CO 2 Exchange Over Sea Ice: Questions and Answers After 6 years of Observation

CO2 Exchange Over Sea Ice: Questions and Answers After 6 years of Observation

Tim PapakyriakouCEOS, Faculty of Environment, Earth and Resources

University of ManitobaWinnipeg MB, Canada

[email protected]

09/09/10 Los Alamos: Sea Ice Salinity Workshop 1

Collaborators:• Lisa Miller (DFO, Victoria)• Jean-Louis Tison (ULB)• Bruno Delille (U. Liège)• Graduate Students

Outline

• Background and motivation for work• Overview of methods• Observed fluxes: Chambers• Controlling Factors• Observed fluxes: Micrometeorological• Summary and final comments

09/09/10 Los Alamos: Sea Ice Salinity Workshop 3

Los Alamos: Sea Ice Salinity Workshop 4

Motivation

• Winter and fall sea ice maxima are between 18 and 28 million km2

• Uncertainty of the space/time dimensions of CO2 exchange over sea ice, or whether significant fluxes exist

• Gosink et al. (1976) demonstrate that warm ice (>-7) is reasonably permeable to CO2 diffusion

• No established physically-based methodology to represent CO2 exchange over sea ice, or in ice dominated waters

• Incomplete understanding of carbon budgets and the atmosphere’s role in the carbon dynamics of polar marine ecosystems

09/09/10

Takahashi et al (2009)

Krembs et al. (2002)

Mundy et al. (2007)


Motivation

09/09/10

Drobot et al., 2008

Scales of Ice Concentration

•Open Water

€

FCO2 = kSCO2 pCO2sw − pCO2air[ ]

pCO2sw

pCO2air

S

k

Current Practice

•Scaled according to ice concentration

€

FCO2 = 1− A[ ]∗kSCO2 pCO2sw − pCO2air[ ]

pCO2sw

pCO2air

“ice capping”

A

Vast majority of air-surface exchange studies.(e.g., Bates et al., 2006, GRL)

Immediate Objectives

• Assess the validity of the ice-capping assumption;• Develop an strong understanding air - sea ice CO2 exchange;

– seasonality?– consistently among regions?– parameterizations?

• Subsequent Activity– Assess an alternative:

• flux – aggregation? (Lüpkes and Birnbaum, 2005, B-L Meteor 2005)

€

FCO2 = 1− A[ ]∗keff SCO2 pCO2sw − pCO2air[ ][ ] + A∗Fsi


Methods: CO2 Flux

• Net ecosystem exchange vs. net accumulation

09/09/10


Methods: CO2 Flux

• Net ecosystem exchange vs. net accumulation– cumulative measure of flux– open- and closed-path eddy covariance (micrometeorological)– chamber

09/09/10


Methods: CO2 Flux

• Net ecosystem exchange vs. net accumulation– cumulative measure of flux– open- and closed-path eddy covariance (micrometeorological)– chamber– box model - monitor change in system carbon stock (organic and

inorganic) over a period of time

09/09/10


Methods: Chambers

• Flux is calculated using the time rate of change of gas concentration in the chamber

• Introduces bias errors, including perturbations of local pressure, wind and CO2 concentration fields

• valuable information on feature efflux/uptake characteristics

09/09/10


Methods: Eddy Covariance Technique

• Simultaneous measurement of the high frequency (10 Hz to 20 Hz) fluctuations in vertical wind velocity and CO2 concentration from above the surface;

• Time-averaged covariance is directly proportional to the flux;• Sensor height to upwind fetch ratio of between approximately 1/100 to

1/300• Good measure of ecosystem exchange (10% to 20% accuracy), but little

information on surface sources or sinks

09/09/10


Methods: Open- and Closed Path Systems

• Difference relates to the style of infrared gas analyzer (IRGA):– open path measurements

measures across an ‘open’ optical path

– air is drawn through a tube into a ‘closed’ cell for measurement in closed systems

– both measure molar density• Corrections: (i) ‘spectral loss’

for both systems; (ii) dilution (open path) & sensor heating (open path)

09/09/10


Overview of Chamber Measurements

09/09/10

April – June 2008First year sea ice

Oct. 2007First year pack ice

Nov. – Dec. 2004First and multi year pack ice

AA03 V1 cruise Sept. - Oct. 2003First year pack ice

IPY BASICS


Chamber Measurements: CFL and Antarctic Sites

• Measurements over first-year sea ice cleared of snow;

• Fluxes range between 0.02 μmol m-2 s-1 and -0.06 μmol m-2 s-1

; • Little flux activity was observed

below a surface temperature of -10°C;

• Fluxes responded to the air-surface pCO2 gradient – direction and magnitude;

• Ice temperature exerted a strong control over sea ice pCO2.

09/09/10

black = CFL; blue = AntarcticGeilfus et al., 2010


Chamber Measurements over New Ice

• New ice supported CO2 efflux • SIMBA: flux ranged between 0.03

μmol m-2 s-1 and 0.06 μmol m-2 s-1 with colder ice supporting larger fluxes

• BASICS: flux ranged between 0.12 μmol m-2 s-1 and 0.5 μmol m-2 s-1

09/09/10

A B C D E-2.5

0.0

2.5

5.0over iceover snow

Air-

ice

CO

2 flu

xes

(mm

ol m

-2d

-1)

-9

-8

-7

-6

-5

-4

Inte

rfac

e te

mpe

ratu

re (°

C)OLD ICENew ice

SnowFrost Flowers

EDCBA

Hi between 25 and 27 cm


Role of Snow: Chamber Results

• Springtime measurements in the Sea of Okhotsk (ice thickness between ~ 42 cm and 45 cm) (Nomura et al. 2010);

• Observed fluxes ranged between -1.18 μmol m-1 s-1 and -0.37 μmol m-1 s-1 from sea ice without a snow or slush cover and and responded to the air ice surface pCO2 gradient;

• Fluxes were suppressed beyond critical snow thickness of ~ 9 cm;

• Combined snow and slush of < 9 cm reduced the uptake to ~ 50% of snow-free values, while the effect on efflux was less pronounced;

• Suggested that refreezing of melt water and resulting ice layer impeded the exchange.

09/09/10


Response of Brine pCO2 to ΔT

• Flux is dictated by the response of brine pCO2 to changing temperature:– On freezing/melting:

• increase/decrease in brine pCO2 within sea ice

• results from decrease/increase in brine solubility with increasing/decreasing brine salinity.

• thermal cycling in brine could induce covection

09/09/10

Thomas et al. (2010)

Geilfus et al. (2010)

Controls Over Observed Fluxes


Response of Brine pCO2 to CaCO3 Formation

• On cooling/warming:– possible

precipitation/dissolution of CaCO3

– first salt to precipitate at T=-2.2– observational evidence of ikaite

in both Arctic and Antarctic sea ice (Dieckman et al. 2008; 2010)

– thermal cycling in brine could cause CaCO3 to precipitate and dissociate;

09/09/10

2 HCO3- + Ca2+ ↔ CaCO3 + CO2

CaCO3

Assur (1960)


Los Alamos: Sea Ice Salinity Workshop 2109/09/10

“Closed system”Precipitation of CaCO3 and release of CO2

BUT during the spring – summer:

Dissolution of CaCO3 and CO2 consumption

Balance null during the period

“Open System”Precipitation of CaCO3

IF CaCO3 and CO2 are released in the water column

Balance null or source if crystal CaCO3 sink fast and deep

IF CaCO3 stays in the ice and CO2 is released in the water column and passed below the pycnocline:

The system is will be a CO2 sink.

-1.9°C

“Surface process”“Open System” Near surface formation in young ice or frost flowers allows CO2 to be released to the atmosphere → Source

2 HCO3- + Ca2+ ↔ CaCO3 + CO2

Courtesy: Tison


Role of Biology

• Primary Production– Sea ice alga typically concentrate within the bottom 20 cm of sea ice;– Production varies regionally within the Arctic and Antarctica;

• Arctic (mean Cla = 31 mg m-2; range between <10 and 300 mg m-2);• Antarctica (mean Cla=170 ; range betweeen <10 and >1000 mg m-2).

Arrigo et al. (2010)• Respiration

– bacteria are ubiquitous in sea ice • bacterial production can exceed primary production in the fall, winter, and summer;• high surface and near surface concentrations can be found in frost flowers

associated with young sea ice.

Deming (2010)

09/09/10



Overview of Eddy Covariance Measurements

1. Semiletov et al. (2004)

• Spring and summer open- path eddy covariance measurements w/o sensor heating correction;

• Landfast sea ice Barrow, Alaska;• CO2 fluxes between -1 μmol m-2 s-1

and 0.6 μmol m-2 s-1;• Fluxes are 2 to 3 larger than

measured by chambers; • Flux follows decreasing brine pCO2

concentration with depth.

09/09/10



2. Zemmelink et al. 2006

• Summertime open-path eddy covariance measurements w/o sensor heating correction;

• MY sea ice Weddel Sea• CO2 fluxes between -0.35 μmol m-

2 s-1 and 0.1 μmol m-2 s-1;• Flux was controlled by algae

production at a slushy snow ice interface.

09/09/10



3. Papakyriakou and Miller (submitted)• Springtime (open-path) eddy covariance measurements over first-year

sea ice in the Canadian Arctic Archipelago;• CO2 fluxes between -3.0 μmol m-2 s-1 and 1 μmol m-2 s-1;

• CO2 evasion is in line with Semiletov et al. (2004), but maximum uptake was a factor of 5 greater.

09/09/10



• Early Season Efflux (ave = 0.36 μmol m-2 s-1):

09/09/10

• Less confident in results: (i) large contribution by heating correction (over corrected?); (ii) turbulent mixing is weak;

• CO2 concentration of ~ two times atmospheric value is required at the snow base to support an average early experiment efflux by diffusion, which is realistic (e.g., Delille et al 2007; Heinesch et al. 2010; Miller et al., submitted);

• Possible CO2 sources: bacterial respiration (DOC, EPS), DIC stock in snow and saline upper ice.

T (C)

Vb (%)



• Late Season Uptake:

09/09/10

• Heating correction is a small component of the flux;

• High wind speed facilitates strong turbulent mixing, ventilates the snow and promotes blowing snow:

• high surface area to mass ratio;• rapid turbulent exchange;

• Occurs when sea ice is hydrologically ‘open’;

• Regional sea ice is tremendously productive, approaching between 28 and 55 mg C m-2 day-1 (data from Lavoi et al., 2005)• Contribution of atmospheric CO2?

T (C)

Vb (%)



09/09/10

• Thermodynamics appears to drive an air-ice pCO2 gradient by affecting brine carbonate speciation, and sea ice permeability; and

• High wind is necessary to facilitate the exchange


Contrast to Closed Path Observations

• Heinesch et al (2010) report closed- path eddy covariance measurements over landfast sea ice near to Barrow;

• Winter fluxes ranged between 0 and 0.8 μmol m-2 s-1, while fluxes in the early spring fluxes range between ± 0.2 μmol m-2 s-1

• Efflux rates are comparable to Papakyriakou and Miller, however maximum uptake is an order of magnitude smaller;

• Consistent relationships to thermodynamics.

• Regional differences in sea ice biogeochemistry, or uncorrected biases associated with the respective systems?

09/09/10


Summary

• CO2 fluxes have been measured using difference techniques, in different hemispheres, and appear to respond in direction and magnitude to temperature sensitivity of brine carbonate equilibria

• Chambers typically show the smallest absolute fluxes.• Early eddy covariance measurements tended to show uptake – perhaps because

of a missing sensor heating correction.• Recent eddy covariance measurements using different systems conform for

efflux, but not uptake. Perhaps variation in regional sea ice biogeochemistry can account for part of the discrepancy.

• Relationships are observed between fluxes and snow sea ice thermodynamics using different EC flux systems.

• Little is known about the role of saline snow on the carbon exchange dynamics. Measurements suggest that mobile snow may play an important role in CO2, particularly when sea ice cold and assumed “closed” to brine drainage.

09/09/10


Answers and Questions

• There appears episodic and sometimes large CO2 exchange over sea ice.• Fluxes appear related to sea ice thermodynamics and temperature-brine

relations.• Which methodology consistently provides accurate flux measurements, or

should they be expected to?• How geographically relevant are the fluxes reported here? • Can sea ice carbon stocks (organic and inorganic) support observed fluxes,

and what exactly is the source of the observed efflux?• What is the fate of atmospheric carbon taken into the snow/sea ice layer?

09/09/10

Winnipeg, MB11/02/09

Thank You. Questions?

Documents

CO 2 Exchange Over Sea Ice: Questions and Answers After 6 years of Observation