References

1 School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332 USA 2 School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, GA 30332 USA 3 Department of Atmospheric Sciences, University of Washington, Box 351640, Seattle, WA 98195-1640 USA 4 Climate Change Research Center, Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, NH 03824 USA

snow. The effect of fossil-fuel combustion on particulate elemental carbon (EC) is assessed by a combination of ambient measurements (∼1 km from the main camp), a series of snow pits (up to 20 km from Summit Camp), and Gaussian plume modeling. Ambient measurements indicate that the air directly downwind of the research station generators experiences particulate absorption coefficient (closely related to EC) values 10 that are up to a factor of 200 higher than the summer 2006 non-camp-impacted ambient average. Local anthropogenic influence on snow EC content is also evident. The average EC concentration in 1-m snow pits in the "clean air" sector of Summit Camp are a factor of 1.8-2.4 higher than in snow pits located 10 km and 20 km to the north ("downwind") and south ("upwind") of the research site. Gaussian plume modeling performed 15 using meteorological data from years 2003-2006 suggests a strong angular dependence of anthropogenic impact, with highest risk to the northwest of Summit Camp and lowest to the southeast. Along a transect to the southeast (5 degree angle bin), the modeled frequency of significant camp contribution to atmospheric EC (i.e. campproduced EC>2006 summer average EC) at a distance of 0.5 km, 10 km, and 20 km 20 is 1%, 0.2%, and 0.05%, respectively. According to both the snow pit and model results, a distance exceeding 10 km towards the southeast is expected to minimize risk of contamination. These results also suggest that other remote Arctic monitoring stations powered by local fuel combustion may need to account for local air and snow contamination in field sampling design and data interpretation.

Introduction
Since its inception in 1989, the United States National Science Foundation research station at the highest point of the Greenland Ice Sheet (elevation: 3200 m), "Summit Camp", has been an extremely valuable research site. The immense effort placed into providing electricity, communications, and shelter at this remote location have paid 5 off in access to rare field measurements supporting numerous scientific disciplines (e.g. glaciology, atmospheric chemistry, and paleoclimatology). While many field measurements at Summit are naturally immune to post-1989 camp activity at Summit station (e.g. deep ice core studies), numerous research studies involve measurements that may be vulnerable to camp emissions such as atmospheric monitoring or sam-10 pling of shallow snow pits. Impacts on the local environment by the Summit research site likely include a modification in nearby snow accumulation as camp structures alter natural drifting patterns, the introduction of foreign bacteria by visiting researchers and their related refuse, and the contamination of the local environment due to emissions from camp fossil fuel burning (camp generators, heavy equipment, snowmobiles, and 15 aircraft).
The focus of our research team's effort at Summit Camp was to measure carbonaceous particulate matter (organic and elemental carbon) in the air and snow. These species are of interest as markers of natural and anthropogenic emissions reaching the Greenland Ice Sheet (e.g. fossil fuel combustion and biomass burning), both in am-20 bient sampling and as a paleorecord of previous source activity. While carbonaceous particulate species have been measured in several past field studies at Summit, no thorough investigation into the potential contamination from camp fossil fuel combustion has taken place. Although Summit Camp seeks to minimize human impacts on the pristine environment (e.g. sleeping in unheated tents during the summer season), cur-25 rently it is necessary to operate two diesel generators (burning modified Jet A-1 fuel) for electricity at all times, diesel-powered heavy equipment to groom an aircraft "ski-way" and dig snow for water use, and gasoline-powered snowmobiles for dragging heavy EGU loads. To protect the designated "clean air" sector located south of camp, staff and researchers cease vehicular use during northerly winds. However, the camp generators are in continuous use and intermittent (every 2-3 weeks during the spring to summer and every 2-3 months during late-summer to early-spring) supply aircraft arrivals occur regardless of wind direction. As these emitting sources could potentially contaminate 5 our atmospheric sampling for organic and elemental carbon, protective measures were integrated into our atmospheric sampling protocol during the field season (cessation of sampling during air traffic and ongoing sector control at all other times). In addition, given that elemental carbon (EC) is expected to be a stable tracer of local combustion emissions, six snow pits were sampled for EC and Gaussian plume modeling was per-10 formed to better understand the footprint of camp contamination. While this study is focused primarily on carbonaceous species, this research is expected to be applicable to other atmospheric species of interest that may be impacted by camp emissions at Summit, Greenland.

15
Extensive sampling of the air and snow for particulate carbonaceous species took place at a research site located approximately 1 km from Summit, Greenland during the summer of 2006. The field methods are described by Hagler et al. (2007a, b), so the sampling procedures will be only briefly discussed. Atmospheric sampling included near-real-time (minutes to hours) measurement of the aerosol absorption coefficient 20 (σ ap ) using a Particle Soot Absorption Photometer (PSAP). Using a calculated mass absorption coefficient of 24 m 2 g −1 (Hagler et al., 2007b), σ ap was converted to an estimated EC concentration.
In addition to the ongoing atmospheric sampling for σ ap , a sector control system was put in place to flag time periods when wind patterns created a potential contamination EGU bell Scientific Inc., CR200 Datalogger), and two modified power strips that provided the capability to shut off time-integrated atmospheric filter sampling. Under periods of stagnation (wind speed <0.5 m/s) or during northerly winds that may transport camp emissions to our southerly research site, the sector control system would shut off integrated samples and assign a "flag" variable the value of 0 (flag = 1 during "on" periods).

5
The sector control program ran and reacted every 10 minutes (a compromise between the need for a short response time and the desire to minimize the cycling off/on of sampling pumps). The potential impact of Summit camp on snow-phase EC was investigated through a series of six 1-m snow pits that were dug and sampled over a two-week period in 10 the month of July. Two snow pits were co-located in the "clean air" region of Summit, a region to the south of the camp that has additional protection of reduced camp vehicular emissions during northerly winds. The remaining four snow pits were located at 10 and 20 km to the north and south of Summit. Each snow pit was sampled at 20 cm increments (5 total layers) for particulate elemental and organic carbon (Hagler et al.,15 2007a). Duplicates were sampled at two layers in each snow pit.
To better understand the impact of Summit camp activity on the local atmosphere and to interpret our snow pit samples, a Gaussian plume model was applied to estimate the regional footprint of Summit camp contamination. The camp emission rate of EC was estimated by assessing concentration spikes in σ ap that occurred throughout the sum-20 mer, ranging ∼2-30 Mm −1 (compared with the summertime average of 0.15 Mm −1 ). A moderate spike of 14.5 Mm −1 that occurred on 10 July was selected as a "best guess" for its mid-range concentration and the absence of flight traffic on that day, with the source emission rate calculated assuming this was a centerline plume concentration hitting the satellite ambient sampling station located 1 km from camp. The measured 25 σ ap was converted to an EC concentration using a previously calculated mass absorption efficiency coefficient of 24 m 2 g −1 (Hagler et al, 2007b), and a camp emission rate of EC (Q) was back-calculated using the standard Gaussian plume model with ground ACPD 8,2008 Site contamination at an Arctic research station EGU reflection using Eq. (1)-(4). EGU 3 Results and discussion

Absorption coefficient and sector control
Throughout the field campaign at Summit, Greenland in the summer of 2006, the need for a sector control system to protect multi-day integrated samples was readily apparent. Sampling time periods flagged for contamination concern were often associated 5 with brief extreme spikes in the absorption coefficient, reaching up to 30 Mm −1 , a factor of 200 higher than the summertime average of 0.15 Mm −1 (Fig. 1). Additionally, it appears that the sector control wind speed and direction parameters selected were effective, with every major concentration spike coinciding with a "flag/shut-off" time period (Fig. 1). Thus, it is expected that previously reported filter measurements for 10 carbonaceous particulate matter (Hagler et al., 2007b)  EGU constituting 1.6% of the total sampling period (Fig. 1). This demonstrates that a higher precision sector control system would likely cause only minor interruptions in ambient sampling.

Snow pits
While it seems to be feasible to avoid camp contamination in atmospheric samples by 5 adding in a sector control system, there is unfortunately no similar method available to prevent the deposition of camp emissions to surface snow. To minimize snow contamination, camp staff members eliminate all vehicular emissions during northerly winds. However, the camp electricity depends on continuous use of diesel generators. In addition, supply flights to the research site take place independent of wind direction. It 10 is of concern that these emissions may impact measurements made of carbonaceous particulate matter in post-1989 snow.
To evaluate the footprint of Summit Camp emissions on surrounding snow, a series of 1-m snow pits were dug and sampled in the clean air sector of Summit and at distances up to 20 km north and south of camp (Table 1). Assessing the snow pit profiles, the 15 two co-located snow pits near Summit Camp appear to be at a generally higher EC concentration than those located at 10 km or further from camp (Fig. 3). Given the coarseness of sampling (20 cm increments) and the difficulty in precisely collecting identical snow layers across multiple pits, a more clear way to compare the snow pit concentrations is to average over the entire depth sampled. In terms of the average 20 EC concentration per pit, a marked difference between the Summit Camp snow pits and those at remote sites is observed (Fig. 4). The average EC concentration of the Summit Camp pits (0.53 µg kg −1 snow) is a factor of 1.8-2.4 higher than EC levels in snow sampled at 10 and 20 km away from camp. One possible explanation of the higher EC measured near Summit Camp is a difference in snow accumulation rates 25 near and far from the research station. However, past research indicates that this is likely only a minor factor; the snow accumulation rate reported at Summit was nearly identical to that at a location 28 km from Summit (Dibb and Fahnestock, 2004 EGU higher EC loading found closer to Summit suggests that future snow pit sampling for species believed to also have camp sources, or secondarily affected by camp pollution, should be performed at some distance from Summit. As snow pits at 20 km are at a similar EC concentration to those at 10 km, it appears that the footprint of Summit is confined to within 10 km. In addition, the snow pits north of Summit camp (average 5 EC of 0.28 µg kg −1 ) are not substantially higher than those to the south (average EC of 0.26 µg kg −1 ), suggesting that the increased camp activity during southerly winds does not translate to long-distance impacts on snow concentrations. Given the difference in camp vs. distant (10 or 20 km) snow pits, one conclusion is that our reported carbonaceous snow concentrations in the clean air sector of camp 10 (Hagler et al., 2007a, b) may have an upward bias. However, it is difficult at this point to determine whether this is an appropriate conclusion, as the impact of camp on snow in the "clean air" sector is likely dependent on the co-occurrence of precipitation and wind direction from camp, as well as on wind speed and surface snow history. Our atmospheric sampling results indicate that camp contamination is highly variable. As 15 seen in the sector control observations, there are a number of periods flagged for camp contamination that did not experience spikes in σ ap ; and, when spikes did occur, they were not at a consistent concentration (Fig. 1). The observed variability in σ ap during flagged time periods is likely due to a number of factors, including plume dispersion under changing atmospheric conditions, wind direction relative to our ambient monitoring 20 site, and variable source emission rates (e.g. supply aircraft traffic).

Gaussian plume modeling
To further understand the air sampling and snow pit results, we estimated the transport of Summit camp plumes to the surrounding snow using a Gaussian plume model. level. While camp plumes can pose a major contamination threat if directly passing over a sampling area, it is important to keep in mind the relative frequency of camp impact in any one direction. At 10 km in the northwest direction of camp, prevailing winds lead to frequent concentration spikes over years [2003][2004][2005][2006] (Fig. 6a). In the opposite 15 direction, model results show that camp-related concentration spikes are still evident but far fewer in number (Fig. 6b). As snow contamination for particulate species are mainly controlled by the occurrence of wet deposition events (Bergin et al., 1995), not all atmospheric concentration spikes are expected to translate to snow contamination. However, a greater frequency of camp plumes traveling in a certain direction certainly 20 increases the risk of sample contamination.
In order to determine "safe" distances and angles for future field work near Summit, the frequency of major camp plume events (i.e. camp contribution exceeding 2006 average EC concentrations) is modeled over all angles and at distances up to 30 km from camp (Fig. 7). It appears that the highest risk of significant camp impact (3% of the 25 time for a given 5 degree angle bin) occurs at a close proximity to camp (0.5 km) in the northwest to north direction of camp. Meanwhile, the southeast direction receives significant camp impact at 0.5 km only ∼0.5% of the time over a given 5 degree angle. In addition, it is clear that moving further in distance from camp lessens risk at ACPD Introduction

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion EGU all directions from camp. At 10 km and 20 km from camp, the maximum (minimum) frequency of significant camp impact reduces to about 1% (0.2%) and 0.2% (0.05%) of the time, respectively. Although there are a number of assumptions influencing the Gaussian plume model estimates, it is interesting that the model results are in a similar range of the summer 2006 observed frequency of camp-related σ ap spikes (1.6%) at 5 ∼1 km southwest of Summit Camp. In interpreting the model results, it is important to point out that the estimated impact of camp contamination depends not only on the camp emission rate but also the typical ambient concentration of the species of interest. Also, it should be noted that the model does not take into account the increased camp activity during southerly winds, which may lead to more highly concentrated plumes 10 transported northward. In general, it appears that the Gaussian modeling supports the insignificant difference in average EC concentrations between snow pits located at 10 km vs. 20 km and North vs. South. Since the frequency of significant camp contamination is already reduced to <1% of the time per 5 degree angle bin over all directions (Fig. 7), the distance 15 1-m snow pits (equal to ∼1 year of snowfall) likely avoided a major camp plume event. The model also indicates that snow contamination at 1 km distance even in the "clean air sector" south of camp is more likely than any point >10 km away given the more highly concentrated plumes close to camp.

20
In all remote and pristine sampling environments, the impact of research site activities on the local environment needs to be taken into consideration to ensure the accuracy of field measurements. At Summit, Greenland, it appears that camp emissions can greatly impact nearby EC concentrations in the air and snow. Extreme and short-term spikes in the absorption coefficient occurred numerous times throughout the summer 25 of 2006 during time periods that were flagged by a sector control system warning of potential approaching camp plumes. For atmospheric samples, it appears that a sector Introduction

Conclusions
References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion

EGU
control system would be a successful means of avoiding camp combustion-related pollution. While longer-term sampling may have a reduction in sampling time by ∼20% (conservative estimate), loss in sample time for shorter field studies will heavily depend on wind patterns and thus may have a considerably higher or lower fraction of downtime compared to the long-term estimate.

5
In terms of snow sampling, Gaussian plume modeling and snow pit results point to a distance of approximately 10 km towards the SE as a good "rule of thumb" to minimize risk of camp impact (0.2% frequency of significant camp plume events). To translate this result to other species potentially impacted by camp generator emissions (e.g. specific organic molecules, isotopes of carbon or nitrogen, sulfate), one needs to consider the generator emission rate of a particular species relative to its expected ambient background concentration. Given a lower generator emission rate and/or higher background concentration compared to EC, the "safe" distance may be closer to Summit camp; and, the converse would be true given a higher emission rate and/or lower background concentrations.

15
While integrating sector control systems into atmospheric studies and traveling far distances to perform snow sampling can improve the quality of field sampling at Summit, a reduction in camp emissions would be a second (and preferable) means to reduce the anthropogenic footprint at such a remote location. A greater reliance on nonemitting power sources (e.g. wind or solar) may be potential technologies to consider, 20 as well as improved energy efficiency in camp structures and fuel-powered vehicles.
Acknowledgements. The authors are grateful for the field support and detailed information on camp power generation from Veco Polar Resources. We thank the New York Air National Guard 109th Airlift Wing for transport of goods and personnel to Summit, Greenland. Financial support for this research came from the National Science Foundation and graduate fellowships to 25 the author (G. S. W. Hagler) from the American Association of University Women and P.E.O. International. 8,2008 Site contamination at an Arctic research station   ACPD 8,2008 Site contamination at an Arctic research station       . Estimated fraction of time (%) that camp contribution at a specific angle and distance will exceed the 2006 measured average EC (7 ng m -3 ). Fig. 7. Estimated fraction of time (%) that camp contribution at a specific angle and distance will exceed the 2006 measured average EC (7 ng m −3 ).