission ratios of nine VOCs species including five OVOCs relative to acetonitrile are determined by two different approaches. The contributions of biomass burning to VOCs mixing ratios are also investigated, and these estimates will help to better understand biomass burning emissions in China.
2. Measurements The Pearl River Delta region is a typical area to investigate the complexity of air pollution problems in the very fast urbanization process in China. A series of field measurement campaigns were organized in 2004, 2006, and 2008 in this area. Aiming at the formation mechanism of ozone and fine particles in fall, 2008 in PRD, two intensive sites: an urban one in Guangzhou city, and a rural one at Jiangmen, were set to measure gaseous and particulate pollutants. A commercial PTR-MS (Ionicon Analytik GmbH) was employed at Jiangmen to conduct VOCs measurements from October 19 to November 18, 2008. The Jiangmen site (22.32°N, 112.53°E) is about 120 km away from Guangzhou city to the southwest. The sampling site has little influence from local emissions and is surrounded by shrubs and eucalyptus forest. PTR-MS is a novel method to measure VOCs in the atmosphere with high sensitivity and fast response. The detailed setup of PTR-MS can be found in the review paper (11). During the field campaign, PTR-MS measured a total of 35 masses in the selected ion mode at a time resolution of 160 s. Instrument background signals were determined by passing sampled air through an activated charcoal trap after every 35 cycles of ambient measurements. The PTR-MS system was calibrated every 3-5 days using a TO15 mixture standard (Air Environmental Inc., Denver, CO) at seven different mixing ratios ranging from 1 to 15 ppbv. Concentrations of acetonitrile and nine other VOCs species (Table 1), with sensitivities well quantified by periodic calibrations, are reported in the following discussion. Measurement precision was better than 10% for all VOCs compounds when mixing ratios were >1 ppbv. The detection limits of the system were compound-specific with values ranging between 20 and 250 pptv (Table 1). VOL. 44, NO. 12, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4577
TABLE 1. Detection Limits, Enhancement Ratios Relative to Acetonitrile, and Biomass Burning Contributions to Various VOCs Species mass (amu)
VOCs species
detection limit (ppt)
79 93 107 45 73 33 59 69 71 42
benzene toluene C8 aromatics acetaldehyde MEK methanol acetone isoprene+furan MVK+MACR acetonitrile
34 138 42 53 186 233 111 32 21 38
enhancement ratios (ppbv/ppbv) STR method
whole period
10/19-11/11
11/12-11/18
1.2 ( 0.9 2.0 ( 1.5 2.6 ( 1.9 3.0 ( 1.3 3.1 ( 2.3 12.6 ( 8.5 6.4 ( 2.6 2.6 ( 0.9 2.3 ( 0.8 -
1.5 ( 1.2 2.5 ( 2.3 2.7 ( 2.2 2.3 ( 1.7 3.6 ( 3.0 11.5 ( 8.0 7.1 ( 4.4 1.7 ( 1.1 1.2 ( 0.7 -
9.5 9.7 14.4 17.7 13.3 13.5 12.3 14.1 14.3 -
6.2 6.4 12.6 12.4 6.4 10.9 8.8 8.3 8.2 -
19.6 19.8 20 34.3 34.4 29.7 21.4 31.2 32.2 -
FIGURE 1. Acetonitrile measured at Jiangmen site. (a) Temporal variations of acetonitrile and MODIS fire counts in Guangdong during the campaign. The gray areas are four acetonitrile enhancement periods. The dashed boxes are days (10/21-10/22, 11/3-11/4) with very low acetonitrile mixing ratios. (b and c) Diurnal variations of acetonitrile during November 12-18 and November 3-4. The gray areas show standard deviations.
3. Results and Discussion 3.1. Time Series of Acetonitrile and Fire Hotspot Data. Acetonitrile (CH3CN) is an excellent indicator for biomass burning (12). It shows significant enhancements in biomass burning plumes, but no distinct increase over background mixing ratios in urban and power plant plumes (12). In addition, acetonitrile reacts slowly with hydroxyl radical and has a lifetime as long as several months (13, 14). Figure 1 shows the temporal variation of acetonitrile measured at the Jiangmen site. Several periods with acetonitrile enhancements were observed during the campaign: October 19, October 23-24, October 29-31, and November 12-18, as illustrated in Figure 1. During these periods, acetonitrile mixing ratios were much higher than the levels on the other days and reached as high as 1.5 ppb. Diurnal variations of acetonitrile during these biomass burning episodes show that acetonitrile peaked at 8:00-10:00 and 18:00-22:00 and remained at relatively high levels throughout the night (Figure 1b). In contrast, acetonitrile mixing ratios were very low (0.1-0.2 ppb) during the periods of October 21-22 and November 3-4, and there was no distinct diurnal variation (Figure 1c). It is believed that the influence of biomass burning was very slight during these days. During the periods of acetonitrile enhancement, crop residue fires were often observed in nearby rice fields, especially in the mornings and late afternoons. It is typical practice in the PRD region to harvest rice in the beginning of November, and then burn the rice straws in the field in the following days (15). Therefore, high acetonitrile mixing ratios during November 12-18 are attributed to intensive 4578
9
biomass burning contributions (%)
linear regression
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 12, 2010
open burning of crop residues in the field. Although field burning of crop residues is illegal in China, the regulations to ban straw burning have not been fully implemented in remote areas, and large quantities of crop residues are still burned in the field (15). Figure 1 also shows daily fire hotspots number detected by the Moderate Resolution Imaging Spectroradiometer (MODIS) instruments aboard the Terra and Aqua satellites in Guangdong (16). Very few fire hotspots were detected during the periods October 21-22 and November 3-9, consistent with the low measured acetonitrile mixing ratios. On the other hand, 10-20 fire hotspots were detected per day during November 12-18, which were also identified as biomass burning episodes using the acetonitrile tracer. However, acetonitrile levels were high but fire counts were low in some days (October 23-24, 30-31), possibly due to higher cloud fractions in these days disturbing the detection of fires (http://ladsweb.nascom.nasa.gov). The good agreement between acetonitrile mixing ratios and fire hotspots numbers, especially during the rice harvest time, indicates that the measurements of acetonitrile at Jiangmen site reflect the regional characteristics of biomass burning in the PRDsor perhaps Guangdong as a whole. It is likely that significant quantities of rice residues are also burned in the field in other parts of Guangdong after harvest. A large number of fire hotspots were also detected in the middle of November in other southern provinces of China, such as Fujian, Jiangxi, and Hunan (Figure S3 in the Supporting Information). This suggests that the open burning of crop residues is a widespread phenomenon in southern China. 3.2. Enhancement Ratios of VOCs in Biomass Burning Plumes. The enhancement ratio (ER) is a useful parameter in studies of biomass burning. It is defined as the excess mixing ratio of a species (∆X, the mixing ratio of species “X” in the plume above the mixing ratio of species “X” before or after the plume) divided by the excess mixing ratio of another species (∆Y), usually a fairly long-lived tracer (17). In this study, two different methods are used to calculate ERs of VOCs species relative to acetonitrile. ERs can be determined by correlating VOCs species with acetonitrile and calculating the slope of regression line between the two compounds when a biomass burning event was occurring (18). Significant peaks of acetonitrile mixing ratios were identified as biomass burning plumes (13), and 11 biomass burning plumes were identified from PTR-MS measurements. Figure 2 shows one of the 11 biomass burning events, occurring on November 13, 2008. Mixing ratios of acetonitrile and other VOCs were stable and low before 7:53 a.m. Acetonitrile mixing ratios increased quickly from 0.27 to 1.45 ppb in about 20 min, then declined slowly afterward. Other VOCs species also peaked at around 8:15 a.m. and followed variation patterns similar to that of acetonitrile. Mixing ratios
FIGURE 2. Biomass burning plume in the morning of November 13, 2008 of eight VOCs species as a function of acetonitrile in the time window of 7:53-8:15 are plotted in Figure 3. For all OVOCs species, correlation coefficients (r) were higher than 0.9. Moderate values of correlation coefficients were observed for aromatic compounds, which may be attributed to the presence of anthropogenic sources of aromatic compounds (e.g., solvent use in factories) in the PRD (9) and the potential mixing of plumes from these sources with biomass burning plumes during transport. Orthogonal distance regression (ODR) was applied to linear fittings and regression results for the fire plume are also shown in Figure 3. The other way to calculate enhancement ratios is a sourcetracer-ratio (STR) method (19). The STR method splits VOCs concentrations ([VOC]) into two parts (eq 1): concentrations attributed to biomass burning ([VOC]bb) and concentrations attributed to other sources([VOC]other). Enhancement ratios (ERs) are determined when the R2 values between VOCs concentrations from other sources and acetonitrile concentrations ([CH3CN]) reach their minimum. In this study, enhancement ratios are calculated from measured data for every two hours (typically 35-40 data points). [VOC] ) [VOC]BB+[VOC]other ) ER × ([CH3CN] [CH3CN]bg) + [VOC]other (1) Only background concentrations of acetonitrile ([CH3CN]bg) are considered in the calculation of ERs. Background concentrations of other VOCs compounds do not affect the calculation of ERs and biomass burning
contributions, so we omit this part (19). Acetonitrile backgrounds are estimated from average concentration measured on very clean days (November 3-4). It is believed that there was no biomass burning influence on these days. The data for October 21 and 22 are not used for the background calculation because air arrived at Jiangmen site from the South China Sea, based on back-trajectory analysis for those two days (Figure S4), and acetonitrile may be uptaken by ocean (12). The average acetonitrile concentration for November 3-4 was 0.19 ( 0.03 ppb, which is only slightly higher than the value determined for the free troposphere over the Pacific Ocean (0.15 ppb) (14). We chose 0.19 ppb to be representative of acetonitrile background concentration in the PRD region. ERs determined from linear regressions and the STR method are shown in Table 1. It should be recognized that there are differences in time spans for calculating ERs in the two methods: linear regressions are based on data points when biomass burning plumes were encountered, whereas the STR method determines ERs for every 2-h period during the campaign (e.g., 8:00-10:00 for a given day). Nevertheless, the agreements between average ERs calculated by the two methods are within 30% for most of the VOCs species, with the exceptions of mass 69 and mass 71 (Table 1). ERs for mass 69 and mass 71 determined by the STR method were significantly lower than those from the linear regressions (p-value < 0.05 for t test). The GC-PTRMS measurements of biomass burning plumes showed that isoprene and furan, methyl vinyl ketone (MVK) and methyl acrolein (MACR) dominated the signals at mass 69 and mass 71, respectively (10). The differences between ERs in mass channels of 69 and 71 may be due to the high reactivity of isoprene, MVK, and MACR. The STR method includes some time intervals when biomass burning plumes have been well mixed with other plumes, which means that a substantial fraction of isoprene, MVK, and MACR may have been removed by photochemical reactions. Although differences exist between ERs estimated by the two methods, emissions of isoprene, MVK, and MACR from biomass burning are not negligible. Therefore, these compounds should be used as unique tracers for biogenic emissions with caution when biomass burning influence is significant. 3.3. Biomass Burning Contributions to VOCs. Biomass burning contributions to concentrations of the nine VOCs were determined by the STR method. With the 2-h ERs determined, the contributions from biomass burning can be calculated according to eq 1. Figure 4 shows the time series of biomass burning contributions to selected VOCs species. Biomass burning contributions varied significantly from day to day. It showed the largest contributions during the four acetonitrile enhancement periods. Biomass burning contributed 19.6%-34.4% of the VOCs mixing ratios in November
FIGURE 3. Correlations of VOCs with acetonitrile in a biomass burning plume in the morning of November 13, 2008. Here, s is slope from linear regressions, and r is correlation coefficient. VOL. 44, NO. 12, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4579
FIGURE 4. Time series of biomass burning contributions to benzene, toluene, acetaldehyde, and acetone. 12-18, while only 6.2-12.6% of the VOCs mixing ratios from October 19 to November 11. There were also differences in biomass burning contributions to the nine VOCs. Biomass burning contributed most to acetaldehyde concentrations, with an average contribution of 17.7% throughout the campaign, as shown in Table 1. Biomass burning contributions to benzene (9.5%) and toluene (9.7%) were the lowest among the nine VOCs species. The relative contribution of biomass burning is dependent on the emission strength of biomass burning itself and on those of other sources as well. As mentioned before, the lowest contributions of biomass burning to aromatics may be due to large emissions from industrial sources in the PRD region (9). 3.4. Comparisons with Other Studies. The contributions of biomass burning estimated in this study are compared with previous ambient studies and emission inventories. Biomass burning contributions to the sum of all anthropogenic compounds (benzene, toluene, C8 aromatics, methanol, acetaldehyde, acetone, and MEK) measured by PTR-MS are used for the comparison. Here, both NMHCs and OVOCs species are included, thus the photochemical removal of NMHCs and formation of OVOCs in the atmosphere could be partially canceled out. The average biomass burning contribution is calculated to be 12.6% on a volume basis and 11.8% on a weight basis during the whole campaign, which are used for the comparison of ambient studies and emission inventories, respectively. The average contributions could be considered as a representative average contribution in the two months of October and November.
FIGURE 5. Average monthly fire counts by MODIS satellites in Guangdong from 2004 to 2008. The dashed line is the average fire counts in the 60 months from 2004 to 2008. The diamond line is average fire counts in October and November in the five years. Previous studies related to biomass burning in PRD or Guangdong spanned the years from 2000 to 2006 (Table 2). Ambient studies focused on one or several specific months, whereas emission inventories were usually estimated for a whole year. So, it is important to investigate interannual and seasonal variations of biomass burning emissions in Guangdong before comparison with other studies. A recent study indicated that open biomass burning emissions in Guangdong declined in recent years (20). The demand of biofuel for cooking was also decreasing due to rapid economic development in rural areas (5). On the other hand, VOCs emissions from anthropogenic sources increased in Guangdong by about 50% from 2000 to 2006 (21). Adding up these facts, contribution of biomass burning to total VOCs emissions ought to decrease gradually from 2000 to 2008. Seasonal variations of fire hotspots detected by satellites are good surrogates for those of biomass burning emissions (22). The monthly average fire counts detected by MODIS satellites from 2004 to 2008 are shown in Figure 5. Most of the fire hotspots were detected in the months from November to March, similar to a recently reported seasonal pattern of fire hotspots in southern China (23). Rice and sugar cane, as the most important crops in Guangdong, are harvested in November and December and their crop residues are burned in the field after straws are air-dried. In addition, forest fires usually occur in the winter months, because Guangdong has a drier climate in winter. Though distinct monthly variations are observed, the average emission strength of biomass burning in October and November is only slightly higher (
