S3 (I) The North American mercury speciation networks: Analysis and modeling results

The Canadian Atmospheric Mercury Measurement Network (CAMNet) and the Mercury Deposition Network (MDN) of the National Atmospheric Deposition Program (NADP) were both established in 1996 for the measurement of total gaseous mercury and mercury wet-deposition, respectively. While these networks offer important long-term monitoring data, a lack of scientific information on atmospheric mercury species, as well as limited coverage over various geographic scales has resulted in gaps with regard to mercury atmospheric fate and transport and total (wet + dry) mercury deposition across North America. Over the past several years, NADP and Environment Canada have worked to remedy the situation. In Canada, atmospheric mercury measurements have been folded into the larger CAPMoN program. In the United States, the Atmospheric Mercury Network (AMNet) has been established by the NADP as a new coordinated and standardized network. Under AMNet, U.S. and Canadian agencies, along with the academic research community, industry, and others are coordinating to monitor the three atmospheric mercury species that contribute to dry deposition - gaseous oxidized mercury (GOM), particulate-bound mercury (PBM2.5), and gaseous elemental mercury (GEM). With the development of AMNet, joint cooperation between U.S. and Canadian scientists has led to a more cohesive network of standardized, full mercury speciation measurements based on harmonized methods for field operations, data management and quality assurance. As an introduction to Special Session, S3, we will present: 1) an overview of the networks, 2) summary atmospheric mercury speciation data, 3) a comparative analysis of sites receiving different air mass types and 4) an approach for the estimation of mercury dry deposition on a network wide basis.

Three atmospheric mercury (Hg) concentrations have been measured since 2006 in northern New York State using a Tekran speciation system. At the rural site the concentrations of the three species were relatively stable (GEM: ~1.4 ng m^-3, GOM and PBM_2.5: ~1.3 and 4.0 ng m^-3) and somewhat lower than in an urban area (GEM: ~1.6 ng m^-3, GOM and PBM_2.5: ~6.0 and 9.0 ng m^-3). Dissimilar GEM diurnal patterns were recorded at the sites in the warmer seasons. GEM concentrations were driven mostly by mixed layer heights at the urban site. However, at the rural site it was correlated with O₃ concentrations. Diurnal GOM concentration variations were observed at both sites in the warmer seasons, when atmospheric chemistry was significant. The observed GOM reaction times (time from the lowest to the highest GOM concentration) were 20 to 100% of the theoretical reaction time. An elevated PBM_2.5 concentration was seen in winter at both sites and was probably related to biomass combustion for space heating. PBM_2.5 was also seen to simultaneously peak with carbonacous particles during a short period in summer 2010 when the site was impacted by forest fires in Quebec. Potential source contribution function (PSCF) was used with rural Hg concentrations to locate Hg sources. The most important sources were coal/oil-fired power plants, oil refineries, steel smelters, and landfills in the Midwestern US although sources as far away as Louisiana were seen. Principle component analysis was used to classified atmospheric Hg sources and chemical/physical processes using the urban site data. There were five important Hg related components extracted, including GEM released from melting snow, gas-phase oxidation, coal combustion, mobile sources, and non-coal combustion. The local coal-fired power plant significantly impacted Hg concentrations in the downwind area.

As part of the Atmospheric Mercury Network (AMNet), a site on the central California coast (Elkhorn Slough – CA48) was established in March 2010 for monitoring atmospheric mercury species (GEM, GOM, PBM2.5, and mercury dry deposition to a surrogate surface). Data from an annual cycle at this site are presented along with total gaseous mercury (TGM) data taken onboard the R/V Atlantis during the NOAA/CARB CalNex 2010 campaign to investigate the behavior of atmospheric mercury in California. Mean speciated mercury concentrations at CA48 between March-December 2010 were: GEM = 1.44 ± 0.14 ng/m3, GOM = 0.8 ± 1.0 pg/m3, and PBM2.5 = 2.4 ± 2.3 pg/m3. These values demonstrate that the site is characteristic of the marine boundary layer, with little influence from local emissions sources and little photochemical production of GOM. Dry deposition to a surrogate surface was measured weekly from September 2009 – November 2010 at CA48 and an average flux of 0.26 ± 0.16 ng/m2/hr was found. This value compares closely to dry deposition measurements obtained at other coastal sites in the US and is consistent with low concentrations of GOM. In contrast with the observations at Elkhorn Slough, TGM concentrations were elevated at many locations off the coast of Los Angeles where there are several large mercury point sources including an incinerator and oil refineries. On three occasions when the ship was within 1 km of a large incinerator, TGM concentrations were 6-7 ng/m3. Using 2008 emissions inventories for the criteria pollutants (CO, SO2, NOx) and the observed plume enhancement ratios with TGM, the mercury emissions flux from this incinerator was estimated to be 11.5 kg/y. This is about 6 times lower than the published mercury emissions inventory for this facility. On seven additional occasions along the Los Angeles coastline, TGM enhancements were observed along with the criteria pollutants, and these air masses were typical of well-mixed pollutant outflow from the Los Angeles basin with TGM concentrations < 2.5 ng/m3. In general however, TGM enhancement events (> 1.7 ng/m3) were uncommon throughout the ship track, as indicated by a mean TGM concentration of 1.39 ± 0.17 ng/m3. This suggests that California mercury emissions may play a small role in determining mercury concentrations in the coastal atmosphere of California.

Measurements of speciated atmospheric Hg have been conducted at Lulin Atmospheric Background Station (LABS; 120.87°E, 23.47°N, 2862 m a.s.l.) in Taiwan since April 2006 to study the export of East Asian atmospheric mercury in the free troposphere. Between April 2006 and June 2010, the mean concentrations of GEM, RGM, and PHg were 1.81 ng m^-3, 15.2 pg m^-3, and 2.2 pg m^-3, respectively. GEM usually peaked in the afternoon. In contrast, spikes of RGM were frequently observed between midnight and early morning with concurrent decreases in GEM and relative humidity and increases in O₃, suggesting the oxidation of GEM and formation of RGM in free troposphere (FT). Upslope movement of boundary layer (BL) air in daytime and subsidence of FT air at night resulted in these diurnal patterns. Considering only the nighttime data, which were more representative of FT air, seasonal variation in GEM was evident, with lower concentrations usually occurring in summer when marine air masses prevailed. Between fall and spring, air masses passed the East Asian continent prior to reaching LABS. Trajectory cluster analysis identified 9 groups of air mass transport paths, 5 groups mainly passed over the East Asian continent and the other 4 groups mainly passed over the Pacific Ocean/South China Sea. Concentrations of GEM, CO, O₃ and PM₁₀ were significantly elevated in air masses coming from the East Asian continent, demonstrating the influence of human activities. Analysis of GEM/CO correlation further supported the argument. Good GEM/CO correlations were observed in fall, winter, and spring, suggesting influence of anthropogenic emission sources. Our results demonstrate the significance of East Asian Hg emissions, including both anthropogenic and biomass burning emissions, and their long-range transport in the FT. Because of the pronounced seasonal monsoon activity and the seasonal variation in regional wind field, export of the Asian Hg emissions to Taiwan occurs mainly during fall, winter, and spring.

Regional scale photochemical transport models are used to estimate mercury deposition that result from changes in local, regional, and global scale emissions. The Atmospheric Mercury Network began making routine sub-hourly measurements of speciated ambient mercury in 2008 and 2009. This network provides a unique opportunity to assess how well state of the science photochemical transport models estimate observed diurnal and seasonal patterns of ambient speciated mercury: Hg(0), Hg(2+), and Hg(p). In addition, model estimates of total mercury deposition and rainfall are directly compared to measurements at monitors in the Mercury Deposition Network. The Community Multi-scale Air Quality (CMAQ) model version 4.7.1 represents mercury emissions, removal processes, and chemical transformations through dry and wet deposition schemes, gas-phase, and aqueous phase chemistry. The photochemical model is applied with anthropogenic, biogenic, and geogenic emissions, boundary conditions, and meteorology for the entire year of 2005 at 12 km grid resolution covering the continental United States. Particle bound mercury contributes about 4%, Hg(0) 25-29%, and Hg(2+) 67-71% to total mercury deposition over the continental United States. Approximately 65% of mercury is removed through dry processes and 35% through wet processes. The composition of dry deposition is roughly equal between Hg(0) and Hg(2+) while Hg(2+) dominates the composition of mercury wet deposition (>90%). Total wet deposition performance for CMAQ is consistent with previously published modeling results: annual mean bias and error are 34% and 52% respectively. CMAQ compares well with total mercury wet deposition observations in the eastern United States but over-estimates in the western United States. Systematically over-estimated rainfall in the western U.S. in part explains over-estimated wet deposition and under-estimated mercury concentration in wet precipitation. Relationships between rainfall and mercury wet deposition are less clear in the eastern U.S. CMAQ captures the seasonal trends and diurnal patterns of Hg(0) and Hg(2+) ambient concentrations and wet deposition well in the eastern U.S. The modeling system does not capture diurnal patterns of Hg(0) and Hg(2+) in the western U.S. as well as in the eastern U.S.

HYSPLIT-Hg is a specialized version of the NOAA HYSPLIT model being developed to simulate mercury’s atmospheric fate and transport. HYSPLIT-Hg outputs have been compared against hourly-average ambient data from two AMNet sites: Beltsville, Maryland (N39.03, W76.82); and Grand Bay, Mississippi (N30.41, W88.40). The model evaluation exercise was focused on specific episodes of high mercury measurements in 2008-2009. This high-resolution evaluation may be a more demanding model test than comparison against regional, long term averages. Numerous simulations were performed to examine the model’s sensitivity to different inputs and configurations, including variations in meteorological and emissions data, and model physics/chemistry. Perturbations were limited to realistic ranges, and each simulation was characterized by the degree to which model predictions matched measured speciated mercury concentrations. In addition to comprehensive forward fate/transport dispersion simulations with HYSPLIT-Hg, back-trajectories of air parcels arriving at the sites during the episodes were also examined. Different meteorological inputs to forward dispersion and back-trajectory simulations included gridded outputs from large-scale weather models with horizontal resolution of 40 km (Eta Data Assimilation System, “EDAS”), 12 km (North American Mesoscale, “NAM”), and 4 km (Weather Research and Forecast, “WRF”), with analogous variations in vertical resolution. The 4-km WRF datasets were developed specifically for this model evaluation exercise. Potential effects of errors in the meteorological data were also investigated. Emissions inputs investigated included differences in speciation, total emissions, temporal emissions variations, and the types (e.g., natural vs. anthropogenic) and geographical extent (e.g., local vs. regional vs. global) of emissions sources considered. Model physics and chemistry aspects investigated included the rates of chemical reactions involving atmospheric mercury, algorithms for dry and wet deposition processes, the numerical treatment of atmospheric dispersion, model time step, and several other aspects of the simulation. Results to date show that the HYSPLIT-Hg model has reasonable skill in matching observed mercury concentrations, but uncertainties in model inputs (some of which are inherently stochastic) and in model physics and chemistry represent significant challenges. At the same time, results to date show that the model has widely different sensitivities to different types of perturbations, and these variations may be useful in informing the prioritization of efforts to reduce uncertainties.

Dry deposition fluxes during 2008-2009 for speciated mercury, i.e., gaseous oxidized mercury (GOM), particulate bound mercury (PBM), and gaseous elemental mercury (GEM), were estimated at Atmospheric Mercury Network (AMNet) locations using the inferential method, which combined the 2-3 hourly speciated ambient concentration data with calculated hourly speciated dry deposition velocities (V_d). The meteorological input used for V_d calculations were from the archived surface-layer meteorology produced by the Canadian weather forecast model. Annual dry deposition of GOM+PBM was estimated to be in the order of 1-5 g m^-2 at the majority of the sites with GOM deposition being 5-10 times higher than PBM deposition due to their different V_d values. The dry deposition of GOM+PBM estimated here was substantially lower than values produced from regional-scale mercury transport models published in the past. GEM dry deposition was estimated to be higher than those of GOM+PBM; this was also confirmed by a comparison of the estimated annual dry deposition with the annual litterfall deposition. Total dry deposition was found to be equal in importance to annual wet deposition of mercury. Seasonal and geographical patterns will also be discussed in this study.

The Great Salt Lake (GSL) in the western United States has been identified as the most mercury-laden body of water in the United States with a median water mercury concentration of 42 ng L^-1. When Hg enters an aquatic ecosystem it can be converted to methylmercury which bioaccumulates up the food chain. Methylmercury contamination has resulted in many consumption advisories for game fish in lakes and rivers throughout the Intermountain West. In 2005, the Utah Department of Health and the Fish and Wildlife Service placed a similar consumption advisory on waterfowl on the GSL. The primary goal of this study is to identify the pathway of greatest influx of Hg pollution to the GSL to give insight toward the source and an eventual solution to the Hg pollution problem.

Speciated atmospheric mercury measurements have been collected at the UT96 field site on the eastern shore of the GSL from July 1, 2009 to the present. The atmospheric mercury concentrations, along with high-frequency atmospheric turbulence measurements, were used as input to a resistance-in-series dry deposition model (based on Wesley and Hicks 1977). The dry deposition flux of mercury was determined from the modeled dry deposition velocity and the measured concentrations. This dry deposition flux was compared to the wet deposition flux measured by the National Deposition Network and the riverine influx measured by the USGS. It was found that in the first year of sampling (i.e., July 1, 2009 through June 30, 2010) 10.7 µg m^-2 of Hg was deposited into the GSL by dry deposition from the atmosphere.

Dry deposition makes up 60% of the total Hg influx from all measured pathways. The flux from the dry deposition of the global background pool of gaseous elemental mercury (1.5 ± 0.2 ng m^-2) dominated the dry deposition flux, making up 82.5±8.5% of the dry deposition flux and 50% of the total Hg influx to the GSL. Lake sediment cores from the GSL suggest a much larger annual flux of 68 µg m^-2 for the last 27 years. This discrepancy suggests that measurements of coarse particulate mercury and gaseous elemental mercury oxidation within the surface boundary layer should beundertaken at this location.