- Email: [email protected]
Growing season total gaseous mercury (TGM) flux measurements over an Acer rubrum L. stand
Atmospheric Environment 43 (2009) 5953–5961
Contents lists available at ScienceDirect
Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv
Growing season total gaseous mercury (TGM) flux measurements over an Acer rubrum L. stand Jesse O. Bash a, *, David R. Miller b a b
U.S. Environmental Protection Agency, National Exposure Research Laboratory, Research Triangle Park, NC 27711, USA Department of Natural Resources and the Environment, University of Connecticut, Storrs, CT 06269, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 26 May 2009 Received in revised form 12 August 2009 Accepted 13 August 2009
Relaxed eddy accumulation (REA) measurements of the total gaseous mercury (TGM) flux measurements were taken over a deciduous forest predominantly composed of Red Maple (Acer rubrum L.) during the growing season of 2004 and the second half of the growing season of 2005. The magnitudes of the flux estimates were in the range of published results from other micrometeorological mercury fluxes taken above a tall canopy and larger than estimates from flux chambers. The magnitude and direction of the flux were not static during the growing season. There was a significant trend (p < 0.001), from net deposition of TGM in early summer to net evasion in the late summer and early fall before complete senescence. A growing season atmosphere-canopy total mercury (TGM) compensation point during unstable daytime conditions was estimated at background ambient concentrations (1.41 ng m3). The trend in the seasonal net TGM flux indicates that long term dry deposition monitoring is needed to accurately estimate mercury loading over a forest ecosystem. Published by Elsevier Ltd.
Keywords: Biogeochemical cycling Relaxed eddy accumulation Micrometeorological fluxes Mercury dry deposition Natural mercury emissions
1. Introduction Deposition and cycling of mercury in the forest canopy has been identified as an important pathway for mercury accumulation in soils and watersheds (St. Louis et al., 2001; Fay and Gustin, 2007) and a source of natural mercury emissions to the atmosphere (Lindberg et al., 1998; Graydon et al., 2006). Since the first micrometeorological measurements of the mercury fluxes over a forest canopy (Lindberg et al., 1998), the direction and magnitude of the mercury biogeochemical cycle in the forested ecosystems has been an area of active debate (Lee et al., 2000; Demers et al., 2007; Graydon et al., 2006). Several natural mercury emission models have been developed to provide mercury emissions estimates from natural surfaces for the community multiscale air quality (CMAQ) (Byun and Schere, 2006) model (Bash et al., 2004; Gbor et al., 2006) and the SARMAP air quality model (SAQM) (Xu et al., 1999). These models estimate the terrestrial elemental mercury flux from forest canopies as a function of evapotranspiration where mercury is the soil water solution is assumed to be transported via the transpiration stream. The contribution of regional sources to local mercury deposition and concentrations indicate that the atmospheric lifetime of mercury over terrestrial systems may also be shorter than
* Corresponding author. E-mail address: [email protected] (J.O. Bash). 1352-2310/$ – see front matter Published by Elsevier Ltd. doi:10.1016/j.atmosenv.2009.08.008
previously estimated (Keeler et al., 2006), or that alternative sources are needed to explain current levels of mercury concentrations (Sigler and Lee, 2006a). A shorter atmospheric lifetime of mercury over terrestrial systems suggests that there is likely an active atmosphere-terrestrial cycling of mercury. Previous TGM flux measurements indicate that evasion from forest ecosystems may provide contribution to the atmospheric mercury pool equivalent to direct anthropogenic emissions (Lindberg et al., 2007; 1998). Previous micrometeorological TGM flux measurements over forest canopies have only spanned several weeks primarily during daytime hours (Lindberg et al., 1998) and used the modified Bowen ratio (Lindberg et al., 1998). The difficulties in successfully measuring a small scalar gradient over a forest canopy make micrometeorological gradient methods, i.e. modified Bowen ratio, uncertain for extended periods. In the experiment reported here, a relaxed eddy accumulation (REA) system was designed to quantify the total mercury flux over a hardwood forest (Bash and Miller, 2008). The REA technique measures mean concentrations in up- and downdraft air parcels rather than a fixed mean gradient and often results in larger concentration gradients (Bowling et al., 1998). This is beneficial in well mixed areas where a representative gradient must be measured over a vertical distance that is often larger than is feasible or when the precision of the chemical analyzer is limited (Bowling et al., 1998). State-of-the-science techniques were used to investigate the fetch and turbulent averaging periods for the
5954
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
use of micrometeorological fluxes at the study site (Bash and Miller, 2008). The purpose of this study is to report on the dynamics and trends of the net total gaseous mercury (TGM) flux over a forest canopy taken during the growing season using the REA technique. Hourly TGM flux data from the automated REA system was collected from June to November 2004, and again from August to October 2005. The REA system was designed for automated collection of air–surface fluxes over tall vegetation for dry deposition monitoring, for details see Bash and Miller (2008). 2. Methods 2.1. Site description Flux measurements were taken on a 40 m tall micrometeorology instrument tower in a Red Maple (Acer rubrum L.) forest on the University of Connecticut research farm in Coventry Connecticut (Lat. 41 470 3000 N, Long. 72 220 2900 W, 162 m in elevation). The single inlet REA system is comprised of a Tekran model 2537A mercury (Tekran Instrument Corporation, Knoxville, TN, USA) analyzer, a Campbell Scientific CR5000 data logger, a Campbell Scientific CSAT3 sonic anemometer (Campbell Scientific Inc., Logan, UT, USA), and a laptop computer. The REA system was mounted 25 m above the forest floor, 5 m above the forest canopy. In addition to the REA system, a suite of meteorological instruments were colocated measuring incoming solar radiation, temperature, humidity, rainfall, the soil heat flux, latent heat flux, and leaf wetness. 2.2. Sampling protocol The REA system was partially assembled in the laboratory with new Teflon tubing and acid-cleaned Teflon valve bodies and fittings, double bagged and fully assembled on the tower following ultra clean techniques. The mercury analyzer was calibrated using a permeation source biweekly, and often more frequently due to interruption in the power supply from inclement weather. Particle filters were replaced on a weekly basis and analyzed for mercury content. Twice a week the REA system was pressure checked for leaks and a system blank was taken by drawing mercury free air through the up- and downdraft sampling trains. Sampling lines were insulated to prevent condensation but power limitations at the site prohibited the heating of the sampling lines to a constant temperature. The sampling lines and the orientation and level of the sonic anemometer were also inspected twice a week. During the growing season the REA system was taken down from the tower for maintenance approximately every six weeks. Maintenance primarily consisted of replacing the tubing, calibrating the mercury analyzer in the laboratory, and washing the Teflon valve bodies and unions in a 1.8% (V/V) nitric acid bath to avoid biases in the flux measurements that may be due to mercury build up in the system. Precipitation samples were taken at the University of Connecticut experimental agricultural station in Storrs CT located approximately 3 miles East of the research tower. They were analyzed for mercury content following EPA method 1631 (U.S. Environmental Protection Agency, 1999) from June 2004 to September 2006. Samples and blanks were taken on weekly intervals until June 2005, after which samples and blanks were collected on an event bases. Event based precipitation samples were collected following EPA method 1669 (U.S. Environmental Protection Agency, 1996; Landis and Keeler, 1997). Foliar concentrations of mercury were analyzed during the autumn leaf fall. A total of fifty litter fall samples were collected on
a weekly basis during the fall senescence from October 8th through October 22nd, 2004. Samples of recently fallen leaves were collected in 16 collection baskets arranged along north–south and west–east transects centered on the tower. Leaves were not allowed to come into contact with the forest floor. All samples were collected following ‘‘clean hands dirty hands’’ protocols and double bagged (U.S. Environmental Protection Agency, 1996).
2.3. REA measurements The design and operating characteristics of the single inlet, zero dead band REA system are reported in detail in Bash (2006) and Bash and Miller (2008). In general, REA is a conditional sampling technique combining fast response, 10 Hz sampling frequency, vertical anemometry to sense upward and downward air motions, with fast switching, 30 ms maximum response time, of intake air to isolate the air from the upward and downward motions. The mercury vapor carried in the isolated upward and downward moving air is then accumulated in separate sampling lines. Mercury free air is introduced to the up- and downdraft sampling trains when they are not sampling the ambient atmosphere to maintain a constant flow rate (Bash and Miller, 2008). The mercury concentrations in the sampling lines are measured with the available slow response, in this case the Tekran model 2537A, mercury vapor analyzer. The mercury analyzer was configured to sample mercury concentrations at a 5 min interval, although capable of faster rates, to improve the signal to noise ratio of the measurement. The flux is calculated following Businger and Oncley (1990).
þ FHg ¼ bsw CHg CHg
(1)
Where the mercury flux, FHg (ng m2 h1), is the product of the mean mercury concentration difference in the up- and downdrafts, þ (ng m3) respectively. s (m h1) is the CHg (ng m3) and CHg w standard deviation of the vertical wind speed, and b (unitless) is the relaxation coefficient calculated following Bowling et al. (1998). The ambient mercury concentration can be re-constructed from a single inlet, zero dead band REA system from the up- and downdraft concentrations and the fractions of sampling period in which up- and downdrafts were sampled as follows: þ C Hg ¼ CHg
dt dt þ CHg dt þ dt
(2)
where C Hg is the ambient mercury concentration, dt is the duration of the sampling period, and dt þ and dt are the durations of the up- and downdraft sampling periods respectively.
b¼
w0 T 0
sw Tu Td Þ
(3)
where w0 T 0 (K m s1) is the eddy covariance sensible heat flux. w0 (m s1) is the vertical velocity perturbation; T0 (K) is the temperature perturbation; and Tu and Td (K) are the mean temperatures of the up- and downdrafts respectively. b was calculated using three months of sensible heat flux data and found to be 0.474 with an r2 of 0.96 for the site. The measured fluxes were corrected for density perturbations caused by vapor density fluctuations in the air flow through the thermal mass flow meters used in the system (Webb et al., 1980; Pattey et al., 1992; Lee et al., 2000). As with all turbulent flux calculations the REA technique assumes that the wind vector and scalar concentrations are stationary over the flux averaging period. Also horizontal
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
homogeneity in the measurement field, mass conservation, statistical assumptions in the averaging process, and the Reynolds’ postulates are assumed (Foken and Wichura, 1996). The next two sections cover techniques employed to determine conditions in which these assumptions are applicable. However, the 10 min upand downdraft sampling periods were not altered over the measurement campaign. This effectively imposes the assumption that the ambient mercury concentration remains stationary during the 20 min up- and downdraft sampling sequence. This assumption is valid under most convective daytime conditions but likely increases the uncertainty in the flux measurements during transition and nighttime conditions.
2.4. Footprint climatology Horizontal homogeneity required to satisfy the assumptions needed for micrometeorological flux calculations is typically quantified by estimating the flux footprint under a variety of stability regimes (Horst and Weil, 1994; Wesely and Hicks, 2000). This assumption was evaluate using an Eulerian analytic flux footprint model following Amiro (1998) at the measurement site. It was used to estimate the probable sources of the fluxes measured and to quantify when fetch requirements satisfied the assumptions of the REA technique. The footprint was calculated using a year of meteorological data collected in 2004 at the measurement site. During unstable conditions the prevailing wind directions were from the southwest and the north northwest. The majority of the fluxes were estimated to originate from areas within the forest for all wind directions except from approximately due west (Bash and Miller, 2008). Under stable conditions less than half of the TGM flux was estimated to originate from the forest for wind blowing from the west. Thus fluxes measured under these conditions were not included in the analysis as they likely include a composite of sources and sinks that may not be representative of the air–forest flux at this location. The prevailing wind direction was from the southwest, an area of relatively good fetch. The footprint analysis
5955
was used to assess the quality of measured fluxes and filter data from periods of poor fetch.
2.5. Processing of turbulent statistics The micrometeorological assumption of a stationary time series and averaging windows suitable for the calculation of turbulent statistics were quantified through the use of spectral analysis. The variance of the vertical component of the wind speed, s2w, is dependent on stability and scale (Mahrt et al., 2001). The averaging times for the sw (Equation (1)) in the REA system and for eddy covariance heat flux, used in the estimation of the relaxation coefficient, were calculated using multiresolution decomposition (MRD) analyses after Howell and Mahrt (1997) and Vickers and Mahrt (2003). The MRD spectra was used similarly to a Fourier spectral analysis, to determine the spectral gap between fluxes on the turbulent time scale and motions on mesoscale or synoptic scales. However, MRD spectra do not assume periodicity or a stationary time series (Howell and Mahrt, 1997). This helps improve the similarity relationships between scalars essential to the eddy covariance (EC) and REA techniques. MRD spectra and cospectra were calculated following Howell and Mahrt (1997). The spectral gaps calculated using the MRD technique were located between 15 and 60 min for convective daytime conditions and the spectral gap ranged from 5 to 10 min during periods of strong nighttime stability over the forest canopy. The nighttime spectral gap in s2w typically varied less than 30% from the 20 min averaging window indicating that the wind vector can be considered stationary for a 20 min averaging window (Foken and Wichura,1996). Similarity was assumed between s2w and s2Hg because a fast response TGM analyzer is not yet available. In 2004, MRD for unstable periods indicated that a 30 min averaging period for the site captured most of the eddy flux satisfying turbulent transfer assumptions. In the 2005 and 2006 field campaigns, MRD spectra and cospectra were also performed for stable nighttime conditions as well as convective daytime conditions. The MRD spectra indicated that during typical nighttime conditions a 10 min averaging period captured most of the nighttime turbulent transfer.
Fig. 1. Daily mean flux and atmospheric TGM concentrations for the 2004 and 2005 growing seasons the error bars represent inherent errors in REA flux estimates following Bowling et al., 1998.
5956
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
Fig. 2. Monthly hourly median fluxes of TGM for the 2004 growing season, the error bars represent one median absolute deviation.
The planar fit method, following Lee et al. (2004), was used to rotate the scalar and momentum fluxes into the long term mean wind vector and estimate the bias in the vertical wind speed component. The more commonly used natural wind coordinate system forces the cross-wind momentum flux, v0w0, to zero and often produces physically unrealistic results (Finnigan, 2004). Over the rough forest canopy in this study it would be unrealistic to assume v0 w0 ¼ 0. Rotations were conducted on approximately a monthly basis due to leveling of the sonic anemometer and seasonal changes in the canopy structure. 3. Results 3.1. Growing season daily average flux TGM flux measurements were calculated on hourly intervals for the 2004 and the second half of the 2005 growing seasons. The magnitude of the flux measurements were comparable to
micrometeorological fluxes taken by Lindberg et al. (1998) and larger than fluxes measured with chamber techniques (Graydon et al., 2006; Hanson et al., 1995). A significant (Tukey’s test: p < 0.001) linear change from deposition (a negative flux value) to evasion (positive flux value) of the mercury flux was measured approximately a month following leaf out to senescence for both the 2004 and 2005 growing seasons. Fig. 1 shows this trend was similar during both years of data collection despite the climatic differences of the two years. The trend was significant between all months (Tukey’s test: p < 0.05) except from June to July and September to October. Approximately twice the amount of summer precipitation occurred in 2004 (35.4 cm) than in 2005 (18.9). During 2004 the mercury flux was not measured for a month following leaf out. The net growing season flux, estimated using both 2004 and 2005 measurements, to the forest ecosystem was 5.57 2.5 mg m2 if the flux following the month after leaf out is assumed to be equal to the first month of flux measurements. This may be an underestimation of the net growing season
Fig. 3. Monthly hourly median fluxes of TGM for August, September, and October or the 2005 growing season, the error bars represent one absolute median deviation.
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
5957
deposition. If the measured trend in the flux, Fig. 1, is extrapolated to leaf out, approximately the fist week of May at the experiment site, a larger net deposition estimates would be expected for the month following leaf out. 3.2. Diel cycles in the TGM flux The data displays a strong diel pattern in the growing season flux. The absolute magnitude of the TGM flux was generally greatest during daytime periods, when atmospheric instability was common, and generally small at night as expected due to larger atmospheric and physiological conductance during daytime conditions. This diurnal effect was independent of the direction of the net flux as demonstrated in Figs. 2 and 3. The increase in the TGM flux over the growing season and the midseason change from net deposition to net evasion are clearly illustrated for both growing seasons by the monthly median daily TGM fluxes in Figs. 2 and 3. In 2005 the median daily TGM flux for October was lower than that of September. If the forest canopy is assumed to be the source of the TGM evasion, this may be due to the an earlier litter fall at the tower site, approximately 1.5 weeks, in October 2005 than in 2004 due to heavy rain and strong winds associated with a strong fall frontal system. 3.3. Canopy compensation points A canopy compensation point was determined by the relationship between the flux and the concentration (Walker et al., 2006). And in this case the compensation point is defined as the intercept in a least squares linear regression of the ambient TGM concentrations versus the TGM flux. It is important to note, that this compensation point represents the ambient concentration at which the net flux, including contributions from vegetative, soil, and leaf litter components, was zero. Compensation points were determined for daytime conditions where irradiation was greater than 5 W m2 (Fig. 4) and for nighttime conditions (Fig. 5). The scatter in the daytime relationship may be due to the relatively larger contributions of the various sources/sinks of mercury in the forest under unstable conditions versus nighttime conditions where much of the lower canopy and soil is often decoupled from the atmosphere (Kaimal and Finnigan, 1994). Daytime compensation points were estimated for three six-week intervals, Table 1. The daytime compensation point was estimated to increase over the
Fig. 5. Nighttime TGM flux verses atmospheric TGM concentrations for the 2004 growing season.
growing season until approximately a month before senescence however this trend was not significant at p < 0.05. When the canopy began to senesce during the 2004 growing season the net flux was evasive and there was not a significant correlation between the TGM flux and ambient concentrations. The relationship of environmental variables with the TGM flux can elucidate some of the potential mechanisms driving the flux and compensation point. Spearman’s rank coefficient correlations and scatter plots of the mercury flux, concentration and environmental variables during the daytime hours of the three six week intervals are shown in Fig. 6a–c. During the early summer interval, June 18th through July 29th, the mercury flux was negatively correlated with evapotranspiration and variables positively related to stomatal conductance (ambient temperature, incoming solar radiation and leaf wetness; Jarvis, 1976), and positively correlated with the soil heat flux, Fig. 6a. This may indicate that mercury is being taken up by the canopy through stomatal pathways. TGM fluxes during mid and late summer, July 30th through September 9th, were only significantly correlated with the ambient temperature and ambient TGM concentrations, Fig. 6b. The TGM flux was negatively correlated with the leaf wetness measurements (p < 0.05) from the late summer to senescence, September 10th through October 22nd (n ¼ 132), Fig. 6c. These relationships indicate that the diel uptake observed in the early to mid summer could be explained by stomatal uptake of atmospheric TGM and that the observed evasion from late summer to fall does not appear to be stomatally mediated and may be related to photoreduction and evasion of mercury during the drying of the canopy as observed by Graydon et al. (2008).
Table 1 Daytime compensation points and ambient fluxes estimated and measured ambient concentrations for three periods during the 2004 measurement campaign, q25 and q75 are the 25th and 75th quantiles.
Fig. 4. Daytime TGM flux verses atmospheric TGM concentrations for the 2004 growing season.
Time window
Comp. Point, ng m3
Median (q25, q75) TGM fluxes ng m2 h1
Median (q25, q75) TGM concentrations ng m3
June 18th–Jul 29th Jul 30th–Sept 9th Sept 10th–Oct 22nd
0.89 (p < 0.01) 1.79 (p < 0.05) N.S.
79(156,2) 14 (61,80) 71 (17,145)
1.56 (1.42,1.71) 1.68 (1.55,1.90) 1.34 (1.24,1.56)
5958
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
Fig. 6. Scatter plots, Spearman’s rank correlation coefficient, and probabilities of the total gaseous mercury flux and concentration, ambient temperature (Ta), leaf wetness, soil heat flux (G), incoming solar radiation (Rs), Evapotranspiration (E.T.), and sensible heat flux (H) for six-week early summer (a), mid-to late summer (b), and late summer to fall (c) sampling periods.
3.4. Wet deposition and litter fall fluxes 3.4.1. Wet deposition Volume-weighted wet deposition concentrations were greatest in June and July in both 2004 and 2005, in agreement with results reported by VanArsdale et al. (2005). Annual wet deposition measurements of 7.03 mg m2 in 2004 and 6.57 mg m2 in 2005 were collected. There were no mercury deposition network monitoring stations in CT during this time but these results are in general agreement with the interpreted values from mercury deposition network reported for 2004 and 2005. 3.4.2. Foliar mercury concentrations and litter fall deposition The maximum one-sided leaf area index at the site was 5.3 m2 m2 with a dry matter mass of 378 97.6 g m2 measured during the fall senescence. The total seasonal deposition from litter fall was 12.4 3.2 mg m2 measured using an array of 16 litter trap samplers and a total of 50 collected samples. This total mercury in leaf litter fall was nearly double the measured total wet deposition at the experimental farm in 2004 indicating that dry deposition and/or vegetative uptake has enriched the canopy mercury pool. Litter fall represented the largest input of mercury to the measurement site forest floor during the growing season in
agreement with Graydon et al. (2008). The total mercury loading during the growing season reported here may be underestimated as precipitation throughfall was not measured under the canopy and is generally larger than wet deposition in open areas (Graydon et al., 2008). 4. Discussion 4.1. Growing season trends The trend from net deposition in the spring to net evasion in the autumn over the growing season shows that the cycling of mercury in forested natural ecosystems is more seasonal than indicated by short-term deposition measurements. Increases in the compensation point, although not statistically significant, indicate that the trend in the growing season flux may be related to the accumulation of mercury in the vegetation as reported by Rea et al. (2002) or to a capacitance of mercury storage in vegetation from soil sources before it is reduced (Battke et al., 2008). The undercanopy air–soil flux measured at the same time in this forest stand by Sigler and Lee (2006b) did not share the same trend as the above canopy flux; thus we hypothesize that the trend in this data is primarily driven by air–canopy processes.
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
Fig. 6. (continued).
5959
5960
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
Table 2 Soil mercury concentrations, pH, and percent organic matter by depth. Soil horizon and depth
Upland soils (NE of tower)
Transitional soils (near the base of the tower)
Wetland soils (SW and W of the tower)
Leaf litter
Hg (ng g1) pH % Organic matter
68.78 28.62 4.26 84.6%
104.39 34.43 4.18 79.7%
79.96 41.53 4.79 37.6%
O (5 cm)
Hg (ng g1) pH % Organic matter
439.14 296.07 3.81 35.7%
274.72 81.20 4.06 72.8%
193.74 47.50 4.99 37.67%
A (20 cm)
Hg (ng g1) pH % Organic matter
69.72 45.59 4.57 1.9%
42.75 27.27 4.81 8.0%
70.81 39.20 5.25 32.2%
B (50 cm)
Hg (ng g1) pH % Organic matter
4.32 6.81 5.08 0.7%
BDL 4.86 1.5%
BDL 5.51 0.3%
Air–soil flux measurements at the base of the tower using a dynamic flux chamber were positively correlated with the gaseous mercury concentrations in the soil air spaces at 2 cm, and the gradients in the gaseous soil concentrations at this site mirrored the concentrations of mercury in the bulk soil samples collected (Sigler and Lee, 2006b). This indicates that the air–soil flux beneath the forest canopy at this site is driven by mercury bound in the upper layers of the soil where the bulk soil concentration of mercury was the highest, Table 2. The relative contribution of the forest soils in the air–canopy flux of mercury and the cycling of mercury between soil and vegetation reservoirs, via the transpiration stream or in-canopy advection, merits further investigation. The net growing season air–canopy flux estimated by the REA technique cannot fully explain the quantity of mercury accumulated by the canopy foliage. There are several possibilities that can explain this discrepancy. (1) The undercanopy air–ground flux was almost entirely evasive and there could be foliar uptake of mercury from ground sources that would not be measured using above canopy micrometeorological techniques. (2) The REA relied on open path sonic anemometers to accurately measure the direction of the vertical wind vector. Precipitation events interfered with data collection from these sensors and the mercury flux during and immediately following precipitation events remains unresolved. (3) The Tekran 2537A reports concentrations as TGM but it is widely considered to reflect primarily gaseous elemental mercury (Hg0) concentrations due to line losses. Thus this investigation may not have accounted for Hg2þ if there were substantial Hg2þ deposition contributions to the canopy at this site. Steps were taken to minimize wall losses of Hg by utilizing the minimal amount of tubing necessary and insulating the sampling lines, as detailed in Bash and Miller (2008).
5. Conclusions Growing season flux measurements of TGM from a deciduous forest in this study demonstrate that continuous flux monitoring is needed to capture the dynamics and seasonality of the air–canopy mercury flux. A significant trend exists in the net growing season TGM flux evolving from strong deposition just after leaf out to strong evasion in the late summer. This suggests that the air– canopy compensation point increases over the growing season.
Previously deposited mercury in vegetation and soil has been shown to evade under typical environmental conditions (Graydon et al., 2006; Xin et al., 2007), and it is hypothesized that this transition in the direction of the net flux is related to the accumulation of mercury in the tree foliage. Estimates of the compensation point indicate that there may be an increasing trend in the compensation point. Micrometeorological flux estimates are better suited to estimate the magnitude of the net air–surface exchange and not an ideal technique to elucidate trends in compensation points because they measure the net contribution of all canopy and soil sources and sinks. A strong diel cycle exists in the mercury flux throughout the season. The magnitude of the above canopy TGM flux measured using the REA technique was larger than previously recorded TGM flux measurements using flux chamber techniques, such as (Gustin et al., 1999) but similar in magnitude to flux gradient measurements taken over forest canopies by Lindberg et al. (1998). These measurements indicate that there is a strong seasonality in the atmospheric mercury flux at this study site. There is a negative correlation between mercury flux and the atmospheric TGM concentration indicating that a compensation point exists at background concentration levels. The relative contribution of canopy and soil sources and sinks to the overall atmospheric-forest flux needs further quantification and will likely require in situ flux, careful laboratory measurements and modeling to resolve the sources and sinks of mercury in the forest ecosystem. 6. Disclaimer Although this work was reviewed by EPA and approved for publication, it may not necessarily reflect official Agency policy. Mention of commercial products does not constitute endorsement by the Agency. Acknowledgements We greatly appreciated assistance in the field from Luke Simmons and Kate Knight. This research was funded by the Connecticut River Airshed Watershed Consortium. Support was also received from the University of Connecticut, Storrs Agricultural Experiment Station. The authors would like to thank the thoughtful suggestions provided by two anonymous reviewers. References Amiro, B.D., 1998. Footprint climatologies for evapotranspiration in a boreal catchment. Agric. For. Meteorol. 90, 195–201. Bash, J.O., Miller, D.R., 2008. A relaxed eddy accumulation system for measuring surface fluxes of mercury. J. Atmos. Oceanic Technol. 25 (2), 244–257. Bash, J.O., 2006. Ph.D. dissertation. Measurements of Total Mercury Flux Over a Forest Canopy for Model Development. University of Connecticut. Bash, J.O., Miller, D.R., Meyer, T.H., Bresnahan, P.A., 2004. Northeast United States and Southeast Canada natural mercury emissions estimated with a surface emission model. Atmos. Environ. 38, 5683–5692. Battke, F., Ernst, D., Fleischmann, F., Halbach, S., 2008. Phytoreduction and volatilization of mercury by ascorbate in Arabidopsis thaliana, European beech and Norway spruce. Appl. Geochem. 23 (3), 494–502. Bowling, D.R., Turnipseed, A.A., Delany, A.C., Baldocchi, D.D., Greenberg, J.P., Monson, R.K., 1998. The use of relaxed eddy accumulation to measure biosphere-atmosphere exchange of isoprene and other biological trace gases. Oecologia 116, 306–315. Businger, J.A., Oncley, S.P., 1990. Flux measurements by conditional sampling. J. Atmos. Oceanic Technol. 7, 349–352. Byun, D., Schere, K.L., 2006. Review of the governing equations, computational algorithms, and other components of the models-3 community multiscale air quality (CMAQ) modeling system. Appl. Mech. Rev. 59, 51–77. Demers, J.D., Driscoll, C.T., Fahey, T.J., Yavitt, J.B., 2007. Mercury cycling in litter and soil in different forest types in the Adirondack region, New York, USA. Ecol. Appl. 17 (5), 1341–1351.
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961 Fay, L., Gustin, M., 2007. Assessing the influence of different atmospheric and soil mercury concentrations on foliar mercury concentrations in a controlled environment. Water, Air, Soil, Pollut. 181 (1–4), 373–384. Finnigan, J.J., 2004. A re-evaluation of long-term flux measurement techniques. Part II: coordinate systems. Boundary Layer Meteorol. 113 (1), 1–41. Foken, Th., Wichura, B., 1996. Tools for quality assessment of surface-based flux measurements. Agric. For. Meteorol. 78, 83–105. Gbor, P.K., Wen, D., Meng, F., Yang, F., Zhang, B., Sloan, J.J., 2006. Improved model for mercury emission, transport and deposition. Atmos. Environ. 40 (5), 973–983. Graydon, J.A., St. Louis, V.L., Hintelmann, H., Lindberg, S.E., Sandilands, K.A., Rudd, J.W.M., Kelly, C.A., Hall, B.D., Mowat, L.D., 2008. Long-term wet and dry deposition of total and methyl mercury in the remote boreal ecoregion of Canada. Environ. Sci. Technol. 42, 8345–8351. Graydon, J.A., St. Louis, V.L., Lindberg, S.E., Hintelmann, H., Krabbenhoft, D.P., 2006. Investigation of mercury exchange between forest canopy vegetation and the atmosphere using a dynamic chamber. Environ. Sci. Technol. 40, 4680–4688. Gustin, M.S., Lindberg, S.E., Marsik, F., Casimir, A., Ebinghaus, R., Edwards, G., Hubble-Fitzgerald, C., Kemp, R., Kock, H., Leonard, T., London, J., Majewski, M., Montecinos, C., Owens, J., Pilote, M., Poissant, L., Rasmussen, P., Schaedlich, F., Schneeberger, D., Schroeder, W., Sommar, J., Turner, R., Vette, A., Wallschlaeger, D., Xiao, Z., Zhang, H., 1999. Nevada STORMS project: measurement of mercury emissions from naturally enriched surfaces. J. Geophys. Res. 104 (D17), 21831–21844. Hanson, P.J., Lindberg, S.E., Tabberer, T.A., Owens, J.G., Kim, K.-H., 1995. Foliar exchange of mercury vapor: evidence for a compensation point. Water, Air, Soil, Pollut. 80, 611–620. Horst, T.W., Weil, J.C., 1994. How far is enough? the fetch requirements for micrometeorological measurements of surface fluxes. J. Atmos. Oceanic Technol. 11, 1018–1025. Howell, J.F., Mahrt, L., 1997. Multiresolution flux decomposition. Boundary Layer Meteorol. 83, 117–137. Jarvis, P., 1976. The interpretation of the variations in leaf water potential and stomatal conductance found in canopies in the field. Royal Soc. London Proc. Ser. B 273, 593–610. Kaimal, J.C., Finnigan, J.J., 1994. Atmospheric Boundary Layer Flows: Their Structure and Measurement. Oxford University Press, New York. 289pp. Keeler, G.J., Landis, M.S., Norris, G.A., Christianson, E.M., Dvonch, J.T., 2006. Sources of mercury wet deposition in Eastern Ohio, USA. Environ. Sci. Technol. 40, 5874–5881. Landis, M.S., Keeler, G.J., 1997. Critical evaluation of a modified automatic wet-only precipitation collector mercury and trace element determinations. Environ. Sci. Technol. 31, 2610–2615. Lee, X., Massman, W., Law, B. (Eds.), 2004. The Handbook of Micrometeorology: a Guide for Surface Flux Measurements and Analysis. Springer, p. 268. Lee, X., Benoit, G., Hu, X., 2000. Total gaseous mercury concentrations and flux over a coastal saltmarsh vegetation in Connecticut, USA. Atmos. Environ. 34, 4205–4213.
5961
Lindberg, S.E., Bullock, R., Ebinghaus, R., Engstrom, D., Feng, X.B., Fitzgerald, W., Pirrone, N., Prestbo, E., Seigneur, C., 2007. A synthesis of progress and uncertainties in attributing the sources of mercury in deposition. Ambio 36 (1), 19–32. Lindberg, S.E., Hanson, P.J., Meyers, T.P., Kim, K.-H., 1998. Air/surface exchange of mercury vapor over forests – the need for a reassessment of continental biogenic emissions. Atmos. Environ. 32, 895–908. Mahrt, L., Moore, E., Vickers, D., Lensen, N.O., 2001. Dependence of turbulent and mesoscale velocity variances on scale and stability. J. Appl. Meteorol. 40, 628–641. Pattey, E., Deshardins, R.L., Boudreau, F., Rochette, P., 1992. Impact of density fluctuations on flux measurements of trace gases: implications for the relaxed eddy accumulation technique. Boundary Layer Meteorol. 59, 195–203. Rea, A.W., Lindberg, S.E., Scherbatskoy, T., Keeler, G.J., 2002. Mercury accumulation in foliage over time in two northern mixed-hardwood forests. Water, Air, Soil, Pollut. 133, 49–67. Sigler, J.M., Lee, X., 2006a. Recent trends in anthropogenic mercury emissions in the Northeast United States. J. Geophys. Res. 111, D14316. Sigler, J.M., Lee, X., 2006b. Gaseous mercury in background forest soil in the northeastern United States. J. Geophys. Res. 111, G072007. St. Louis, V.L., Rudd, J.W.M., Kelly, C.A., Hall, B.D., Rolfhus, K.R., Scott, K.J., Lindberg, S.E., Dong, W., 2001. Importance of the forest canopy to fluxes of methylmercury and total mercury to boreal ecosystems. Environ. Sci. Technol. 35 (15), 3089–3098. U.S. Environmental Protection Agency (EPA), 1996. Method 1669. Sampling Ambient Water for Trace Metals at EPA Water Quality Criteria Levels. EPA. EPA-821-R-96-011. U.S. Environmental Protection Agency (EPA), 1999. Method 1631, Rrevision B: Mercury in Water by Oxidation, Purge, and Trap, and Cold Vapor Atomic Fluorescence Spectrometry. Office of Water, EPA. EPA-821-R-99-005. VanArsdale, A., Weiss, J., Keeler, G., Miller, E., Boulet, G., Brulotte, R., Poissant, L., 2005. Patters of mercury deposition and concentrations in northeastern North America (1996–2002). Ecotoxicology 14 (1–2), 37–52. Vickers, D., Mahrt, L., 2003. The cospectral gap and turbulent flux calculations. J. Atmos. Oceanic Technol. 20 (5), 660–672. Walker, J.T., Robarge, W.P., Wu, Y., Meyers, T.P., 2006. Measurement of bi-directional ammonia fluxes over soybean using modified Bowen-ratio technique. Agric. For. Meteorol. 138, 54–68. Webb, E.K., Pearman, G.I., Leuning, R., 1980. Correction of flux measurements for density effects due to heat and water vapor transfer. Q.J.R. Meteorol. Soc. 106, 85–100. Wesely, M.L., Hicks, B.B., 2000. A review of the current status of knowledge in dry deposition. Atmos. Environ. 34, 2261–2282. Xin, M., Gustin, M., Johnson, D., 2007. Laboratory investigation of the potential for re-emission of atmospherically derived Hg from soils. Environ. Sci. Technol. 41 (14), 4946–4951. Xu, X., Yang, X., Miller, D., Helble, J.J., Carley, R.J., 1999. Formulation of bi-directional atmosphere-surface exchanges of elemental mercury. Atmos. Environ. 33 (27), 4345–4355.
Contents lists available at ScienceDirect
Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv
Growing season total gaseous mercury (TGM) flux measurements over an Acer rubrum L. stand Jesse O. Bash a, *, David R. Miller b a b
U.S. Environmental Protection Agency, National Exposure Research Laboratory, Research Triangle Park, NC 27711, USA Department of Natural Resources and the Environment, University of Connecticut, Storrs, CT 06269, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 26 May 2009 Received in revised form 12 August 2009 Accepted 13 August 2009
Relaxed eddy accumulation (REA) measurements of the total gaseous mercury (TGM) flux measurements were taken over a deciduous forest predominantly composed of Red Maple (Acer rubrum L.) during the growing season of 2004 and the second half of the growing season of 2005. The magnitudes of the flux estimates were in the range of published results from other micrometeorological mercury fluxes taken above a tall canopy and larger than estimates from flux chambers. The magnitude and direction of the flux were not static during the growing season. There was a significant trend (p < 0.001), from net deposition of TGM in early summer to net evasion in the late summer and early fall before complete senescence. A growing season atmosphere-canopy total mercury (TGM) compensation point during unstable daytime conditions was estimated at background ambient concentrations (1.41 ng m3). The trend in the seasonal net TGM flux indicates that long term dry deposition monitoring is needed to accurately estimate mercury loading over a forest ecosystem. Published by Elsevier Ltd.
Keywords: Biogeochemical cycling Relaxed eddy accumulation Micrometeorological fluxes Mercury dry deposition Natural mercury emissions
1. Introduction Deposition and cycling of mercury in the forest canopy has been identified as an important pathway for mercury accumulation in soils and watersheds (St. Louis et al., 2001; Fay and Gustin, 2007) and a source of natural mercury emissions to the atmosphere (Lindberg et al., 1998; Graydon et al., 2006). Since the first micrometeorological measurements of the mercury fluxes over a forest canopy (Lindberg et al., 1998), the direction and magnitude of the mercury biogeochemical cycle in the forested ecosystems has been an area of active debate (Lee et al., 2000; Demers et al., 2007; Graydon et al., 2006). Several natural mercury emission models have been developed to provide mercury emissions estimates from natural surfaces for the community multiscale air quality (CMAQ) (Byun and Schere, 2006) model (Bash et al., 2004; Gbor et al., 2006) and the SARMAP air quality model (SAQM) (Xu et al., 1999). These models estimate the terrestrial elemental mercury flux from forest canopies as a function of evapotranspiration where mercury is the soil water solution is assumed to be transported via the transpiration stream. The contribution of regional sources to local mercury deposition and concentrations indicate that the atmospheric lifetime of mercury over terrestrial systems may also be shorter than
* Corresponding author. E-mail address: [email protected] (J.O. Bash). 1352-2310/$ – see front matter Published by Elsevier Ltd. doi:10.1016/j.atmosenv.2009.08.008
previously estimated (Keeler et al., 2006), or that alternative sources are needed to explain current levels of mercury concentrations (Sigler and Lee, 2006a). A shorter atmospheric lifetime of mercury over terrestrial systems suggests that there is likely an active atmosphere-terrestrial cycling of mercury. Previous TGM flux measurements indicate that evasion from forest ecosystems may provide contribution to the atmospheric mercury pool equivalent to direct anthropogenic emissions (Lindberg et al., 2007; 1998). Previous micrometeorological TGM flux measurements over forest canopies have only spanned several weeks primarily during daytime hours (Lindberg et al., 1998) and used the modified Bowen ratio (Lindberg et al., 1998). The difficulties in successfully measuring a small scalar gradient over a forest canopy make micrometeorological gradient methods, i.e. modified Bowen ratio, uncertain for extended periods. In the experiment reported here, a relaxed eddy accumulation (REA) system was designed to quantify the total mercury flux over a hardwood forest (Bash and Miller, 2008). The REA technique measures mean concentrations in up- and downdraft air parcels rather than a fixed mean gradient and often results in larger concentration gradients (Bowling et al., 1998). This is beneficial in well mixed areas where a representative gradient must be measured over a vertical distance that is often larger than is feasible or when the precision of the chemical analyzer is limited (Bowling et al., 1998). State-of-the-science techniques were used to investigate the fetch and turbulent averaging periods for the
5954
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
use of micrometeorological fluxes at the study site (Bash and Miller, 2008). The purpose of this study is to report on the dynamics and trends of the net total gaseous mercury (TGM) flux over a forest canopy taken during the growing season using the REA technique. Hourly TGM flux data from the automated REA system was collected from June to November 2004, and again from August to October 2005. The REA system was designed for automated collection of air–surface fluxes over tall vegetation for dry deposition monitoring, for details see Bash and Miller (2008). 2. Methods 2.1. Site description Flux measurements were taken on a 40 m tall micrometeorology instrument tower in a Red Maple (Acer rubrum L.) forest on the University of Connecticut research farm in Coventry Connecticut (Lat. 41 470 3000 N, Long. 72 220 2900 W, 162 m in elevation). The single inlet REA system is comprised of a Tekran model 2537A mercury (Tekran Instrument Corporation, Knoxville, TN, USA) analyzer, a Campbell Scientific CR5000 data logger, a Campbell Scientific CSAT3 sonic anemometer (Campbell Scientific Inc., Logan, UT, USA), and a laptop computer. The REA system was mounted 25 m above the forest floor, 5 m above the forest canopy. In addition to the REA system, a suite of meteorological instruments were colocated measuring incoming solar radiation, temperature, humidity, rainfall, the soil heat flux, latent heat flux, and leaf wetness. 2.2. Sampling protocol The REA system was partially assembled in the laboratory with new Teflon tubing and acid-cleaned Teflon valve bodies and fittings, double bagged and fully assembled on the tower following ultra clean techniques. The mercury analyzer was calibrated using a permeation source biweekly, and often more frequently due to interruption in the power supply from inclement weather. Particle filters were replaced on a weekly basis and analyzed for mercury content. Twice a week the REA system was pressure checked for leaks and a system blank was taken by drawing mercury free air through the up- and downdraft sampling trains. Sampling lines were insulated to prevent condensation but power limitations at the site prohibited the heating of the sampling lines to a constant temperature. The sampling lines and the orientation and level of the sonic anemometer were also inspected twice a week. During the growing season the REA system was taken down from the tower for maintenance approximately every six weeks. Maintenance primarily consisted of replacing the tubing, calibrating the mercury analyzer in the laboratory, and washing the Teflon valve bodies and unions in a 1.8% (V/V) nitric acid bath to avoid biases in the flux measurements that may be due to mercury build up in the system. Precipitation samples were taken at the University of Connecticut experimental agricultural station in Storrs CT located approximately 3 miles East of the research tower. They were analyzed for mercury content following EPA method 1631 (U.S. Environmental Protection Agency, 1999) from June 2004 to September 2006. Samples and blanks were taken on weekly intervals until June 2005, after which samples and blanks were collected on an event bases. Event based precipitation samples were collected following EPA method 1669 (U.S. Environmental Protection Agency, 1996; Landis and Keeler, 1997). Foliar concentrations of mercury were analyzed during the autumn leaf fall. A total of fifty litter fall samples were collected on
a weekly basis during the fall senescence from October 8th through October 22nd, 2004. Samples of recently fallen leaves were collected in 16 collection baskets arranged along north–south and west–east transects centered on the tower. Leaves were not allowed to come into contact with the forest floor. All samples were collected following ‘‘clean hands dirty hands’’ protocols and double bagged (U.S. Environmental Protection Agency, 1996).
2.3. REA measurements The design and operating characteristics of the single inlet, zero dead band REA system are reported in detail in Bash (2006) and Bash and Miller (2008). In general, REA is a conditional sampling technique combining fast response, 10 Hz sampling frequency, vertical anemometry to sense upward and downward air motions, with fast switching, 30 ms maximum response time, of intake air to isolate the air from the upward and downward motions. The mercury vapor carried in the isolated upward and downward moving air is then accumulated in separate sampling lines. Mercury free air is introduced to the up- and downdraft sampling trains when they are not sampling the ambient atmosphere to maintain a constant flow rate (Bash and Miller, 2008). The mercury concentrations in the sampling lines are measured with the available slow response, in this case the Tekran model 2537A, mercury vapor analyzer. The mercury analyzer was configured to sample mercury concentrations at a 5 min interval, although capable of faster rates, to improve the signal to noise ratio of the measurement. The flux is calculated following Businger and Oncley (1990).
þ FHg ¼ bsw CHg CHg
(1)
Where the mercury flux, FHg (ng m2 h1), is the product of the mean mercury concentration difference in the up- and downdrafts, þ (ng m3) respectively. s (m h1) is the CHg (ng m3) and CHg w standard deviation of the vertical wind speed, and b (unitless) is the relaxation coefficient calculated following Bowling et al. (1998). The ambient mercury concentration can be re-constructed from a single inlet, zero dead band REA system from the up- and downdraft concentrations and the fractions of sampling period in which up- and downdrafts were sampled as follows: þ C Hg ¼ CHg
dt dt þ CHg dt þ dt
(2)
where C Hg is the ambient mercury concentration, dt is the duration of the sampling period, and dt þ and dt are the durations of the up- and downdraft sampling periods respectively.
b¼
w0 T 0
sw Tu Td Þ
(3)
where w0 T 0 (K m s1) is the eddy covariance sensible heat flux. w0 (m s1) is the vertical velocity perturbation; T0 (K) is the temperature perturbation; and Tu and Td (K) are the mean temperatures of the up- and downdrafts respectively. b was calculated using three months of sensible heat flux data and found to be 0.474 with an r2 of 0.96 for the site. The measured fluxes were corrected for density perturbations caused by vapor density fluctuations in the air flow through the thermal mass flow meters used in the system (Webb et al., 1980; Pattey et al., 1992; Lee et al., 2000). As with all turbulent flux calculations the REA technique assumes that the wind vector and scalar concentrations are stationary over the flux averaging period. Also horizontal
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
homogeneity in the measurement field, mass conservation, statistical assumptions in the averaging process, and the Reynolds’ postulates are assumed (Foken and Wichura, 1996). The next two sections cover techniques employed to determine conditions in which these assumptions are applicable. However, the 10 min upand downdraft sampling periods were not altered over the measurement campaign. This effectively imposes the assumption that the ambient mercury concentration remains stationary during the 20 min up- and downdraft sampling sequence. This assumption is valid under most convective daytime conditions but likely increases the uncertainty in the flux measurements during transition and nighttime conditions.
2.4. Footprint climatology Horizontal homogeneity required to satisfy the assumptions needed for micrometeorological flux calculations is typically quantified by estimating the flux footprint under a variety of stability regimes (Horst and Weil, 1994; Wesely and Hicks, 2000). This assumption was evaluate using an Eulerian analytic flux footprint model following Amiro (1998) at the measurement site. It was used to estimate the probable sources of the fluxes measured and to quantify when fetch requirements satisfied the assumptions of the REA technique. The footprint was calculated using a year of meteorological data collected in 2004 at the measurement site. During unstable conditions the prevailing wind directions were from the southwest and the north northwest. The majority of the fluxes were estimated to originate from areas within the forest for all wind directions except from approximately due west (Bash and Miller, 2008). Under stable conditions less than half of the TGM flux was estimated to originate from the forest for wind blowing from the west. Thus fluxes measured under these conditions were not included in the analysis as they likely include a composite of sources and sinks that may not be representative of the air–forest flux at this location. The prevailing wind direction was from the southwest, an area of relatively good fetch. The footprint analysis
5955
was used to assess the quality of measured fluxes and filter data from periods of poor fetch.
2.5. Processing of turbulent statistics The micrometeorological assumption of a stationary time series and averaging windows suitable for the calculation of turbulent statistics were quantified through the use of spectral analysis. The variance of the vertical component of the wind speed, s2w, is dependent on stability and scale (Mahrt et al., 2001). The averaging times for the sw (Equation (1)) in the REA system and for eddy covariance heat flux, used in the estimation of the relaxation coefficient, were calculated using multiresolution decomposition (MRD) analyses after Howell and Mahrt (1997) and Vickers and Mahrt (2003). The MRD spectra was used similarly to a Fourier spectral analysis, to determine the spectral gap between fluxes on the turbulent time scale and motions on mesoscale or synoptic scales. However, MRD spectra do not assume periodicity or a stationary time series (Howell and Mahrt, 1997). This helps improve the similarity relationships between scalars essential to the eddy covariance (EC) and REA techniques. MRD spectra and cospectra were calculated following Howell and Mahrt (1997). The spectral gaps calculated using the MRD technique were located between 15 and 60 min for convective daytime conditions and the spectral gap ranged from 5 to 10 min during periods of strong nighttime stability over the forest canopy. The nighttime spectral gap in s2w typically varied less than 30% from the 20 min averaging window indicating that the wind vector can be considered stationary for a 20 min averaging window (Foken and Wichura,1996). Similarity was assumed between s2w and s2Hg because a fast response TGM analyzer is not yet available. In 2004, MRD for unstable periods indicated that a 30 min averaging period for the site captured most of the eddy flux satisfying turbulent transfer assumptions. In the 2005 and 2006 field campaigns, MRD spectra and cospectra were also performed for stable nighttime conditions as well as convective daytime conditions. The MRD spectra indicated that during typical nighttime conditions a 10 min averaging period captured most of the nighttime turbulent transfer.
Fig. 1. Daily mean flux and atmospheric TGM concentrations for the 2004 and 2005 growing seasons the error bars represent inherent errors in REA flux estimates following Bowling et al., 1998.
5956
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
Fig. 2. Monthly hourly median fluxes of TGM for the 2004 growing season, the error bars represent one median absolute deviation.
The planar fit method, following Lee et al. (2004), was used to rotate the scalar and momentum fluxes into the long term mean wind vector and estimate the bias in the vertical wind speed component. The more commonly used natural wind coordinate system forces the cross-wind momentum flux, v0w0, to zero and often produces physically unrealistic results (Finnigan, 2004). Over the rough forest canopy in this study it would be unrealistic to assume v0 w0 ¼ 0. Rotations were conducted on approximately a monthly basis due to leveling of the sonic anemometer and seasonal changes in the canopy structure. 3. Results 3.1. Growing season daily average flux TGM flux measurements were calculated on hourly intervals for the 2004 and the second half of the 2005 growing seasons. The magnitude of the flux measurements were comparable to
micrometeorological fluxes taken by Lindberg et al. (1998) and larger than fluxes measured with chamber techniques (Graydon et al., 2006; Hanson et al., 1995). A significant (Tukey’s test: p < 0.001) linear change from deposition (a negative flux value) to evasion (positive flux value) of the mercury flux was measured approximately a month following leaf out to senescence for both the 2004 and 2005 growing seasons. Fig. 1 shows this trend was similar during both years of data collection despite the climatic differences of the two years. The trend was significant between all months (Tukey’s test: p < 0.05) except from June to July and September to October. Approximately twice the amount of summer precipitation occurred in 2004 (35.4 cm) than in 2005 (18.9). During 2004 the mercury flux was not measured for a month following leaf out. The net growing season flux, estimated using both 2004 and 2005 measurements, to the forest ecosystem was 5.57 2.5 mg m2 if the flux following the month after leaf out is assumed to be equal to the first month of flux measurements. This may be an underestimation of the net growing season
Fig. 3. Monthly hourly median fluxes of TGM for August, September, and October or the 2005 growing season, the error bars represent one absolute median deviation.
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
5957
deposition. If the measured trend in the flux, Fig. 1, is extrapolated to leaf out, approximately the fist week of May at the experiment site, a larger net deposition estimates would be expected for the month following leaf out. 3.2. Diel cycles in the TGM flux The data displays a strong diel pattern in the growing season flux. The absolute magnitude of the TGM flux was generally greatest during daytime periods, when atmospheric instability was common, and generally small at night as expected due to larger atmospheric and physiological conductance during daytime conditions. This diurnal effect was independent of the direction of the net flux as demonstrated in Figs. 2 and 3. The increase in the TGM flux over the growing season and the midseason change from net deposition to net evasion are clearly illustrated for both growing seasons by the monthly median daily TGM fluxes in Figs. 2 and 3. In 2005 the median daily TGM flux for October was lower than that of September. If the forest canopy is assumed to be the source of the TGM evasion, this may be due to the an earlier litter fall at the tower site, approximately 1.5 weeks, in October 2005 than in 2004 due to heavy rain and strong winds associated with a strong fall frontal system. 3.3. Canopy compensation points A canopy compensation point was determined by the relationship between the flux and the concentration (Walker et al., 2006). And in this case the compensation point is defined as the intercept in a least squares linear regression of the ambient TGM concentrations versus the TGM flux. It is important to note, that this compensation point represents the ambient concentration at which the net flux, including contributions from vegetative, soil, and leaf litter components, was zero. Compensation points were determined for daytime conditions where irradiation was greater than 5 W m2 (Fig. 4) and for nighttime conditions (Fig. 5). The scatter in the daytime relationship may be due to the relatively larger contributions of the various sources/sinks of mercury in the forest under unstable conditions versus nighttime conditions where much of the lower canopy and soil is often decoupled from the atmosphere (Kaimal and Finnigan, 1994). Daytime compensation points were estimated for three six-week intervals, Table 1. The daytime compensation point was estimated to increase over the
Fig. 5. Nighttime TGM flux verses atmospheric TGM concentrations for the 2004 growing season.
growing season until approximately a month before senescence however this trend was not significant at p < 0.05. When the canopy began to senesce during the 2004 growing season the net flux was evasive and there was not a significant correlation between the TGM flux and ambient concentrations. The relationship of environmental variables with the TGM flux can elucidate some of the potential mechanisms driving the flux and compensation point. Spearman’s rank coefficient correlations and scatter plots of the mercury flux, concentration and environmental variables during the daytime hours of the three six week intervals are shown in Fig. 6a–c. During the early summer interval, June 18th through July 29th, the mercury flux was negatively correlated with evapotranspiration and variables positively related to stomatal conductance (ambient temperature, incoming solar radiation and leaf wetness; Jarvis, 1976), and positively correlated with the soil heat flux, Fig. 6a. This may indicate that mercury is being taken up by the canopy through stomatal pathways. TGM fluxes during mid and late summer, July 30th through September 9th, were only significantly correlated with the ambient temperature and ambient TGM concentrations, Fig. 6b. The TGM flux was negatively correlated with the leaf wetness measurements (p < 0.05) from the late summer to senescence, September 10th through October 22nd (n ¼ 132), Fig. 6c. These relationships indicate that the diel uptake observed in the early to mid summer could be explained by stomatal uptake of atmospheric TGM and that the observed evasion from late summer to fall does not appear to be stomatally mediated and may be related to photoreduction and evasion of mercury during the drying of the canopy as observed by Graydon et al. (2008).
Table 1 Daytime compensation points and ambient fluxes estimated and measured ambient concentrations for three periods during the 2004 measurement campaign, q25 and q75 are the 25th and 75th quantiles.
Fig. 4. Daytime TGM flux verses atmospheric TGM concentrations for the 2004 growing season.
Time window
Comp. Point, ng m3
Median (q25, q75) TGM fluxes ng m2 h1
Median (q25, q75) TGM concentrations ng m3
June 18th–Jul 29th Jul 30th–Sept 9th Sept 10th–Oct 22nd
0.89 (p < 0.01) 1.79 (p < 0.05) N.S.
79(156,2) 14 (61,80) 71 (17,145)
1.56 (1.42,1.71) 1.68 (1.55,1.90) 1.34 (1.24,1.56)
5958
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
Fig. 6. Scatter plots, Spearman’s rank correlation coefficient, and probabilities of the total gaseous mercury flux and concentration, ambient temperature (Ta), leaf wetness, soil heat flux (G), incoming solar radiation (Rs), Evapotranspiration (E.T.), and sensible heat flux (H) for six-week early summer (a), mid-to late summer (b), and late summer to fall (c) sampling periods.
3.4. Wet deposition and litter fall fluxes 3.4.1. Wet deposition Volume-weighted wet deposition concentrations were greatest in June and July in both 2004 and 2005, in agreement with results reported by VanArsdale et al. (2005). Annual wet deposition measurements of 7.03 mg m2 in 2004 and 6.57 mg m2 in 2005 were collected. There were no mercury deposition network monitoring stations in CT during this time but these results are in general agreement with the interpreted values from mercury deposition network reported for 2004 and 2005. 3.4.2. Foliar mercury concentrations and litter fall deposition The maximum one-sided leaf area index at the site was 5.3 m2 m2 with a dry matter mass of 378 97.6 g m2 measured during the fall senescence. The total seasonal deposition from litter fall was 12.4 3.2 mg m2 measured using an array of 16 litter trap samplers and a total of 50 collected samples. This total mercury in leaf litter fall was nearly double the measured total wet deposition at the experimental farm in 2004 indicating that dry deposition and/or vegetative uptake has enriched the canopy mercury pool. Litter fall represented the largest input of mercury to the measurement site forest floor during the growing season in
agreement with Graydon et al. (2008). The total mercury loading during the growing season reported here may be underestimated as precipitation throughfall was not measured under the canopy and is generally larger than wet deposition in open areas (Graydon et al., 2008). 4. Discussion 4.1. Growing season trends The trend from net deposition in the spring to net evasion in the autumn over the growing season shows that the cycling of mercury in forested natural ecosystems is more seasonal than indicated by short-term deposition measurements. Increases in the compensation point, although not statistically significant, indicate that the trend in the growing season flux may be related to the accumulation of mercury in the vegetation as reported by Rea et al. (2002) or to a capacitance of mercury storage in vegetation from soil sources before it is reduced (Battke et al., 2008). The undercanopy air–soil flux measured at the same time in this forest stand by Sigler and Lee (2006b) did not share the same trend as the above canopy flux; thus we hypothesize that the trend in this data is primarily driven by air–canopy processes.
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
Fig. 6. (continued).
5959
5960
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961
Table 2 Soil mercury concentrations, pH, and percent organic matter by depth. Soil horizon and depth
Upland soils (NE of tower)
Transitional soils (near the base of the tower)
Wetland soils (SW and W of the tower)
Leaf litter
Hg (ng g1) pH % Organic matter
68.78 28.62 4.26 84.6%
104.39 34.43 4.18 79.7%
79.96 41.53 4.79 37.6%
O (5 cm)
Hg (ng g1) pH % Organic matter
439.14 296.07 3.81 35.7%
274.72 81.20 4.06 72.8%
193.74 47.50 4.99 37.67%
A (20 cm)
Hg (ng g1) pH % Organic matter
69.72 45.59 4.57 1.9%
42.75 27.27 4.81 8.0%
70.81 39.20 5.25 32.2%
B (50 cm)
Hg (ng g1) pH % Organic matter
4.32 6.81 5.08 0.7%
BDL 4.86 1.5%
BDL 5.51 0.3%
Air–soil flux measurements at the base of the tower using a dynamic flux chamber were positively correlated with the gaseous mercury concentrations in the soil air spaces at 2 cm, and the gradients in the gaseous soil concentrations at this site mirrored the concentrations of mercury in the bulk soil samples collected (Sigler and Lee, 2006b). This indicates that the air–soil flux beneath the forest canopy at this site is driven by mercury bound in the upper layers of the soil where the bulk soil concentration of mercury was the highest, Table 2. The relative contribution of the forest soils in the air–canopy flux of mercury and the cycling of mercury between soil and vegetation reservoirs, via the transpiration stream or in-canopy advection, merits further investigation. The net growing season air–canopy flux estimated by the REA technique cannot fully explain the quantity of mercury accumulated by the canopy foliage. There are several possibilities that can explain this discrepancy. (1) The undercanopy air–ground flux was almost entirely evasive and there could be foliar uptake of mercury from ground sources that would not be measured using above canopy micrometeorological techniques. (2) The REA relied on open path sonic anemometers to accurately measure the direction of the vertical wind vector. Precipitation events interfered with data collection from these sensors and the mercury flux during and immediately following precipitation events remains unresolved. (3) The Tekran 2537A reports concentrations as TGM but it is widely considered to reflect primarily gaseous elemental mercury (Hg0) concentrations due to line losses. Thus this investigation may not have accounted for Hg2þ if there were substantial Hg2þ deposition contributions to the canopy at this site. Steps were taken to minimize wall losses of Hg by utilizing the minimal amount of tubing necessary and insulating the sampling lines, as detailed in Bash and Miller (2008).
5. Conclusions Growing season flux measurements of TGM from a deciduous forest in this study demonstrate that continuous flux monitoring is needed to capture the dynamics and seasonality of the air–canopy mercury flux. A significant trend exists in the net growing season TGM flux evolving from strong deposition just after leaf out to strong evasion in the late summer. This suggests that the air– canopy compensation point increases over the growing season.
Previously deposited mercury in vegetation and soil has been shown to evade under typical environmental conditions (Graydon et al., 2006; Xin et al., 2007), and it is hypothesized that this transition in the direction of the net flux is related to the accumulation of mercury in the tree foliage. Estimates of the compensation point indicate that there may be an increasing trend in the compensation point. Micrometeorological flux estimates are better suited to estimate the magnitude of the net air–surface exchange and not an ideal technique to elucidate trends in compensation points because they measure the net contribution of all canopy and soil sources and sinks. A strong diel cycle exists in the mercury flux throughout the season. The magnitude of the above canopy TGM flux measured using the REA technique was larger than previously recorded TGM flux measurements using flux chamber techniques, such as (Gustin et al., 1999) but similar in magnitude to flux gradient measurements taken over forest canopies by Lindberg et al. (1998). These measurements indicate that there is a strong seasonality in the atmospheric mercury flux at this study site. There is a negative correlation between mercury flux and the atmospheric TGM concentration indicating that a compensation point exists at background concentration levels. The relative contribution of canopy and soil sources and sinks to the overall atmospheric-forest flux needs further quantification and will likely require in situ flux, careful laboratory measurements and modeling to resolve the sources and sinks of mercury in the forest ecosystem. 6. Disclaimer Although this work was reviewed by EPA and approved for publication, it may not necessarily reflect official Agency policy. Mention of commercial products does not constitute endorsement by the Agency. Acknowledgements We greatly appreciated assistance in the field from Luke Simmons and Kate Knight. This research was funded by the Connecticut River Airshed Watershed Consortium. Support was also received from the University of Connecticut, Storrs Agricultural Experiment Station. The authors would like to thank the thoughtful suggestions provided by two anonymous reviewers. References Amiro, B.D., 1998. Footprint climatologies for evapotranspiration in a boreal catchment. Agric. For. Meteorol. 90, 195–201. Bash, J.O., Miller, D.R., 2008. A relaxed eddy accumulation system for measuring surface fluxes of mercury. J. Atmos. Oceanic Technol. 25 (2), 244–257. Bash, J.O., 2006. Ph.D. dissertation. Measurements of Total Mercury Flux Over a Forest Canopy for Model Development. University of Connecticut. Bash, J.O., Miller, D.R., Meyer, T.H., Bresnahan, P.A., 2004. Northeast United States and Southeast Canada natural mercury emissions estimated with a surface emission model. Atmos. Environ. 38, 5683–5692. Battke, F., Ernst, D., Fleischmann, F., Halbach, S., 2008. Phytoreduction and volatilization of mercury by ascorbate in Arabidopsis thaliana, European beech and Norway spruce. Appl. Geochem. 23 (3), 494–502. Bowling, D.R., Turnipseed, A.A., Delany, A.C., Baldocchi, D.D., Greenberg, J.P., Monson, R.K., 1998. The use of relaxed eddy accumulation to measure biosphere-atmosphere exchange of isoprene and other biological trace gases. Oecologia 116, 306–315. Businger, J.A., Oncley, S.P., 1990. Flux measurements by conditional sampling. J. Atmos. Oceanic Technol. 7, 349–352. Byun, D., Schere, K.L., 2006. Review of the governing equations, computational algorithms, and other components of the models-3 community multiscale air quality (CMAQ) modeling system. Appl. Mech. Rev. 59, 51–77. Demers, J.D., Driscoll, C.T., Fahey, T.J., Yavitt, J.B., 2007. Mercury cycling in litter and soil in different forest types in the Adirondack region, New York, USA. Ecol. Appl. 17 (5), 1341–1351.
J.O. Bash, D.R. Miller / Atmospheric Environment 43 (2009) 5953–5961 Fay, L., Gustin, M., 2007. Assessing the influence of different atmospheric and soil mercury concentrations on foliar mercury concentrations in a controlled environment. Water, Air, Soil, Pollut. 181 (1–4), 373–384. Finnigan, J.J., 2004. A re-evaluation of long-term flux measurement techniques. Part II: coordinate systems. Boundary Layer Meteorol. 113 (1), 1–41. Foken, Th., Wichura, B., 1996. Tools for quality assessment of surface-based flux measurements. Agric. For. Meteorol. 78, 83–105. Gbor, P.K., Wen, D., Meng, F., Yang, F., Zhang, B., Sloan, J.J., 2006. Improved model for mercury emission, transport and deposition. Atmos. Environ. 40 (5), 973–983. Graydon, J.A., St. Louis, V.L., Hintelmann, H., Lindberg, S.E., Sandilands, K.A., Rudd, J.W.M., Kelly, C.A., Hall, B.D., Mowat, L.D., 2008. Long-term wet and dry deposition of total and methyl mercury in the remote boreal ecoregion of Canada. Environ. Sci. Technol. 42, 8345–8351. Graydon, J.A., St. Louis, V.L., Lindberg, S.E., Hintelmann, H., Krabbenhoft, D.P., 2006. Investigation of mercury exchange between forest canopy vegetation and the atmosphere using a dynamic chamber. Environ. Sci. Technol. 40, 4680–4688. Gustin, M.S., Lindberg, S.E., Marsik, F., Casimir, A., Ebinghaus, R., Edwards, G., Hubble-Fitzgerald, C., Kemp, R., Kock, H., Leonard, T., London, J., Majewski, M., Montecinos, C., Owens, J., Pilote, M., Poissant, L., Rasmussen, P., Schaedlich, F., Schneeberger, D., Schroeder, W., Sommar, J., Turner, R., Vette, A., Wallschlaeger, D., Xiao, Z., Zhang, H., 1999. Nevada STORMS project: measurement of mercury emissions from naturally enriched surfaces. J. Geophys. Res. 104 (D17), 21831–21844. Hanson, P.J., Lindberg, S.E., Tabberer, T.A., Owens, J.G., Kim, K.-H., 1995. Foliar exchange of mercury vapor: evidence for a compensation point. Water, Air, Soil, Pollut. 80, 611–620. Horst, T.W., Weil, J.C., 1994. How far is enough? the fetch requirements for micrometeorological measurements of surface fluxes. J. Atmos. Oceanic Technol. 11, 1018–1025. Howell, J.F., Mahrt, L., 1997. Multiresolution flux decomposition. Boundary Layer Meteorol. 83, 117–137. Jarvis, P., 1976. The interpretation of the variations in leaf water potential and stomatal conductance found in canopies in the field. Royal Soc. London Proc. Ser. B 273, 593–610. Kaimal, J.C., Finnigan, J.J., 1994. Atmospheric Boundary Layer Flows: Their Structure and Measurement. Oxford University Press, New York. 289pp. Keeler, G.J., Landis, M.S., Norris, G.A., Christianson, E.M., Dvonch, J.T., 2006. Sources of mercury wet deposition in Eastern Ohio, USA. Environ. Sci. Technol. 40, 5874–5881. Landis, M.S., Keeler, G.J., 1997. Critical evaluation of a modified automatic wet-only precipitation collector mercury and trace element determinations. Environ. Sci. Technol. 31, 2610–2615. Lee, X., Massman, W., Law, B. (Eds.), 2004. The Handbook of Micrometeorology: a Guide for Surface Flux Measurements and Analysis. Springer, p. 268. Lee, X., Benoit, G., Hu, X., 2000. Total gaseous mercury concentrations and flux over a coastal saltmarsh vegetation in Connecticut, USA. Atmos. Environ. 34, 4205–4213.
5961
Lindberg, S.E., Bullock, R., Ebinghaus, R., Engstrom, D., Feng, X.B., Fitzgerald, W., Pirrone, N., Prestbo, E., Seigneur, C., 2007. A synthesis of progress and uncertainties in attributing the sources of mercury in deposition. Ambio 36 (1), 19–32. Lindberg, S.E., Hanson, P.J., Meyers, T.P., Kim, K.-H., 1998. Air/surface exchange of mercury vapor over forests – the need for a reassessment of continental biogenic emissions. Atmos. Environ. 32, 895–908. Mahrt, L., Moore, E., Vickers, D., Lensen, N.O., 2001. Dependence of turbulent and mesoscale velocity variances on scale and stability. J. Appl. Meteorol. 40, 628–641. Pattey, E., Deshardins, R.L., Boudreau, F., Rochette, P., 1992. Impact of density fluctuations on flux measurements of trace gases: implications for the relaxed eddy accumulation technique. Boundary Layer Meteorol. 59, 195–203. Rea, A.W., Lindberg, S.E., Scherbatskoy, T., Keeler, G.J., 2002. Mercury accumulation in foliage over time in two northern mixed-hardwood forests. Water, Air, Soil, Pollut. 133, 49–67. Sigler, J.M., Lee, X., 2006a. Recent trends in anthropogenic mercury emissions in the Northeast United States. J. Geophys. Res. 111, D14316. Sigler, J.M., Lee, X., 2006b. Gaseous mercury in background forest soil in the northeastern United States. J. Geophys. Res. 111, G072007. St. Louis, V.L., Rudd, J.W.M., Kelly, C.A., Hall, B.D., Rolfhus, K.R., Scott, K.J., Lindberg, S.E., Dong, W., 2001. Importance of the forest canopy to fluxes of methylmercury and total mercury to boreal ecosystems. Environ. Sci. Technol. 35 (15), 3089–3098. U.S. Environmental Protection Agency (EPA), 1996. Method 1669. Sampling Ambient Water for Trace Metals at EPA Water Quality Criteria Levels. EPA. EPA-821-R-96-011. U.S. Environmental Protection Agency (EPA), 1999. Method 1631, Rrevision B: Mercury in Water by Oxidation, Purge, and Trap, and Cold Vapor Atomic Fluorescence Spectrometry. Office of Water, EPA. EPA-821-R-99-005. VanArsdale, A., Weiss, J., Keeler, G., Miller, E., Boulet, G., Brulotte, R., Poissant, L., 2005. Patters of mercury deposition and concentrations in northeastern North America (1996–2002). Ecotoxicology 14 (1–2), 37–52. Vickers, D., Mahrt, L., 2003. The cospectral gap and turbulent flux calculations. J. Atmos. Oceanic Technol. 20 (5), 660–672. Walker, J.T., Robarge, W.P., Wu, Y., Meyers, T.P., 2006. Measurement of bi-directional ammonia fluxes over soybean using modified Bowen-ratio technique. Agric. For. Meteorol. 138, 54–68. Webb, E.K., Pearman, G.I., Leuning, R., 1980. Correction of flux measurements for density effects due to heat and water vapor transfer. Q.J.R. Meteorol. Soc. 106, 85–100. Wesely, M.L., Hicks, B.B., 2000. A review of the current status of knowledge in dry deposition. Atmos. Environ. 34, 2261–2282. Xin, M., Gustin, M., Johnson, D., 2007. Laboratory investigation of the potential for re-emission of atmospherically derived Hg from soils. Environ. Sci. Technol. 41 (14), 4946–4951. Xu, X., Yang, X., Miller, D., Helble, J.J., Carley, R.J., 1999. Formulation of bi-directional atmosphere-surface exchanges of elemental mercury. Atmos. Environ. 33 (27), 4345–4355.