How ocean CO2 fluxes are

estimated/measured

Colm Sweeney[csweeney@ldeo.columbia.edu]

Princeton Universityand

Lamont-Doherty Earth Observatory

Outline

IV. Improving our estimates of air-sea fluxes- Time-space distribution of pCO2 - Parameterization of gas transfer velocity

III. Surface measurements:-Measurements of surface pCO2

-Methods for interpolation

II. The air-sea flux measurement-Covariance-Gradient technique

I. Concept-Ocean carbon chemistry primer-The air-sea flux

Ocean Carbon Chemistry Primer

CO2(gas)

CO2 + H2O H2CO3

H3CO2 H+ + HCO3

-

HCO3- H+

+ CO32-

Carbonic acid

Bicarbonate

Carbonate

CO2 + CO32- 2 HCO3

-

TCO2

Ocean Carbon Chemistry Primer

CO2(gas)

CO2 + H2O H2CO3

H3CO2 H+ + HCO3

-

HCO3- H+

+ CO32-

Carbonic acid

Bicarbonate

Carbonate

CO2 + CO32- 2 HCO3

-

280 atm 560 atm

8 mol kg-1

1617 mol kg-1

268 mol kg-1

15 mol kg-1

1850 mol kg-1

176 mol kg-1

1893 mol kg-1 2040 mol kg-1

100% pCO2 8% TCO2

TCO2

Taken from Feely et al. (2001)

Concept

k =f(u*) Sc-n

u* – frictional velocity

s – solubilitySc – schmit number (v/D)n – 0.4 – 0.67 (high slope…low slope)

Net air-sea gas flux: Fgas=ks(pCO2w-pCO2a)

I=ks(pCO2a)River input:0.6 PgC yr-1

pCO2~2 atm

Keeling et al.

E=ks(pCO2w)

Bomb 14C

Broecker and Peng (1994)

Transfer velocitykav = 22 cm/hru* = 7.4 m/s

Semi-infiniteHalf space

Early estimates air-sea CO2 exchange

Natural14CO2/12CO2 in gassing

14CO2/12CO2 out gassing

n+14N14C

Decay: 14C 14N + e-

Pre-industrial assumption:14CO2 in = 14CO2 out + Decay

Solve for I

0.061 mol m-2 yr-1 uatm-1

=21.4 cm hr-1

seameansurfatm

VTCOCCC

][ 2

141414 CIA

CIA

C

Early estimates air-sea CO2 exchange

Natural14CO2/12CO2 in gassing

14CO2/12CO2 out gassing

n+14N14C

Decay: 14C 14N + e-

Pre-industrial assumption:14CO2 in = 14CO2 out + Decay

Solve for I

0.061 mol m-2 yr-1 uatm-1

=21.4 cm hr-1

222Rn 218Po + 4He

[Rn]mixed layer Rn

[Rn]no loss Rn+ gas exchange

226Raaq 222Rngas + 4He

Outgassing of Radon

=

0.062 mol m-2 yr-1 uatm-1

=21.9 cm hr-1

[Rn]

Flux Measurements in the Atmosphere

Direct covariance technique

Covariance flux of H2O and CO2

Fair-sea=<c'w'>

3-D SonicAnemometers

IR Detector(Sample)

H2O/CO2

samples

IR Detector(Motion Detection)

Std

Res

Pump

Gradient Flux Technique

z

czwuFnet

)(c

2/1

Frictional velocity

MeasuredGradient (3-13m)

Gradient Function-empirically determinedbased on Monin Obukhov (MO)similarity theory

McGillis et al. (2001)

Covariance intake

z

c

GasEx-98 Comparison-estimates of transfer velocity

GasEx-2001

Estimates of gas transfer velocity

Rayleigh DistributionFor ocean wind speedsP(u)

k- short term

21)660/(

])([/

Scauk

uuPkan

nav

Bomb 14Ckav=22 cm /hr

Estimates of CO2 fluxes from measurements of pCO2

1. Shipboard measurements of atmospheric and surface ocean pCO2

2. The ocean pCO2 climatology

3. Flux calculations using the climatology

Shipboard measurements of atmospheric and surface ocean pCO2

Equilibration of air sample

IR DetectorAir flow

Re-circulation

Drain

Takahashi pCO2 database

1,183,000 measurements- Since ~1968

Monthly distribution of pCO2

The climatology1. Exclude all El-Nino years.

- dramatic change in annual fluxes have been observedEl-Nino periods based on SIO<-1.5 and SST changes.

2. Normalize pCO2 single reference year (1995)- In warm waters (lat. <45) pCO2 remains constant

3. Interpolate data on to 4ox 5ox 365 day grid-finite differencing algorithm is used with a 2-D transport model from Toggwieler et al. (1989) to propagate the influence of observed data at one day time steps. Distribution is solved iteratively

Time

pCO2

The pCO2 Climatology

Global CO2 flux

Test of interpolation

pCO2

T0.28 C~0.8 PgC

=3.5%

Sampling resolution 250K samples(Takahashi ’97)

500K samples(Takahashi ’99)

940K samples(Takahashi ’02)

-3

-2

-1

0

1

2

N of 50N 14N-50N 14N-14S 14S-50S S of 50S Global

250K

500K

900KPgC

yr-1

Change in fluxes with increases in samples

Gas Transfer Velocity and Fluxes

Estimates using different gas exchange-wind speed relationships

Relationship Equation Flux

(Pg C yr-1)

Liss & Merlivat [1986] k= 0.17 U10 (U10

< 3.6 m s-1)

k= 2.85 U10 - 9.65 (3.6 m s-1<U10

< 13 m s-1)

k= 5.9 U10 - 49.3 (U10

> 13 m s-1)

-1.0

Wanninkhof [1992] [W-92] k= 0.39 U102 (long term averaged winds) -1.8

Wanninkhof&McGillis (1999)

[W&M-99]

k= 1.09 U10 - 0.333 U10

2 + 0.078 U103

(long term averaged winds)

-3.0

Nightingale et al. [2000] k= 0.333 U10 + 0.222 U10

2 -1.5

NCEP-41 year average windsb

[W-92]

k= 0.39 U102 (long term averaged winds) -2.2

Feely et al., 2001

Long vs. short term winds

-4-3-2-1012

N of50N

14N -50N

14N-14S

14S-50S

S of50S

Global

W-92/41-yr

W-92/1995

W-99/41-yr

W-99/1995

PgC

yr-1

( uunn )

NCEP(1995) 41 Year average Monthly

Sources of uncertainty

• Seasonal distribution of pCO2 (0.8 PgC)

• Estimate of skin temperature (-0.6 to –0.1 PgC)

• Estimates of the transfer velocity (20-40%)

• Estimates of windspeed (2 m/s)

How can we do better?

Factors influencing CO2 flux estimates

Wind

k pCO2

Air-Sea CO2Flux

SST

Transport

BiologyWindWaves

BubblesSurfaceFilm

Near SurfaceTurbulence

Bock et al. (1999)

Better spatial-temporal coverage

2. Predictions using synoptic data sets:

1. Deployment of ships and moorings:

time

space1 m2 1 km2 GlobeOcean

BasinRegional(106 km2)

centuries

decadal

Inter-annual

seasonal

daily

Remote sensing

Space and time coverage of ocean carbon observing networks

hourly

Process Studies

Repeat Trans-basin

Sections

VOS

surface pCO2

Shipboard

Time-Series

Moored

Time-Series

Factors influencing surface water pCO2

dSdTALKdTCOdTd

S

pCO2

TALK

pCO22

TCO2

pCO2

T

pCO2pCO2

Temperature (C) -2 –30 (ln pCO2/T) = 0.0423oC-1 400%

Variable Range Relation Effect

TCO2(mol kg-1) 1900-2200 (ln pCO2/Tln TCO2) = 10 400%

Alkalinity(mol kg-1) 2150-2350 (ln pCO2/Tln TALK) = -9.4 -200%

Salinity(mol kg-1) 33.5-37 (ln pCO2/Tln S) = 0.94 ~10%

Alkalinity and salinity are proportional and can be accounted for

Summer Fall

Winter Spring

Stephens et al., 1996

Temperature correlations

Prediction of pCO2

~

Bermuda

Courtesy of Nick Bates

~100 uatm

~9.5 C4.23% C-1

160 uatmDue to

temperature

dSdTALKdTCOdTd

S

pCO2

TALK

pCO22

TCO2

pCO2

T

pCO2pCO2

uatmdTCOTCO

pCO60~2

2

2

TCO2=33 mol/kg

Temp vs. Biology

Takahashi et al. (2002)

Tem

p. (

C)

CO2+H2O O2+CH2O Upwelling

PalmerSta.

MODIS

May 2001

Sea Surface Temperature

May 2001

Chlorophyll

PAR December 2000

Derived from GSFC Data Assimilation Office 3 hr retrievals.

http://modis-ocean.gsfc.nasa.govhttp://opp.gsfc.nasa.gov

Predicting pCO2

NPP

SST

Zmix

Estimates of gas transfer velocityWind

k

WindWaves

BubblesSurfaceFilm

Near SurfaceTurbulence

0

20

40

60

80

0 50 100 150

k(600)

[cm

·h-1]

Rn [mm·h-1]

k(600)0.929 0.679Rn 0.0015Rn2

Gas exchange vs. rain rate (MP distribution)

Ho et al. 1997

Summary

III. Improving our estimates of air-sea fluxes- Time-space distribution of pCO2

- Deployment of ships and buoys- Use of satellite measurements to calculate change in TCO2

- Parameterization of gas transfer velocity- micro-scale measurements

II. Estimates using surface pCO2:- Provide us with estimates of fluxes on a monthly basis based climatology adjusted for a single non-El Nino year- Errors in flux estimates occur due to lack of direct pCO2, wind speed and understanding of the gas transfer velocity

I. The air-sea flux measurement- Provide true short-term (~1 hr) measurements of flux which can be associated with wind speeds measured on that same time scale. - Are limited to areas of high pCO2

Inventory methods

• Estimates of integrated change in carbon inventory1) Time series approach

– Comparing measurements made between two time intervals

– Compare residuals of multiple parameter regressions using T, S, TALK and nutrients

2) C* Method– Estimate of the total inventory of anthropogenic carbon

in any given region

Hydrographic samplisg stations

C* Method (Gruber et al.)

C*

170O2

116CO2

Soft tissue

[O2]sat-O2 [O2]meas =0

T170 O216 NO32-

Carbonate

pCO2(i)=280

CaCO3

Ca2++CO32-

Cant = Cm – ∆Cbio – Ceq280 – Cdiseq = ∆C* - ∆Cdiseq

∆Cbio=rC:OO2+ ½(rN:OO2+CO32-)

Cdiseq

Ceq280

∆Cbio

Anthropogenic CO2

(mol kg-1)

(m

ol k

g-1)

Pre-industrial CO2

International CLIVAR/CO2 Lines (including US)

CO2 Clivar Repeat Hydro.