1. Mercury Chemistry in the Global Atmosphere: Constraints from Mercury Speciation Measurements Noelle Eckley Selin EPS Grad Student Seminar Series 14 February 2006

2. Why study Mercury (Hg)? Mercury is a global environmental pollutant –Current levels in atmosphere are 3x pre- industrial levels –Accumulates in food webs as methyl mercury; risk to humans & environment (neurotoxin) National, regional, and international policy interest U.S. EPA recommended limit for mercury in hair: 1 ug/g Noelle’s hair: 1.1 ug/g EPA benchmark dose (10% of births show neurological defects): 11 ug/g www.greenpeace.org/usa/mercury

3. THE MERCURY CYCLE: CURRENT Wet & Dry Deposition 2600 ATMOSPHERE 5000 SURFACE SOILS 1,000,000 OCEAN 289,000 Net Wet & Dry Deposition 1900 Net Oceanic Evasion 1500 Net burial 200 Land emissions 1600 Quantities in Mg/year Uncertainty ranges in parentheses Adapted from Mason & Sheu, 2002 Anthropogenic Emissions 2400 Extraction from deep reservoirs 2400 River 200 (1800-3600) (700-3500) (1680-3120) (1300-2600) (700-3500)

4. Hg 0 1.7 ng/m3 Gaseous Phase Aqueous Phase Hg 0 Henry’s Constant 0.11 M/atm Particulate Phase Oxidation Hg 2+ 10-200 pg/m3 Hg P 1-100 pg/m3 Hg 2+ k=8.7(+/-2.8) x 10 -14 cm 3 s -1 (Sommar et al. 2001) k=9.0(+/-1.3) x 10 -14 cm 3 s -1 (Pal & Ariya 2004) Too high? (Calvert and Lindberg 2005) k=3(+/-2) x 10 -20 cm 3 s -1 (Hall 1995) Reported rate constants up to k=1.7 x 10 -18 cm 3 s -1 Henry’s Constant 1.4x10 6 M/atm OH O3O3 Oxidation HO 2 ? Reduction SO 3 k=1.1-1.7 x 10 4 M -1 s -1 (Pehkonen & Lin 1998) Shouldn’t occur (Gårdfeldt & Jonsson 2003) k=0.0106 (+/- 0.0009) s -1 (vanLoon et al. 2000) Occurs only where high sulfur, low chlorine Oxalate?

5. Approach Use observations from latitudinal gradient, seasonal cycles, and short-term variability to constrain uncertainties in Hg chemistry and deposition, using GEOS-Chem mercury simulation and sensitivity simulations

6. Hg(0) 4500 (trop: 3900) Hg(II) 760 (trop:240) OH:8400 Dry Deposition Land (Natural) Emission Anthropogenic Emission Land Re-emission Hg(P) 1.9 (trop:1.9) 720 200 O3:2500 1500 1300 500 2000 Dry Deposition Wet Deposition 4400 1500 19011 Inventories in Mg Rates in Mg/yr k=3 x 10 -20 cm 3 s -1 hv (cloud):5932 k=6.9 x 10 -14 cm 3 s -1 Ocean Emission Mercury Budget in GEOS-Chem

7. Constraints from annual mean TGM + Average concentration at 22 land-based sites Measured: 1.60 ng/m3 Modeled: 1.60 ng/m3 High Atlantic cruise data? Oxidation rate constant (OH) adjusted to correspond to mean concentrations. Shown above: oxidation rate k=6.5 x 10-14 cm3 s-1 k=8.7(+/-2.8) x 10-14 cm3 s-1 (Sommar et al. 2001)

8. Constraints from Interhemispheric Gradient Measurement-based estimates of interhemispheric gradient: Lamborg et al. (2002): 1.2-1.8 Temme et al. (2003): 1.49 (+/- 0.12) GEOS-Chem interhemispheric gradient: 1.21 GEOS-Chem TGM lifetime: 0.92 yr Consistent with TGM lifetime of 1 year Interhemispheric gradient Constrains TGM lifetime *=land-based stations; +=Temme, 2003 (Atlantic); Δ=Fitzgerald, 1995 (Pacific); ◊=Laurier, 2004 (Atlantic); red line=GEOS-Chem global average

9. Constraints from Seasonal Variations Measurements Model Measurements Model (OH, O3, reduction) OH only O3 only 12 sites

10. Constraints from Time Series at Okinawa [Jaffe et al. 2005] Diurnal variation of RGM: daytime production plus rapid sink (uptake onto sea-salt?) measurements, standard model, O3 only, without sea salt morning increase a constraint on OH oxidation  One grid box upwind

11. RGM model-measurement comparison at Okinawa A sea-salt sink for RGM? Previous GEOS-Chem vs. measurements at Okinawa by Jaffe et al. (2005): model overestimates measurements by a factor of 3 (note difference in scale), but captures some day-to- day variation in observations Revised Model and measured RGM including an implied sink for RGM (sea salt uptake?) are consistent with order of magnitude of Okinawa observations (same scale)

12. Okinawa Data: Hg(0) vs CO and Asian Emissions model (red), measured [Jaffe et al 2005] (black) Hg(0)/CO ratio: check on Asian emissions Slope 0.0053 in measurements 0.0036 in model Pacyna et al 2003: 770 Mg/year Jaffe: 1460 Mg/year (based on data) GEOS-Chem Asia: (GEIA 2000 inventory) Hg(0): 586 Mg Hg(II): 365 Mg land reemission: 342 Mg total Hg(0)-Asia: 928 Mg Consistent with Jaffe underestimate of Asian emissions – but land reemission is a substantial portion!

13. Constraints from Annual Average RGM Variable measurements; 2 cruises average of all measurements 17.4 pg/m3, GEOS-CHEM 8.3 pg/m3 however, skewed by a few high measurements Limitations from RGM – HgP partitioning

14. Constraints from Time Series at Mt Bachelor [Swartzendruber et al. 2005] RGM concentrations higher in the free troposphere Negative correlation between Hg(0) and RGM at night @ Mt Bachelor (r=-0.67 for meas, r=-0.71 for GEOS-Chem). Negative correlation between relative humidity and RGM, reproduced in model (downwelling?)

15. Constraints from Wet Deposition Data from U.S. Mercury Deposition Network (2006) Moderate correlation (r2=0.52 for 2003, 0.66 for 2004) GEOS-Chem underestimates wet deposition over U.S. by c. 25% 2 patterns: latitudinal variation (OH oxidation) and regional enhancement (sources) Comparison with measurements% deposition from U.S. Sources

16. Conclusions and Future Work GEOS-Chem model suggests that –OH, O3 reactions, coupled with reduction, provide best explanation for Hg observations –Rapid RGM uptake onto sea-salt aerosol –Elevated RGM in free troposphere & stratosphere Future work: land emissions parameterization Acknowledgments: Prof. Daniel Jacob (advisor); Bob Yantosca (Harvard); Rokjin Park (Harvard); Sarah Strode (U.Wa); Lyatt Jaegle (U.Wa)