Atmospheric Mercury Wet Deposition along the Northern Gulf of Mexico

Continuous event-based rainfall samples were collected at three sites throughout the Pensacola airshed from 2005 - 2011. Samples were analyzed for total mercury (Hg), a suite of trace metals (TMs), and major ions in order to understand how thunderstorms affected their wet deposition and concentrations in rainfall, estimate the contributions from regional coal combustion and other anthropogenic sources to Hg and TMs in rainfall along the Gulf Coast, and investigate the possible influence that a local 950 megawatt coal-fired power plant had on rainfall chemistry in the Pensacola airshed. Mercury was measured with a Tekran 2600 using a method that was a variation of the standard method used by the US Environmental Protection Agency (EPA) to measure total Hg in water which allowed for the analysis of TMs from the same bottle without having to worry about contamination from reagents during sample preparation. Trace metals were measured used an Agilent 7500cs quadrupole inductively coupled plasma mass spectrometer (ICP-MS) while utilizing an octopole reaction cell (ORC) which allowed for the detection of key coal-combustion tracers like arsenic (As) and selenium (Se). Our findings show that summertime rainfall Hg concentrations are higher compared to other months despite higher rainfall amounts. In contrast, other measured pollutant TMs and ions did not show a consistent seasonal pattern. By incorporating Automated Surface Observing System data from nearby Pensacola International Airport and WSR-88D radar data from Eglin Air Force Base, we were able to classify the storm type (thunderstorms or non-thunderstorms) and analyze altitudes of hydrometeor formation for individual rain events. This showed that mid-altitude and high-altitude composite reflectivity radar values were higher for both thunderstorm and non-thunderstorm ”warm season” (May - Sept) rain events compared to ”cool season” (Oct - Apr) events including cool season thunderstorms. Thus, warm season events can scavenge more soluble reactive gaseous Hg from the free troposphere. Two separate multiple linear regression analyses were conducted on log-transformed data using interaction and non-interaction terms to understand the relationship between precipitation depth, season, and storm-type on sample concentrations. The regressions without interaction terms showed that the washout coefficients for more volatile TMs like Hg and Se were less pronounced compared to other pollution-type elements and that their concentrations were therefore less diluted for a given increase in precipitation depth, but otherwise displayed similar coefficients for season and storm-type. The regression model with interaction terms revealed a more interesting dynamic where thunderstorms caused enhanced Hg concentrations in rainfall regardless of season or precipitation depth while showing a more volume-dependent relationship with TM concentrations as concentrations increased with increasing rainfall amounts relative to non-thunderstorm events. This suggests a vacuum cleaner effect such that for increasing storm strength, non-Hg aerosol TMs in the boundary layer are further entrained into a storm cell. With this understanding, a positive matrix factorization (PMF) analysis was conducted using the EPA PMF 5.0 software to estimate the contribution of different sources to Hg deposition. Our results suggest that approximately 84% (72 - 89%; 95% CI) of Hg in rainfall along the northern Gulf of Mexico is due to long-range transport from distant sources while a negligible amount (0 - 21%; 95% CI) comes from regional coal combustion. However, we found that anthropogenic sources like regional coal combustion and ore smelting were significant contributors to rainfall concentrations of other pollution-type TMs like copper, zinc, As, Se, and non-sea salt SO42-. Using modeled wind profiles via the HYSPLIT trajectory model, we assessed whether plumes from a local coal-fired power plant (”Plant Crist”) could be detected in the rainfall chemistry of rain events occurring downwind of the plant. We limit this analysis to cool season rain events between June 2007 (when the model began) and December 2011 (when the study ended) because modeled wind profiles showed better agreement with observations during this time period compared to the warm season. We also limit this analysis to cool season events since the spatial distribution of rainfall throughout the area is more even during this time which makes sample comparisons between sites easier since Hg/TM concentrations are affected by precipitation depth. Furthermore, we focus on Hg and other pollution-type TMs and major ions such as As, Se, and non-sea salt SO42- in this analysis as they serve as tracers of coal combustion. For our ”unpaired-site” analysis, we analyzed, for each individual site, the rainfall chemistry in a given sample as a function of the proportion of rain events associated with that sample that occurred downwind of Plant Crist. Using this method, we were not able to find evidence that the plant had a significant influence on Hg/TM concentrations or Hg/TM:Al enrichment ratios in rainfall. Similarly, for our ”paired-site” analysis, we consider the differences in rainfall chemistry between two sites - an upwind and downwind site pair - that were impacted by the same rain event where the downwind site was exposed to plumes from Plant Crist while the upwind site was not. As with the unpaired-site analysis, we did not find significant differences in the rainfall chemistry between upwind-downwind site pairs with regards to sample concentrations or enrichment ratios. A multiple linear regression analysis was then conducted using interaction terms to understand the relationship between the operation of a wet flue-gas desulfurization system (which began operation at the plant during the middle of the study), the relative exposure a rainfall sample had to the plumes coming from the plant, and the log-transformed precipitation depth on log-transformed sample concentrations. Besides for As, the first regression analysis did not find coefficient values of any statistical significance for any of the variables that would indicate that the scrubber affected the rainfall chemistry at the two urban sites nearest to the plant. The calculations for As gave mixed results as the coefficients for the non-interaction terms suggested that the scrubber and the plumes emanating from Plant Crist affected the concentration of As in rainfall while the interaction terms suggested that they did not. We perform another multiple linear regression analysis, but remove the complicating effects of precipitation depth on Hg/TM concentrations and instead analyze the effects that the scrubber and the plumes coming from the plant might have had on Hg/TM:Al ratios. Again, these results were inconclusive as the regression coefficients suggested that the scrubber helped reduce Hg and TM emissions from the plant while also suggesting that samples with more exposure to the plant’s plumes had lower enrichment ratios. We propose that we were unable to detect a chemical signal from Plant Crist in our rain samples due to a few possible reasons including quick scavenging of TMs from the plume at the onset of a rain event before reaching our sites, the reliance on radar data to determine start and stop times for rain events at the sites as opposed to on-site measurements, and relatively low spatiotemporal resolution for the wind trajectory model.