Air-Sea CO2 Fluxes in the Caribbean Sea

1
Air-Sea CO2 Fluxes in the Caribbean Sea Rik
Wanninkhof1, Are Olsen2, Joaquin Trinanes3, and
Betty Huss1
Project Ocean Carbon Sources and Sinks (using
VOS)- Nick Bates, BBSR Steve Cooke, AOML
Richard Feely, PMEL Frank Millero, RSMASTaro
Takahashi, LDEO and Rik Wanninkhof, AOML
Objective - Investigate methods to create flux
maps - Determine seasonal and inter-annual
variability Methology FCO2 k s ?pCO2
2
Status of Estimating Air-Sea CO2 Fluxes The
global CO2 climatology provides global pattern
and the global uptake by the ocean It does
not yield changes in regional fluxes over time
For an accurate global carbon accounting regional
estimates are necessary To attribute changes in
atmospheric growth rate we need annual estimates
of sources and sinks Goal Constrain
regional fluxes to 0.2 Pg C year-1
For this goal we need ?pCO2 to following
resolution ( equivalent to 0.1 Pg C yr-1)
d(?pCO2) a A-1 U-2
3
Creation of Flux Maps
Obtain data from ships (and other platforms)-
COSP program Create regional algorithms with
biogeochemical and physical parameters that
are measured at higher frequency and with
regional coverage Use high data coverage to
create flux maps that captures spatial and
temporal variability
Remotely sensed products used Q-Scat Wind for
gas transfer Reynolds SST for pCO2w (tied to
ship SST) Position different biogeochemical
provinces Chlorophyll
productivity Mixed layer depth
entrainment Salinity delineation of
provinces Sea surface height entrainment
4

• Testbed Caribbean Sea
• High resolution measurements (14 day repeat
  occupation)
• Good satellite coverage and auxiliary
  measurements (e.g. skin temp)

http//www.aoml.noaa.gov/ocd/gcc/explorer_introduc
tion.php
5
Caribbean Sea A well behaved test case pCO2w
a b SST C Lat D long
Yearly algorithms are very similar Small changes
can in large part be explained by rise in
atmospheric CO2
2002 fCO2sw 10.18 SST 0.5249 lat -
0.2921 lon 52.19, N 40204 rms 5.7
uatm, r2 0.87 2003 fCO2sw 10.45 SST
0.3629 lat - 0.5144 lon 36.52, N
60373 rms5.9 uatm, r2 0.92 2004 fCO2sw
10.64 SST 0.9745 lat - 0.3687 lon
30.68, N 86695 rms 7 uatm, r2 0.85
6
Interpolation Schemes
Kriging interpolation
Interpolation fCO2w f( SST, position)
7
Regional Patterns of Seasonal fluxes in 2004
Regional extent of algorithms based on data from
NOAA ship BROWN
8
Annual fluxes 2002- 2004
2002 -4.8x1011 mol 0.4 global
uptake 2003 -0.48x1011 mol 0.04 global
uptake 2004 -13.6x1011 mol 1 global
uptake (Global uptake 1.6 Pg C -1.3 1014 mol)
9

• Weekly Fluxes
• Small annual regional fluxes are caused
  compensating effect of local and temporal
• sinks and sources
• Weekly fluxes can be as as great as annual
  fluxes
• Most year-to-year variability is in late winter/
  early spring

10
Near Future Operational Future Operational
Autonomous platforms
SENUS- A self navigating Autonomous sailboat D.
Palmer AOML E. Schlieben FAU Utilize in situ
sensors like currently deployed on buoys
In dire need of VOS !!! Instrumentation Establis
hing an arrangement with General Oceanics to
build underway pCO2 systems
11
(Near) Future directions Flux maps of the North
AtlanticHeike Lueger- in collaboration with
Carbo-Oceans partners
Database Years 2002-2004 Data points 157,987
Nuka Arctica
Skogafoss
Falstaff
Ron Brown
12
Provinces in the North Atlantic
ARCT Atlantic Arctic SARC Atlantic
Subarctic NADR North Atlantic Drift GSTR Gulf
Stream NWCS Northwest Atlantic Continental
Shelf
ARCT / SARC
NADR
NWCS
GSTR
following Longhurst (1995)
13
Subpolar Region -
pCO2w 12SST - 0.4SST2 0.6MLD 0.7LON
0.8LAT 161 RMS14 R2 0.6
MLD- Argo profiling float data
Data points 5,177 (no April) Years
2003-2004 VOS Nuka Arctica Skogafoss
14
Conclusions

• 3-years of data suggest that the pCO2- SST
  algorithms are robust for the Caribbean
• Over 3 years very small net fluxes into the
  ocean but large changes over the year
• Most of the interannual variability appears in
  the Spring
• Method shows promise in the North Atlantic by
  inclusion of other parameters
• Completion of the surface ocean CO2 observing
  system is slowed by lack of platforms
• Alternate platforms should be investigated