Quantifying Seasonal to Interannual Changes in AirSea Carbon Dioxide Fluxes

1
Quantifying Seasonal to Interannual Changes in
Air-Sea Carbon Dioxide Fluxes

• Rik Wanninkhof1, Taro Takahashi2, Christopher L.
  Sabine3, Richard A. Feely3, Frank J. Millero4,
  Nick Bates5, Joaquin Trinanes6
• 1Atlantic Oceanographic and Meteorological
  Laboratory, NOAA, 4301 Rickenbacker Causeway,
  Miami FL 33149, E-mail Rik.Wanninkhof_at_noaa.gov
• 2Lamont-Doherty Earth Observatory of Columbia
  University - Palisades NY 10964,
• 3Pacific Marine Environmental Laboratory, NOAA,
  7600 Sand Point Way NE, Seattle WA 98115
• 4Rosenstiel School of Marine and Atmospheric
  Science, University of Miami, 4600 Rickenbacker
  Causeway, Miami FL 33149
• 5 Bermuda Institute of Ocean Science St.
  George's, GE 01 Bermuda
• 6 Cooperative Institute for Marine and
  Atmospheric Science, University of Miami, 4600
  Rickenbacker Causeway, Miami FL 33149

2
Outline

• Approach
• Highlights of the updated climatology of
  Takahashi et al. (2008)
• Change of method to quantify seasonal air-sea CO2
  fluxes
• Social Global interactions and collaborations
• Scientific Flux maps by proxy
• Initial products and results

3
Approach Estimate flux by bulk method Fnet
k s (pCO2w-pCO2a) Obtain the net air-sea CO2
flux by measuring the air water partial pressure
difference of pCO2 (?pCO2) and relating k to
windspeed. OCO programs focussed on measuring
?pCO2 Surface water pCO2 measurements from
ships Wanninkhof, Goni (AOML), Feely (PMEL),
Takahashi (LDEO), Bates (BIOS), Millero
(RSMAS) High-resolution ocean and atmosphere pCO2
time series measurements Sabine (PMEL) , Chavez
(MBARI), Bates (BIOS) Global Carbon Data
Management and Synthesis Project Sabine,
Feely, Hankin (PMEL), Wanninkhof, Peng (AOML),
Key (Princeton), Kozyr (ORNL), Millero
(RSMAS U.Miami), Dickson (SIO)
4
Net flux versus anthropogenic CO2 flux
Models Anthropogenic Air-sea CO2
flux Observation Air-sea CO2 estimates are net
fluxes
Net flux Pre-anthro flux Anthro flux -
1.6 0.4 - 2.0
Fnet k s (pCO2w-pCO2a) Fanthro k s
(pCO2w-ant-pCO2a-ant) pCO2a pCO2a-ant
pCO2a-pre-ant pCO2w pCO2w-ant
pCO2w-pre-ant Seasonal to Interannual changes
in air-sea CO2 fluxes are due to climate
induced changes in the natural cycle
Sarmiento, J. L., and Gruber, N., 2002
5
(Revised July 1, 2008) Climatological Mean and
Decadal Change in Surface Ocean pCO2, and Net
Sea-air CO2 Flux over the Global Oceans Taro
Takahashi, Stewart C. Sutherland, Rik Wanninkhof,
Colm Sweeney, Richard A. Feely, David W. Chipman,
Burke Hales, Gernot Friederich, Francisco Chavez,
Christopher Sabine, Andrew Watson, Dorothee C. E.
Bakker, Ute Schuster, Nicolas Metzl, Hisayuki
Yoshikawa-Inoue1, Masao Ishii, Takashi
Midorikawa, Yukihiro Nojiri, Arne Körtzinger,
Tobias Steinhoff, Mario Hoppema, Jon Olafsson,
Thorarinn S. Arnarson, Bronte Tilbrook, Truls
Johannessen, Are Olsen, Richard Bellerby, C. S.
Wong, Bruno Delille, N. R. Bates and Hein J. W.
de Baar
The annual mean for the contemporary net CO2
uptake flux over the global oceans is estimated
to be -1.6 1.0 Pg- C yr-1. Taking the
pre-industrial steady state ocean source of 0.4
0.2 Pg-C yr-1 into account, the total ocean
uptake flux including the anthropogenic CO2 is
estimated to be - 2.0 1.0 Pg-C yr-1 in 2000.
The entire database including the coastal and El
Nino period equatorial Pacific data (Takahashi et
al., 2008) are available at the Carbon Dioxide
Information and Analysis Center at the Oak Ridge
National Laboratory, Oak Ridge, TN (LDEO database
(NDP- 088) at http//cdiac.ornl.gov/oceans/doc.htm
l).
6
2008 Climatology Focus on Southern Ocean
Data Subpolar Southern Ocean The observations
made in the Southern Ocean south of 50?S
increased significantly not only in numbers from
0.4M in the database used in Takahashi et al.
(2002) to about 1.1 M in this database, but also
in geographic coverage
Figure 1 (A) Sampling locations. The black
dots indicate the measurements in the 0.94M
database used in Takahashi et al. (2002) and the
red dots are new measurements added to the
database (3.0 M) used in this study. (B) Number
of months in each 4 x 5 box area where at
least one surface water pCO2 measurement has been
made since the early 1970s. White areas have no
measurements.
The Drake Passage areas are the only southern
high latitude boxes that have 12-months of data.
7
Comparison with previous work
February
Difference between the surface water pCO2 values
from this study (3.0 M) and the 2002 study
(0.94M) are shown (A) February and (B) August.
August
SEA-AIR CO2 FLUX (Pg-C yr-1) Zone
2008 2002 14?S-50?S -1.05 -1.06
50?S-62?S -0.06 -0.34 S of 62?S 0.01 -0.04
8
Rate of increase of surface water pCO2

On the basis of the surface water pCO2 data
measured during the past three decades, the mean
rate of increase in about 27 of the global ocean
areas is estimated to be in a range of 1.26 (
0.55) to 2.13 ( 0.64) µatm yr-1
boxes Area rate ppm/yr SD N. Pacific
(10?N-60?N) 28 28.9 1.28 0.46 Equatorial
Pacific (5?N-5?S) 2 7.4 1.26 0.55 N.
Atlantic (15?N-70?N) 36 16.8 1.80 0.37 S.
Pacific (15?S-55?S) 6 2.9 1.53 0.35
Southern O. (50?S-60?S) 6 30.6 2.13 0.64
------------------------------------- -------
------- ------- ------- Global with Southern
Ocean 78 86.6 (27) 1.69 0.51 Global without
Southern Ocean 72 56.0 (17) 1.45 0.47
9
Rate of increase of surface water pCO2
Rates of increase in surface water pCO2 south of
50?Sthe winter months (Julian dates from 172 to
326) is greatest for SST 1.5-2.5 C
Rates of increase in surface water pCO2 in the
subpolar region (south of 50?S) of the Southern
Ocean in the winter months (Julian dates from 172
to 326) during the period from 1986 to 2006. The
top panel is for SST between 0.8 to 1.5?C the
middle panel for SST between 1.5 and 2.5?C and
the bottom panel for SST between 4.5 and 5.5?C.
Black dots indicate measurements made in non-El
Nino periods and green dots indicate those made
in El Nino periods. There is no discernible
difference between these periods. The mean rates
are estimated by linearly regressing all
individual data and N is the data count.
10
Attribution of Uncertainties in Global Fluxes
Fnet k s (pCO2w-pCO2a)
The net global sea-air flux may be subject to
random errors of 0.18 Pg-C yr-1 ( 13)
from the ?pCO2 measurements, 0.42 Pg-C yr-1
( 30) in the scaling factor for the gas
transfer piston velocity parameterization,
0.28 Pg-C yr-1 ( 20) in wind speeds
0.5 Pg-C yr-1 ( 35) for the mean rate of
change in ocean water
The climatological mean for the net CO2 uptake by
the global ocean in the reference year 2000 is
estimated to be in a range of 1.6 1.0 Pg-C
yr-1. Counting the pre-industrial steady state
sea-air CO2 flux of 0.4 0.2 Pg-C yr-1 the
oceanic uptake flux of CO2 including
anthropogenic CO2 is 2.0 1.0 Pg-C yr-1
11
Improvements in gas transfer velocity estimates
Fnet k s (pCO2w-pCO2a)
0.42 Pg-C yr-1 ( 30) in the scaling factor
for the gas transfer piston velocity
parameterization,

• Decent global constraint- weak regional
  constraints- local gas transfer studies (SO Gas
  Ex)
• Improvement in gas exchange-wind speed
  relationships Hybrid model
• Improvement in gas exchange relationships
  physically-based gas transfer model derived from
  the NOAA COARE air-sea flux parameterization
  (SOLAS-COST, k-conundrum workshop)
• Utilize Objectively Analyzed Air-Sea Flux
  (OAFlux) database (Yu and Weller) and include k.

• k a b ltU10gt c ltU102 gt d ltU103gt.
• Chemical enhancement Solid wall Stress Energy
  dissipation

k ua (?w/?a)-1/2 (hw Scw½ ln
(zw/dw)/?)-1 and hw ? Rr1/4/???
12
A Change in Strategy to Obtain Seasonal Fluxes
To constrain regional fluxes to 0.2 Pg C yr-1 we
need CO2 observations monthly measurements
every 500-1000 km 5-10 fold increase in ships
of opportunity (globally) 10-20 fold increase
in moorings in the right locations
Adaptation Utilize known (and to be determined)
correlations of pCO2 with physical and
biogeochemical parameters that are available at
higher resolution to establish flux fields.
Different approaches neural networks, self
organizing maps, models, semi-empirical
approaches. And- we cannot do so alone
13
Global relationships with SST- a first step
14
Using SST Proxy-for interannual variability

• Over last two decades amplitude of variability
  0.2 Pg C
• Strong correlation with climate indices such as
  ENSO
• SST good proxy in subtropical gyre but not at
  high latitude (need
• additional parameters (e.g. mixed layer depth)

15
Utilization in model comparison/validations
Thomas et al. (2008) Apparent decrease in sink in
North Atlantic over last decade caused by changes
in wintertime NAO from predominantly postive in
the early 90ties to neutral in recent years
16
Proposed application Incorporate in Carbon
Tracker product
Atmospheric inversions have poor ocean
constraints and hide discrepancies in ocean
sources and sinks
17
Buoys Assess frequency response of pCO2 for
different locations
Spectral analysis indicates that variability is
dominated by two frequencies Tropical
Instability Waves with frequencies of 17-35
days Kelvin Waves with frequencies of 53-60
days Kelvin waves show the most energy in SST,
but the TIWs show more energy in xCO2 Sabine et
al.
18
Data assembly and Collation SOCAT- Surface
Ocean CO2 Atlas Coordination through to IOCCP
spearheaded by Carbo-Ocean effort, data served
using LAS though CDIAC

• Envisioned as a sustained effort of providing
  documented and quality
• controlled data and data products
• Large community buy-in
• NOAA-COD and CDIAC only sustained component

4 million data points and counting
19

• Closing Thoughts and Issues
• A significant portion of the mooring and ship of
  opportunity work done
• in projects funded by NSF-polar, NASA,
  GCC-Polar programs/OAR and
• are of finite duration
• Continued issues finding and maintaining ships
  of opportunity
• Weak interactions with other NOAA-COD activities
• No sustained funding for data collation
• Limited resources available for incorporating
  COARE-k estimates in appropriate databases