Gerhard Lammel: “Trends and Advances in Atmospheric and Environmental Chemistry" Budgeting atmospheric processes Halogenated SOCs and multicompartmental substances Air-surface mass exchange processes 4 Trace substance mass budgets, surface cycling: Emissions, deposition, re-volatilisation 4.1 Mass budget equation, residence time dmi/dt = sources – sinks = Ei – Si = Ei – (ki degrad (1) + ki dep (1)) mi = mi/τ [g/s] dci/dt = Ei – Si = Fi em/h - (ki degrad (1) + ki dep (1)) ci = ci/τair [g/m³/s] • Chemical loss processes of i are 1st order in ci • Source processes of i are 0th order in ci Depositional loss processes are here expressed as 1st order in ci for simplicity For dmi/dt = 0, the system is called to be chemically in a steady state dmi/dt = (Fi in + Ei) – (Fi out + Si) with: Fi in, Fi out = fluxes over boundary Ei, Si = internal sources and sinks mi = Mg i/Mg air mtrop Mg i, Mg air = molar masses (Mair = 29 g/mol) = spatial average of mixing ratio mtrop = mass of tropospheric air = 4.25× 1015 t Si = (j kij (2) Nj /V + ji (1)) Ni/V = kV (1) Ni/V with: kji (2), ji (1)= rate coefficients, photolysis rates Ni/V, Nj/V = reaction partner number concentrations kV (1) = tropospheric average chemical sink rate coefficient If well mixed or almost well mixed: advective losses Fi out Fi out ~ mi = kF mi; with: kF = empiric parameter dmi/dt = Fi in + Ei + (kF + kV (1)) mi τi = (kF + kV (1))-1; with: τi = residence time (not equal to but < ‚lifetime‘!) assuming (in 1st approx.) that kV (1) ≠ f(mi), i.e. no chemical feedbacks leading to Nj/V = f(Ni/V) Variability and atmospheric residence time: Averaging over long times (> mixing times) steady state-assumption holds: dmi/dt = + - (kF + kV (1)) mi  0 Ni/V = + (Ni/V)‘(x, y, z, t); with: = temporally and spatially mean number concentration (Ni/V)‘ = local and temporal number concentration x, y, z = space coordinates Empiric finding (Junge, 1974) for the relative standard deviation i = i*((Ni/V)‘) / = 0.14 / τi with: i*((Ni/V)‘) = absolute standard deviation of (Ni/V)‘  The residence time, τi, can be inferred from variability, as i = f(τi) Example: non-methane hydrocarbons (NMHC) Global budget (Tg/a) Natural 1150 terrestrial vegetation 2 marine biosphere Anthropogenic 120 of which are: 52 % transport 7 % fossil fuels, stationary 5 % chemical, petrochemical industries 9 % oil and gas production 27 % solvents (Ehhalt, 1986; Guenther et al., 1996) • location: mostly from ground, from stacks, from aircrafts • temporal profile, e.g. diel, weekly, seasonal, historical trends • spatial distributions 4.2 Emissions Emissions, CH4 (Crutzen & Gidel, 1983) Global Model results Tg/10°lat lat increase sources (Model results: Horowitz et al., 2003; Crutzen & Gidel, 1983) Global distributions CO (ppbv) @ 970 and 510 hPa, monthly mean Distributions – spatial, seasonal Measurements: polluted air masses background background in the marine boundary layer 60°N (courtesy of Chi & Andreae 2013) Regional distribution CO (ppbv) @ 300 m 60°N/90°E (ZOTTO) Distributions – spatial, seasonal Land use and atmosphere in South America (Ranson & Wickland, 2001) Difference Vegetation Index CO column density wet season dry season Biomass burning Example of the spatial distribution of nighttime active fires given by the ATSR satellite sensor example of burnt area provided by the GBA- 2000 product (Tansey et al., 2008) Higher levels in the N Pacific (Iwata et al., 1993) N-S gradient in the Bering and Chukchi Seas (Jantunen & Bidleman, 1995) Many (most) semivolatile and persistent organic substances are accumulating in high latitudes (despite source distribution). Example -hexachlorocyclohexane (-HCH) Halogenated SOCs and multicompartmental substances Introduction: concerns persistence, bioaccumulation and effects Decreasing trends in air, water and sediments not found in biota: (AMAP, 2004) cfish monitoring Decreasing trends in air and water not necessarily followed in organisms: Bioaccumulation along food chains = volatilisation and dry deposition of (gaseous) molecules - gas flux through interface F = -vwg (cw - cwi) = -vgw (cgi - cg) [mol/m²/s] with cwi = cg / Kaw - all physics hidden in v; equilibrium established at the interface itself 6.2.2.2 SOCs surface exchange 6.2.2.2.1 Air-sea exchange Mass transfer coefficient kmt (often Kol) H’= dimensionless Henry coefficient Kaw Faw = kmt (cw-ca/Kaw)3mm 0.2mm Schwarzenbach et al., 1993 (courtesy of M. Tsapakis) Two film model (or: two film theory of gas absorption) - existence of 2 stagnant layers on either side of the interface = transition zone from fully turbulent to molecular conditions - provide resistance additively Raw = Ra + Rw  1/vaw = 1/vw + 1/(vaKaw) (Liss & Slater, 1974; Schwarzenbach et al., 2002) Two film model (followed) - there is a water-phase and an air-phase controlled regime existing - 1/vaw = 1/vw + 1/(va Kaw) means that in the water-phase controlled regime the overall transfer velocity is independent, in the air-phase related regime linearly dependent of Kaw. This asymmetry, however, is only related to the decision to relate all concentrations to the reference phase water. Schematic of overall air-water exchange velocity vaw as a function of the air-water partition coefficient Kaw. Symmetry as equally frequent renewed surfaces are assumed (not realistic). F = -kmt w (cw - cwi) = -kmt a (cai - ca) [mol/m²/s] defined positive for flux from air to water cai = Kaw cwi with actual bulk (cw,ca) and equilibrium (cwi, cai) concentrations in water and air F = -kmt net [cw - caRgT/(cwH(T,s))] ; salinity s, H [Pa m³/mol]), Rg = 8.206 m³Pa/K/mol H(T,s) = RgT Kaw(T,s) = RgT Kaw(T) × 10Ks cs ; Setschenow constant Ks [L/mol] Ks = 0.04 log Kow + 0.114 (Ni & Yalkowsky, 2003) Resistance by boundary layers: reciprocal transfer coefficients (‚piston velocity‘ kmt w, kmt g [m/s]) R = 1/kmt net = 1/kmt w + RgT/(kmt g H(T,s)) [s/m] consideration of 1 side sufficient for most gases However: Stagnant film model (which ignores non-diffusive mass transport) implies F = (D/z)c with diffusion layer thickness z (can be estimated). F ~ D is not really true. Other existing conceptual models of air-sea mass exchange: - Surface renewal model (interface periodically renewed by turbulence eddies) implies F = 2(D/ta)0.5c ~ D0.5. However, the model is not useful, as the characteristic surface renewal time,ta, is not known or experimentally accessible. - Turbulent fluid flow based model - Turbulence enhancement by bubbles model: bubbles are created by breaking waves (u > 13 m/s) - Surface film effects: no direct inhibition but indirect (hydrodynamic: wave dampening (3.6 < u < 13 m/s), suppression of surface renewal…) effects These conceptual models are not predictive / fail as they are limited to individual processes which in reality combine (review: Johnson, 2000) Overall air-water transfer velocity as a function of Henry‘s law coefficient for 2 very different wind regimes. Result: magnitude of air-water transfer velocity v is determined mainly by wind speed, less by substance (Henry coefficient) Result of lab and field measurements: vw is positively correlated with wind speed, faster for u10> 10 m s-1  Parameterisations in models are empirically based. Wind dependence: - 3 linear regimes for the piston velocity kw (Liss & Merlivat, 1986) - quadratic kw CO2(u) = 0.31 u² (Sc/660)0.5 cm/s (Wanninkhoff, 1992) Empiric relationships to for H2O, CO2, and derived for unknown molecule i: kmt a H2O = 0.83 cm s-1  kmt g i = 0.83 (18/Mgi)0.5 cm s-1 ; molecular mass Mgi (g/mol) kmt w CO2 = 0.0056 cm s-1  kmt w i = 0.0056 (44/Mgi)0.5 cm s-1 (Atlas & Giam, 1986) kmt a H2O = 0.2 u10 + 0.23 kmt g i = (Dg i/Dg H2O)0.67 vg H2O (Mackay & Yeun, 1983) kmt w = 36× (0.2 u10 + 0.3) × (Di g/DH2O g)0.61 ; wind velocity in 10 m height u10 kmt a = 0.01× (0.45 u10 1.64) × (Sci/ScCO2)-0.5 ; Schmidt number Sc:=  /D Kinematic viscosity  := fluid viscosity µ/fluid density  (Murphy, 1995) Model experiment with transient historic emissions 1950-1990 (Stemmler & Lammel, 2009) Net-deposition = Deposition – Volatilisation (kg/m²/s) Volatilisation rate v determined by sea surface temperature t, wind speed u and concentration in seawater c. Here: p(T), H(T), Kow, koc(T) and c of DDT Coefficients of determination R2 (linear correlation) used to find out which of the parameters explains most of the variance of the volatilisation rate. (Stemmler & Lammel, 2011)