Black Carbon Aerosol at McMurdo Station, Antarctica

Transcription

1 TECHNICAL PAPER Hansen, Lowenthal, Chow, and Watson ISSN J. Air & Waste Manage. Assoc. 51: Copyright 2001 Air & Waste Management Association Black Carbon Aerosol at McMurdo Station, Antarctica Anthony D.A. Hansen Lawrence Berkeley National Laboratory, Berkeley, California Douglas H. Lowenthal, Judith C. Chow, and John G. Watson Desert Research Institute, Reno, Nevada ABSTRACT Aerosol light absorption as black carbon (BC) was measured from November 19, 1995, to February 6, 1996, at a location 0.65 km downwind of the center of McMurdo Station on the Antarctic coast. The results show a bimodal frequency distribution of BC concentrations. Approximately 65% of the measurements were found in a mode at a low range of concentrations centered at ~20 ng/m 3. These concentrations are higher than those found at other remote Antarctic locations and probably represent contamination from the station. The remaining measurements were in a high-concentration mode (BC ~300 ng/m 3 ), indicating direct impact of local emissions from combustion activities at the station. High values of BC were associated with winds from the direction of the station, and the BC flux showed a clear directionality. Maximum BC concentrations occurred between 7:00 and 11:00 a.m. The polluted mode accounted for more than 80% of the BC frequency-weighted impact at this location. INTRODUCTION The combustion of carbonaceous fuels for the production of energy inevitably results in the emission of air pollutants. One of these is aerosol black carbon (BC). This material contains elemental carbon (EC) in a microcrystalline graphitic form, possessing a very large optical absorption IMPLICATIONS This paper illustrates the use of continuous, near-real-time aethalometer measurements to elucidate the variations of combustion-related aerosol emissions at a remote Antarctic research station. BC concentrations varied as a function of both wind direction and source strength over the diurnal cycle. Air quality in urban and residential settings is affected by a complex mixture of combustionderived emissions from mobile and stationary sources. The ability to measure BC on time scales as short as 5 min represents a powerful tool for characterizing the temporal and spatial variability of these sources. cross section. Aerosol BC is a unique tracer for combustion emissions: it has no noncombustion source. It is inert and long-lived in the atmosphere, and can be transported over great distances. 1 It may promote reactions between other species through catalytic action on its highly porous surface. It may affect the optical properties of the atmosphere when suspended, and of surfaces when deposited. This last point is of potential importance in the polar regions. When combustion emission aerosols are deposited to the naturally white snow and ice surface, a change in surface albedo will result. Aerosol BC is usually associated with other carbonaceous species such as polycyclic aromatic hydrocarbons and other volatile and particulate organic compounds, which may have toxic effects in highly stressed ecosystems. Thus, the measurement of BC in remote and otherwise pristine areas is an important indicator of short- and long-range environmental impact from combustion emissions. 2,3 The aethalometer instrument (Magee Scientific Co.) measures BC automatically in near-real-time using an optical technique that has been calibrated against chemical analysis. 4-6 Its deployment in areas such as Antarctica provides information on the local and global distribution of combustion emissions. 7,8 METHODS This project was performed in conjunction with a study to determine the sources of PM 10 (aerosol particles with aerodynamic diam less than 10 µm) at McMurdo station Aethalometer measurements were conducted during austral summer, McMurdo Station (77 51'S, 166º41'E) is located at the tip of the Hut Point peninsula at the southern tip of Ross Island. The aethalometer was deployed next to Scott s Hut (named for the British explorer, Robert Scott), located ~0.6 km west of the center of McMurdo. There are no stationary combustion sources in the immediate vicinity of the sampler, but the wind frequently comes from the direction of the station and its associated sources. The location of the sampling site with Volume 51 April 2001 Journal of the Air & Waste Management Association 593

2 respect to the major emissions sources in McMurdo is shown in Figure 1. The intent of this experiment was to measure BC continuously on a 5-min average time base without any selection, to analyze the data with respect to the meteorological record, and to examine the statistical distribution of the BC measurements. The aethalometer was installed in a small weather shelter, with a non-size-selective aerosol inlet at a height of ~2 m above the ground. The ambient air was drawn through the inlet at a flow rate of 18 standard L/min, and collected on a quartz fiber filter. The instrument integrates the optical absorbance of the filter over the 5-min measurement period, and interprets the increases in blackness in terms of an accumulation of optically absorbing BC mass. The filter tape is automatically advanced when the deposit reaches a preset absorbance value, well before optical saturation. The instrument performs diagnostic checks and provides an estimate of its internal noise and BC resolution. Observations of surface wind speed and direction were recorded every 3 hr in central McMurdo. Aethalometer BC and meteorological data files were transmitted to the United States every 2 weeks during the course of the experiment. The data files were checked for internal consistency Figure 1. McMurdo Station. Shown are locations of the sampling site at Hut Point, the power plant, and other large buildings. 594 Journal of the Air & Waste Management Association Volume 51 April 2001

3 and instrumental performance, and were then merged with the wind speed and direction records. RESULTS AND DISCUSSION The aethalometer was operated from 3:00 p.m. on November 19, 1995, until 4:00 p.m. on February 6, The total possible number of 5-min sample periods was 22,764; of these, data were recorded for 22,489 periods (99% instrument availability). The 1% nonrecording time was due to automatic filter tape advancing. The instrument s internal assessment of noise was equivalent to a level of 3.8 ng BC/m 3. A cutoff of 5 ng/m 3 BC was taken as the lower limit of valid data for the 5-min averaging period. Thirteen percent of the measurements fell below this limit, probably representing times with very clean air. For statistical purposes, data below the detection limit were assigned a value of 2.5 ng/m 3, that is, onehalf the detection limit. BC concentrations showed variability between two modes: one with low BC concentrations of ng/m 3, and the other with concentrations of hundreds to many thousands ng/m 3, reflecting strong local influence. Ships were docked at the ice pier in Winter Quarters Bay, located just upwind of the sampling site, on 18 days during the experiment. Table 1 lists average BC concentrations for periods when ships were present and when they were not. The presence of ships and their attendant activity, including heavier motor vehicle traffic on the road to Hut Point, resulted in higher BC concentrations. This was particularly the case after January 27, when a fuel resupply ship and the R/V Greenwave, which resupplies the station and carries cargo back to the United States, were docked at the pier with their engines running. Because of the close proximity of the ice pier to the sampler, ship emissions must be considered local contamination with respect to the more numerous but more distant sources in McMurdo. The 5237 sample periods affected by the presence of ships have been excluded from the following analysis. Table 1. Average BC concentrations (ng/m 3 ) when ships were present or absent from the ice pier. Date Average BC Ship Present 11/19/95 01/08/ No 01/09/96 01/11/ Yes a 01/12/96 01/16/ No 01/17/96 01/21/ Yes b 01/22/96 01/26/ No 01/27/96 02/02/ Yes c 02/03/96 02/06/ Yes d a Icebreaker Polar Star; b R/V Nathaniel B. Palmer (engines off, generators running); c Fuel ship (engines running); d R/V Greenwave. Comparison of BC and EC As part of the McMurdo source apportionment study, 48-hr duration PM 10 samples were collected at Hut Point on quartz fiber filters. 9,10 These samples were analyzed for organic carbon and EC by thermal optical reflectance. 12 The EC and corresponding 48-hr average BC concentrations are compared in Figure 2. The correlation coefficient (r) is 0.92 and the average ratio of BC to EC is 0.76 ± 0.31 (average ± standard deviation). For comparison, the average ratio of BC to EC and r for hr average samples from seven U.S. cities were 0.76 ± 0.16 and 0.97, respectively. 13,14 Because BC and EC are derived from light absorption and thermal evolution, respectively, there is no reason why they should be equivalent. It is remarkable that while the average EC concentration at McMurdo was an order of magnitude lower than that found in U.S. urban areas, the relationships between BC and EC were similar. Frequency Distribution of Measurements Individual 5-min measurements were grouped according to BC value in a frequency distribution. The frequency distribution of measured BC concentrations is shown by the unshaded bars in Figure 3. The cumulative frequency distribution is shown by the dashed line. The distribution is bimodal. The lower clean mode has its peak at ~20 ng/m 3. This is comparable to the summer monthly median of ~10 ng/m 3 reported for Barrow, AK, but much higher than austral summer BC concentrations reported for the South Pole (1 ng/m 3 ) 8 and Halley Station (1 2 ng/m 3 ). 15 Concentrations in this mode probably represent diffuse contamination from the station. The upper polluted mode has its peak at ~300 ng/m 3. This mode reflects the direct impact of emissions from sources in McMurdo. The other data shown in Figure 3 represent the weighted impact, the product of BC concentration and its frequency of occurrence. This is the appropriate variable for estimating the depositional impact. If the deposition velocity is concentration-independent, then the total deposition to the surface is proportional to this product. (In this study, we were not able to assess the effects of variable wind velocity and surface turbulence on deposition to the surface.) The weighted impacts are shown in Figure 3 by the shaded bars for the frequency distribution, and by the solid line for the cumulative frequency distribution. It is clear that the overall impact is dominated by the polluted mode measurements, whose BC concentrations were orders of magnitude greater than those in the clean mode. At the peaks of the two modes, the cumulative (weighted) impact of the upper mode was ~7 times that of the lower mode. About 40% of the weighted impact was due to measurements in the ng/m 3 range, but occasional measurements at higher BC concentrations contributed some 10% to the total impact. Volume 51 April 2001 Journal of the Air & Waste Management Association 595

4 Figure 2. Comparison of 48-hr average BC, measured with the aethalometer, and EC measured on aerosol filter samples by thermal optical reflectance; r is the correlation coefficient and P is its significance level. Figure 3. Frequency distribution of measurements of aerosol BC near McMurdo Station, Antarctica, excluding samples contaminated by ship effluent. Unshaded bars: frequency of occurrence of BC concentrations. Dashed line curve: cumulative frequency distribution. Shaded bars: weighted impact distribution (product of BC concentration and frequency). Solid line curve: cumulative frequency distribution of frequency-weighted BC impact. 596 Journal of the Air & Waste Management Association Volume 51 April 2001

5 Diurnal Cycle of Measurements All sample periods unaffected by ship presence were grouped by the hour of the day to examine any cyclic nature of the McMurdo source emissions. The sparse meteorological data show no diurnal pattern in wind direction or wind speed. There are 24 hours of daylight at McMurdo in summer and there is little diurnal surface temperature variation. Therefore, diurnal patterns in BC concentrations should be due to source emission cycles. Hourly average, geometric mean, and median BC concentrations are shown in Figure 4. The geometric mean and median represent the expected or typical value of a lognormal distribution. The arithmetic mean, which varies more over the diurnal cycle, reflects the influence of higher concentrations caused by meteorological factors and variations in source strength. All three parameters indicate a similar pattern: a peak between 7:00 and 11:00 a.m., lower concentrations in the afternoon hours, and the lowest concentrations in the early morning hours. The morning peak corresponds to the beginning of the workday, when vehicles move around and out of the station. The break for lunch is apparent during the noon hour, and concentrations increase again during the afternoon. Emissions from power generation, water production, and space heating should be relatively constant throughout the day under conditions of 24 hours of daylight. The reported projected diesel fuel (AN-8) consumption at McMurdo for power generation, heating and water production, and equipment operation (including motor vehicles) during austral summer are 4943, 4708, and 2119 gal/day, respectively. 16 Equipment operation accounts for 18% of the total. The operations contribution to BC can also be estimated from the mean hourly BC concentrations shown in Figure 4 by assuming that the average BC concentration from 12:00 to 7:00 a.m. and 8:00 to 11:00 p.m. (129 ng/m 3 ) represents the nonoperations fuel-use component. The operations component can be taken as the deviation from this average from 8:00 to 10:00 a.m. and 1:00 to 7:00 p.m. The ratio of operations BC to total BC over the 24-hr period is 11%, similar to the ratio based on projected fuel consumption. BC Concentration and Flux versus Wind Sector The meteorological data record provided readings of wind speed and direction every 3 hr, although many days did not contain a full complement of eight readings. Figure 5 presents the wind rose or frequency distribution of the wind speeds and directions. The prevailing wind directions are mainly in the east quadrant. Because there were often large changes in wind direction from one entry to the next, we felt that it was inappropriate to apply a single meteorological value to all BC measurements in the preceding Figure 4. Diurnal cycle of average BC concentrations grouped into 1-hr bins. Volume 51 April 2001 Journal of the Air & Waste Management Association 597

6 Figure 5. Wind rose: frequency of wind speeds and associated wind directions at McMurdo Station. Measurements recorded every 3 hours. 3-hr period. BC concentrations were therefore averaged over the half-hour preceding the meteorological entry. Figure 6a plots average, geometric mean, and median BC concentrations against corresponding wind directions grouped into 30 sectors. Note that the winds at the site may be different than those measured in central McMurdo because of the complex local topography. The peak average concentrations are found in the east degree sector, that is, from the direction of the station. There are also elevated concentrations in the south sector, which includes the airfield on (frozen) McMurdo Sound. The ice runway is oriented roughly north-south ~2 mi from McMurdo Station on McMurdo Sound. There are usually several takeoffs and landings of LC-130 and C-141 aircraft daily. Some elevated concentrations (<50 ng/m 3 ) associated with northwesterly winds may reflect some low-level scientific and logistical activity, including helicopter traffic on McMurdo Sound just around and to the northwest of Hut Point. 598 Journal of the Air & Waste Management Association Volume 51 April 2001

7 that is, on the order of 0.02 g BC/sec, or 2 kg/day. The total suspended particle (TSP) combustion emissions from power generation, water production and heating, and operations was estimated to be 66 kg/day. 16 Fuel use was assumed to be within 10% of that estimated for austral summer, TSP emission factors for the power generators were measured directly (11.83 kg/day) and agreed to within a factor of 5 with those published by the manufacturer. 18 All other TSP emission factors were derived from published values for emission factors. 19 Estimates of BC emission rates were not made. BC emissions from internal combustion emissions (power generators and motor vehicles) are found mainly on particles with diameters less than 2.5 µm (PM 2.5 ). 20 TSP and PM 10 emissions from combustion sources may contain coarser material, such as fly ash. Both TSP and PM 10 emissions from the McMurdo power generators were measured, in a ratio of 3.2:1. Applying this factor to total estimated TSP emissions yields a PM 10 emission rate of 20 kg/day. EC comprised ~50% of PM 2.5 emissions from heavy-duty vehicles BC burning Jet-A diesel fuel. 21 Results from our McMurdo PM 10 source apportionment study show that EC comprised 39 47% of PM 2.5 emissions from the power station. Using 50% as the BC fraction of PM 10 emissions, the BC emission rate based on the estimated PM 10 combustion-related emission rate is 10 kg/day. Figure 6. (a) Aerosol BC concentrations by wind direction grouped into 30 sectors. (b) Flux of aerosol BC (concentration multiplied by wind speed) by wind direction grouped into 30 sectors. The ordinate labels refer to the lower bound for each sector. The BC flux was calculated as the product of the BC concentration and the wind speed, and represents the rate of transport of aerosol BC over the measurement point. Given that McMurdo Station is geographically small and that its local topography does not provide for the buildup of emissions under stagnant wind conditions, the BC flux is approximately proportional to the rate of BC emission at locations over which the air trajectory passes. In this way, as the wind vector sweeps the compass, the plot of flux versus wind direction will map the source strength locations. The results are shown in Figure 6b, which plots the arithmetic mean, geometric mean, and median BC flux against wind direction grouped into 30 sectors. Again, the largest fluxes are associated with easterly to southerly winds from the direction of the station. Figure 6b shows that the average flux of BC from the direction of the station is ~700 ng/m 2 /sec. If we assume that this represents the BC flux from the station, that the aerosol is mixed into a shallow layer 30 m high and 1000 m wide, then the total emissions estimate is ( ) , BC Low-Concentration Mode Thus far, we have concentrated mainly on the influence of the station. The resolution of the available meteorological data is too coarse to stratify the 5-min BC observations to definitively exclude the effects of the station. Because of the widespread nature of the sources (aircraft and surface vehicles) and the potential for recirculation of material from McMurdo, even in clean-air sectors, it is unclear whether such discrimination is possible. Nonetheless, Figure 6a shows that the geometric mean of the averages of 5-min observations in the 30-min periods previous to the wind sector measurements was ~15 ng/m 3. We have assumed that the detection limit for a 5-min average measurement is 5 ng/m 3. Long-term averaging of aethalometer measurements reduces the random effect of instrumental noise and enables determination of BC concentrations as low as 0.1 ng/m 3. 8 It may therefore be possible to reliably quantify concentrations lower than 5 ng/m 3 at Hut Point by averaging individual 5-min observations, even if some of these are below the nominal detection limit. Such averages will be statistically meaningful if they are greater than 2 times their standard errors. There were min average BC concentrations corresponding to wind observations in the sector. Of the 60 average concentrations greater than 5 ng/m 3, 14 were greater than 100 ng/m 3. There were 21 average Volume 51 April 2001 Journal of the Air & Waste Management Association 599

8 concentrations below 5 ng/m 3 that were not statistically significant. There were only two cases where average BC concentrations less than 5 ng/m 3 were statistically significant (2.4 ± 1.1 and 4.8 ± 1.2 ng/m 3, respectively). Because Hut Point is generally downwind of the station, it is nearly impossible to select a sufficient number of consecutive 5-min periods to enable detection of BC concentrations near the Antarctic background of ~1 ng/m 3 without introducing contamination from the station. Obviously, long averaging periods, such as those available in the clean air sector at the South Pole, 8 are required to reliably discern this background. SUMMARY Approximately 22,500 5-min average observations of aerosol BC were acquired over a 3-month period at Hut Point, located 0.6 km downwind of McMurdo Station, Antarctica. This material is a unique tracer for combustion emissions, either local or transported long distances in the background atmosphere. The data clearly showed the impact of combustion emissions from station operations. The distribution of BC concentrations was bimodal. The lower mode was centered at ~20 ng/m 3. This value is higher than summer background BC concentrations (~1 ng/m 3 ) reported for other Antarctic sites, suggesting that Hut Point was almost always contaminated by emissions from the station. The upper mode, centered at ~300 ng/m 3 but ranging to thousands of ng/m 3, represented heavily polluted conditions. While the upper mode accounted for roughly one-third of the samples, its cumulative impact was ~7 times greater than that of the lower-mode samples. Wind sector analysis showed that the high values of BC flux were associated with air masses that passed over the station s buildings and installations before reaching the measurement point. The diurnal cycle of BC concentrations corresponded to the daily work schedule at the station. Fuel consumption and estimated emissions at the source are consistent with BC concentrations measured downwind, although the horizontal extent of the plume cannot be adequately assessed with the available meteorological data. ACKNOWLEDGMENTS The authors would like to thank the staff of Antarctic Support Associates for their assistance with logistical arrangements, and especially Mr. Joe Pettit for his supervision of the aethalometer and his diligence in maintaining the measurement program. We also thank John Joseph, LCDR/ USN, and Matt Young, AGC/USN, of the Naval Support Force, Antarctica, for supplying McMurdo meteorological data. This project was supported by National Science Foundation Grant No. OPP The database is available to the scientific community upon request. REFERENCES 1. Bodhaine, B.A.; Harris, J.M.; Ogren, J.A.; Hofmann, D.J. Aerosol Optical Properties at Mauna Loa Observatory: Long-Range Transport from Kuwait?; Geophys. Res. Lett. 1992, 19, Noone, K.J.; Clarke, A.D. Soot Scavenging Measurements in Arctic Snowfall; Atmos. Environ. 1988, 22, Warren, S.G.; Clarke, A.D. Soot in the Atmosphere and Snow Surface of Antarctica; J. Geophys. Res. 1990, 95, Hansen, A.D.A.; Rosen, H.; Novakov, T. Real-Time Measurement of the Absorption Coefficient of Aerosol Particles; Appl. Opt. 1982, 21, Gundel, L.A.; Dod, R.L.; Rosen, H.; Novakov, T. The Relationship between Optical Attenuation and Black Carbon Concentration for Ambient and Source Particles; Sci. Total Environ. 1984, 36, Hansen, A.D.A.; Rosen, H. The Aethalometer An Instrument for the Real-Time Measurement of Optical Absorption by Aerosol Particles; Sci. Total Environ. 1984, 36, Hansen, A.D.A.; Bodhaine, B.A.; Dutton, E.G.; Schnell, R.C. Aerosol Black Carbon Measurements at the South Pole: Initial Results, ; Geophys. Res. Lett. 1988, 15, Bodhaine, B.A. Aerosol Absorption Measurements at Barrow, Mauna Loa, and the South Pole; J. Geophys. Res. 1995, 100, Lowenthal, D.H.; Chow, J.C.; Mazzera, D.M. PM 10 Source Apportionment at McMurdo Station, Antarctica; Antarctic J. 1997, 32, Mazzera, D.M.; Lowenthal, D.H.; Chow, J.C.; Watson, J.G. PM 10 Measurements at McMurdo Station, Antarctica; Atmos. Environ., in press. 11. Mazzera, D.M.; Lowenthal, D.H.; Chow, J.C.; Watson, J.G. Sources of PM 10 and Sulfate Aerosol at McMurdo Station, Antarctica; Chemosphere, in press. 12. Chow, J.; Watson, J.; Pritchett, L.; Pierson, W.; Frazier, C.; Purcell, R. The DRI Thermal/Optical Carbon Analysis System: Description, Evaluation and Applications in U.S. Air Quality Studies; Atmos. Environ. 1993, 27A, Hansen, A.D.A.; Babich, P.C.; Allen, G.A.; Koutrakis, P. Intercomparison of Methods for the Determination of Aerosol Elemental or Black Carbon in Six Major Urban Environments. Presented at the 93rd Annual Meeting & Exhibition of A&WMA, Salt Lake City, UT, June 19 22, Babich, P.; Davey, M.; Allen, G.A.; Koutrakis, P. Method Comparisons for Particulate Nitrate, Elemental Carbon, and PM 2.5 Mass in Seven U.S. Cities; J. Air & Waste Manage. Assoc. 2000, 50, Wolff, E.W.; Cachier, H. Concentrations and Seasonal Cycle of Black Carbon in Aerosol at a Coastal Antarctic Station; J. Geophys. Res. 1998, 103, Master permit application, prepared by Consoer Townsend & Associates AECOM Technology Corp., Fairfax, VA, for Antarctic Support Associates, Englewood, CO; Parfet, P. Antarctic Support Associates, Englewood, CO. Personal communication, Caterpillar Engine Division. Personal communication, Compilation of Air Pollutant Emission Factors. Vol. II, Mobile Sources, 5th ed.; AP-42; U.S. Environmental Protection Agency, U.S. Government Printing Office: Washington, DC, Kuykendal, W.B. Air Emissions Species Manual. Vol. II, Particulate Matter Species Profiles; EPA-450/ b; U.S. Environmental Protection Agency: Research Triangle Park, NC, Lowenthal, D.H.; Zielinska, B.; Chow, J.C.; Watson, J.G.; Gautam, M.; Ferguson, D.H.; Neuroth, G.R.; Stevens, K.D. Characterization of Heavy-Duty Diesel Vehicle Emissions; Atmos. Environ. 1994, 28, About the Authors Anthony D.A. (Tony) Hansen is a staff scientist in the Engineering Division of the Lawrence Berkeley National Laboratory, 1 Cyclotron Rd., Berkeley, CA He is also the owner of Magee Scientific Co., 1829 Francisco St., Berkeley, CA 94703, a company that manufactures instrumentation for aerosol research. Douglas H. Lowenthal (corresponding author) is an associate research professor at the Desert Research Institute, Division of Atmospheric Sciences, 2215 Raggio Pkwy., Reno, NV Judith C. Chow and John G. Watson are research professors at the Desert Research Institute, Division of Atmospheric Sciences. 600 Journal of the Air & Waste Management Association Volume 51 April 2001