Development of Mercury Modeling Schemes Within CMAQ-Hg: Science and Model Implementation Issues Che-Jen Lin, Pruek Pongprueksa, Thomas Ho, Hsing-wei Chu & Carey Jang 2004 CMAS Models-3 Conference October 19, 2004

A potent neural toxin (LD 50 = mg/kg, RfD = mg/kg/day for methyl mercury) An EPA priority air pollutant Persistent – long range transport possible Established contamination episodes globally Sequestration not likely Bioaccumulative – enter the food chain Cycling in the environment Mercury as a Global Pollutant

Global Cycling of Mercury Data from Mason et al., 1994.

Emission Sources Anthropogenic sources –Fuel combustion: air emission –Waste incineration: air emission –Chloralkali process: water/air emission Natural sources –Volcano eruption, weathering, etc. –Vegetation, open water, soil emissions Re-emission –Caused by past mercury emission and deposition – Biotic and abiotic processes cause reduction of deposited Hg(II) back to volatile – Re-emit into the atmosphere

Atmospheric Mercury

“One-Atmosphere” Modeling - Hg Mercury exists at very low concentrations and has “its own” chemistry cycle in the atmosphere Concurrent atmospheric chemical processes involving multiple pollutants affect mercury transport and deposition Coupling of mercury with other atmospheric processes is complex and usually generates non-linear responses Chemical transport modeling of mercury needs to be considered an integral part of the modeling of other atmospheric pollutants and processes (e.g., ozone, PM, acid deposition, etc.)

Modeling Components Landuse DataSynoptic Meteorology Urban/Regional Meteorology Data Dynamic Meteorology Model ( MM5 ) Meteorology-Chemistry Processor ( MCIP2 ) – Hg Implementation Emission Inventory Model ( SMOKE & MIMS ) Model-Ready Emission Inventory Data Model-Ready Meteorology Data Chemical Transport Model (CMAQ-Hg) Emission Inventory DataLand Cover Data Solar Irradiation Data Initial Condition Data Boundary Condition Data Gaseous Poll. Conc. (Hg) Particulate Poll. Conc. (Hg) Visibility & Regional Haze Acid Rain Pollutant Deposition (Hg)

CMAQ-Hg (Bullock & Brehme, 2002) Emission Anthropogenic (Point & Area) Veg./re-emission needed Gas Chemistry O 3, Cl 2, H 2 O 2, and OH New chemistry & kinetics available Aq. Chemistry Ox: O 3, OH, HOCl, and OCl - Speciation controlled Red: HgSO 3, Hg(OH) 2 +hv, HO 2 Speciation controlled Aq. Speciation SO 3 2-, Cl -, OH - Major ligands considered Aq. Sorption Sorption of Hg(II) to ECA, bi- directional non-eq. kinetics w/ linear sorption isotherm High sorption constant implemented Dry Deposition V dep of HNO 3 for RGM deposition No Hg 0 deposition. RGM deposition likely too high V dep of I,J modes for PHg deposition As sulfate deposition Wet Deposition Dissolved and Sorbed Hg(II) aq By precipitation & aqueous concentration

Proposed CMAQ-Hg Implementations Dry deposition velocities of Hg 0 and RGM EI of veg. Hg emission; sea-salt aerosol gen. Photolysis rates of reactive halogens New gaseous phase chemistry and kinetic constants Halogen activation chemistry; Hg sorption in clouds

Mercury Chemical Mechanism

Gaseous Phase Oxidation Oxidants (molec cm -3 ) Typical Location Remark UrbanRemoteMBL O 3 (ppb)15030 Daytime O3O3 3.69x x10 11 Daytime OH5 x x x 10 6 Daytime H2O2H2O2 4.92x x10 10 Daytime Cl x x 10 9 Nighttime Cl1 x x 10 4 Daytime Br x 10 7 Nighttime Br001 x 10 5 Daytime BrO005 x 10 6 Daytime

Sea-Salt Aerosol Inventory Sea-salt aerosol as the primary sources of reactive halogen species Affect the chemistry in coastal areas & in Marine Boundary Layers Sea-salt aerosol generation algorithm Implementation in SMOKE modeling system

Halogen Activation and Chemistry Activation of reactive halogens from sea salt aerosols (Vogt et al., 1996; Glasaw et al., 2002; Knipping and Dabdub, 2002) Acid replacement reactions Oxidation of halides Autocatalytic generation Reaction from ClONO 2 with sea salts Photolysis of reactive halogen species Implementation in CMAQ to provide halogen oxidants for Hg 0

Mercury Emission Inventory Incorporation of vegetation emission in EI processing needed!

CMAQ-Hg Dry Deposition Species considered: RGM and PHg RGM: V dep of HNO 3 calculated by MCIP2 (0.5-8 cm/s during mid-day) used for RGM deposition – may overestimate V dep,RGM (in the range of cm/s) Dry deposition Hg 0 not considered, which may contribute significantly to total dry Hg deposition Implementing dry V dep in MCIP2 recommended

Estimating Mercury V dep MCIP2 supports two dry deposition schemes –RADM by Wesely (1989) –M3DRY by Pleim (1999) V dep = (R a + R b + R c ) -1 R a is the aerodynamic resistance R b is the quasi-laminar boundary layer resistance R c is the canopy (surface) resistance Estimating R c is the key to accurately represent mercury V dep

RADM vs. M3DRY RADM R c = [(r sx + r mx ) -1 + (r lux ) -1 + (r dc + r clx ) -1 + (r ac + r gsx ) -1 ] -1 Requires trace gas properties, horizontal winds, temperature, RH and 2-D met fields for V dep estimate M3DRY R s = {f v / r stb + LAI * [f v (1 – f w ) / r cut + f v f w / r cw ] + (1 – f v ) / r g + f v / (r lc + r g )} –1 Uses common components as in MM5 land-surface model to estimate V dep, corrected for landuse and soil moisture

Hg V dep Implementation - R c TermsFormulationDescriptionRemarks r dc 100[ (G + 10) -1 ] (  ) -1 - Buoyant convection resistance r sx r s D H2O /D x, where r s = r i {1 + [200(G + 0.1) -1 ] 2 } {400[T s (40 - T s )] -1 } - Stomatal resistance for substance x HNO 3 :D H2O /D HNO3 = 1.9 RGM :D RGM = cm 2 /s; D H2O /D RGM = 2.53 GEM :D GEM = cm 2 /s; D H2O /D GEM = 1.82 r clx [k H /(10 5 r clS ) + f 0 /r clO ] -1 - Lower canopy resistance HNO 3 :k H = 1 x M atm -1 ; f 0 (HNO 3 ) = 0.0 r gsx [k H /(10 5 r gsS ) + f 0 /r gsO ] -1 - Ground surf. resistance RGM :k H = 2.75x10 6 M atm –1 ; f 0 (RGM) = 0.1 or 1.0 r mx (k H / f 0 ) -1 - Mesophyll resistance r lux r lu (10 -5 k H + f 0 ) -1 - Leaf cuticular resist. GEM :k H = M atm -1, f 0 (GEM) = 0.0 [1/(3r lu ) k H + f 0 /r luO ] -1 - Dew or rain correction r l uS Leaf cuticular, SO 2 (Dew) [1/ /(3r lu )] -1 - Rain correction r luO [1/ /(3r lu )] -1 - Leaf cuticular, O 3 (Dew) [1/ /(3r lu )] -1 - Rain correction Note: r i, r lu, r clS, r clO, r ac, r gsS, r gsO are parameters depending on land uses and seasons

Sensitivity of RGM Surface Reactivity f 0 = 1.0 f 0 = 0.1 Surface reactivity does not affect the deposition velocity significantly!!

Dry Deposition Velocity: Hg vs. HNO 3 HNO 3 GEM HNO 3 - RGM RGM

Hg(II) Sorption in Aq. Phase Current version of CMAQ-Hg treats Hg(II) sorption as bi-directional sorption kinetics: Distribution of [Hg S 2+ ] and [Hg D 2+ ] estimated from a linear sorption isotherm using a scaled-up sorption constant for EC based on the sorption constant of APM.

Hg Sorption (Cont’d) Data describing water-solid partitioning of Hg(II) in cloud water not widely available Linear sorption isotherm appropriate for describing adsorption in cloud water Sorption constant implemented in CMAQ-Hg probably too high C e : Hg(II) aq q: Hg(II) sorbed q = K s C e

Hg Sorption Implementation Low APM concentration (typically a few mg/L or lower) and small particle size should lead to sorption equilibrium rapidly We recommend implementing Hg(II) sorption equilibrium using insoluble APM in the model: Sorption relationship implemented in model needs further experimental evaluation

Summary CMAQ-Hg serves as an excellent framework for simulation of atmospheric mercury Implementation of new mercury chemistry and reaction kinetics needed in gaseous phase Include vegetation emission in Hg emission processing Formulation and implementation of Hg deposition schemes needed for RGM and Hg 0 More experimental data needed to better describe Hg(II) sorption in aqueous phase Modules to generate sea-salt aerosols and to simulate reactive halogen cycle important for implementing gaseous Hg-halogen chemistry

Acknowledgements Texas Commission on Environmental Quality EPA-Gulf Coast Hazardous Substance Research Center Steve Lindberg, Oak Ridge National Laboratory Daewon Byun, University of Houston