Calculating the air-sea flux of any trace gas: transfer velocity, chemical enhancement and uncertainty

Motivation

Classical air-sea flux theory

Calculating temperature- and salinity-dependent diffusive transfer velocities for any gas

Liss and Slater (1974) revisitedrg/rlSolubilty 'threshold'

Chemical enhancementPrevious studies

Rate 'threshold'

Wider application

Global K climatology for any gas

The effect of bubblesBarry Huebert's DMS observations / NOAA COARE prediction

Global analysis of potential error

Other uncertainties...

Calculating the air-sea flux of any trace gas: transfer velocity, chemical enhancement and uncertainty

Motivation

Lots of researchers need to calculate air-water exchanges from concentration difference measurements:Many not experts in gas exchange

Many for poorly studied gases (i,.e. Not GHGs, noble gases, O2 or DMS)Concentration uncertainty is large so simple (wind driven parameterised) approach to transfer velocity is probably sensible

Serious mistakes are often made in calculationse.g. for CH3OH using kl rather than kg leads to factor of 20 overestimation of flux!

When is it appropriate to consider either kl=Kl or kg=Kg?

Notwithstanding the need to choose the 'best' transfer velocity parameterisations; solubility and diffusivity of the gas, and viscosity of the medium must be quantified for the gas of interest

When is chemical enhancement potentially important?

What else should we be worrying about?

Two-layer model of gas exchange

F = -K.C= -Ka(Cg-Csg) = -Kw(Csl-Cl)

Csg = KH.Csl

Liss, P.S and Slater, P.G., 1974, Nature (247), 181-184

Jhne, B. (2009), Air-sea gas exchange in Encyclopedia of Ocean Science, second edition 147-156.

Two-layer model of gas exchange

F = -K.C= -Ka(Cg-Csg) = -Kw(Csl-Cl)

Csg = KH.Csl

1/Kw = 1/KH.ka + 1/kw 1/Ka = 1/ka + KH/kw

Rl = rl + rg

(Rg = rl'+rg')

r1

r2

R = r1+r2

V

Liquid phase transfer velocity, kl

kl = f(ux). (Sc/Sc0)-0.5

Sc = /D

kl commonly expressed as an exponential function of windspeed (u) scaled by the square root (or other negative exponent) of the ratio of the Schmidt number of the gas in question to a reference Schmidt number (Sc0)

n viscosity of seawater (T, S and composition dependent)

p density (T, S and composition dependent)

D diffusion coefficient of gas in question (dependent itself on the viscosity and also the molecular weight of the medium, and also the molcular weight and molecular volume of the gas in question).

Liquid phase transfer velocity, kl

TemperatureSalinityWind speedSolubility at STPT dependence of solubilityMolecular structureGas-specific dataPhysical forcingsHenry solubility in waterDiffusion coefficients in air and waterViscosity of air and waterSchmidt numbers in air and waterkw Nightingale2000, Wanninkhof92, Woolf97 (bubbles) and various otherska various schemes including Duce91, Jeffrey2010Kw and Ka (1/Kw = 1/kw + H/ka)

All parameters T dependent (and S dependent for water side)

OutputsTemperatureSalinityWind speedSolubility at STPT dependence of solubilityMolecular structureGas-specific dataPhysical forcingsHenry solubility in waterDiffusion coefficients in air and waterViscosity of air and waterSchmidt numbers in air and waterkw Nightingale2000, Wanninkhof92, Woolf97 (bubbles) and various otherska various schemes including Duce91, Jeffrey2010Kw and Ka (1/Kw = 1/kw + H/ka)

All parameters T dependent (and S dependent for water side)

OutputsIn press, Ocean Science...

Henry's law solubility and temp dependence mostly taken from Rolf Sanders compilation (http://www.mpch-mainz.mpg.de/~sander/res/henry.html), or primary literature where not compiled by Sander.

Salinity dependence of KH determined from novel relationship derived from empirical data on gas solubilities in seawater

Vb calculated using 'Schroeder' additive method

Diffusivities of gases in air and water and viscosities of air and water calculated from best available paramterisations

Transfer velocities: various parameterisations of kl and kg implemented. Nightingale et al 2000 (kl) and Jeffrey et al 2010 (kg) used here.

Key assumptions: neutral bouyancy, all the assumptions made by the kl and kg parameterisations selected(!)

TemperatureSalinityWind speedSolubility at STPT dependence of solubilityMolecular structureGas-specific dataPhysical forcingsHenry solubility in waterDiffusion coefficients in air and waterViscosity of air and waterSchmidt numbers in air and waterkw Nightingale2000, Wanninkhof92, Woolf97 (bubbles)ka various schemes including Duce91, Jeffrey2010Kw and Ka (1/Kw = 1/kw + H/ka)

All parameters T dependent (and S dependent for water side)

Outputs For each compound the following data are required:Henry's law solubility (KH)

T-dependence of KH (-solnH/R)

Molecular structure (in order to calculate liquid molar volume at boiling point, Vb)

Wind speed, temperature, salinity

Total transfer velocity

KH

kg

kl

u10

T

S

KH0

-solnH/R

Scg

Scl

Dg

Dl

g

l

gT

lT,S

Sensitivity analysis

gT

lT,S

Vb

CD

kg

kl

Estimated parameter /%uncertainty

Highly soluble gas e.g. NH3

Sparingly soluble gas. e.g CO2

50 50 10 25 5 5 10 10 25 10 10 10 10

0.1 50 10 16 -0.04 4 0.05 -0.05 -1 10 1 -1 9

40 2 1 2 -0.2 2 4 4 -6 20 0.1 -0.1 1

Sparingly soluble gas. e.g CO2

Highly soluble gas e.g. NH3

Estimated parameter /%uncertainty

Dl

Dg

25 25

0.1 3

11 0.3

Table presents percentage change in total transfer velocity over range of parameter uncertainty

Really important to know when to use kw or ka on their own rather than Kl (or Ka)

Application of thin film model of interfacial mass exchange to the air-sea interface

Early estimates of kg and kl for H2O and O2 and some trace gases of interest: SO2, N2O, CO, CH4, CCl4, CCl3F, CH3I, DMS

Showed that rg/rl was small (109

CH2I2photolysis10-4photolysis10-3

SO2--hydration106

CH4Oxidation by OH