www.sciencemag.org/cgi/content/full/science.1189338/DC1
Supporting Online Material for
Decrease in the CO2 Uptake Capacity in an Ice-Free Arctic Ocean Basin
Wei-Jun Cai,* Liqi Chen, Baoshan Chen, Zhongyong Gao, Sang H. Lee, Jianfang Chen,
Denis Pierrot, Kevin Sullivan, Yongchen Wang, Xinping Hu, Wei-Jen Huang, Yuanhui
Zhang, Suqing Xu, Akihiko Murata, Jacqueline M. Grebmeier, E. Peter Jones, Haisheng
Zhang
*To whom correspondence should be addressed. E-mail: [email protected]
Published 22 July 2010 on Science Express
DOI: 10.1126/science.1189338
This PDF file includes:
Materials and Methods
SOM Text
Figs. S1 to S3
Tables S1 and S2
References
Supporting Online Material
1. Materials and methods
1.1. Underway CO2 data of surface water and the atmosphere.
An underway CO2 system of Atlantic Oceanographic and Meteorological Laboratory (AOML,
Miami, Florida) was deployed on Icebreaker Xuelong. The system was described in detail by
Pierrot et al. (2009) (1). It yielded values within 1 μatm compared to the reference system in a
comprehensive inter-comparison performed in Japan in 2003. Surface pCO2 data beyond 78oN
are not reported due to potential data quality issues caused by serious clog of the water flow line
by the heavy ice during the survey in summer 2008. Atmospheric pCO2 measured by us and
recorded at station Point Barrow by NOAA (ftp://ftp.cmdl.noaa.gov/ccg/co2/GLOBALVIEW/) are
similar at about 375 μatm.
1.2. DIC and TA analysis
Discrete water samples were collected using a SeaBird SBE 9/11 + CTD system with a rosette
sampler. Samples were preserved with HgCl2 and stored in a refrigerator onboard ship. The
samples were subsequently shipped via air from Nome, Alaska to Cai’s laboratory in Georgia,
USA for analysis (samples were again kept in a refrigerator and analysis was completed within 7
weeks). DIC was measured in replicate (n≥3) on 0.5-mL of water sample by acidification and,
subsequently, quantification of the released CO2 using an infrared CO2 analyzer (2, 3). TA was
also measured in replicate (n=3 or 4) using open-cell Gran titration on 16-mL of sample. Both
analyses had a precision of better than 0.1% (±2 µmol kg-1) and were calibrated using the
Certified Reference Material (CRM) from A. G. Dickson. A recently opened CRM bottle was
used for calibration each day. 1.3. Carbon uptake rates of phytoplankton
Daily carbon uptake rates were estimated from six light depths (100, 50, 30, 12, 5, and 1 %
penetration of the surface photosynthetically active radiation, PAR), using a 13C isotope tracer
technique. Each light depth was determined after lowering with a secchi disk on the Xuelong
cruise in 2008. Seawater samples of each light depth were transferred from the Niskin bottles to
1L polycarbonate incubation bottles which were covered with stainless steel screens for each
light depth. Water samples were inoculated with a labeled carbon (NaH13CO3) substrate (4).
Bottles were incubated in a deck incubator cooled with surface seawater. The 4-5 hour
incubations were terminated by filtration through pre-combusted (450 °C) GF/F glass fiber filters
(24 mm). The filters were immediately frozen and preserved for mass spectrometric analysis at
the stable isotope laboratory of the University of Alaska Fairbanks, USA. Particulate organic
carbon and abundance of 13C were determined in the Finnigan Delta+XL mass spectrometer after
HCl fuming overnight to remove carbonate. Dark uptake rates were subtracted to correct carbon
values and then carbon uptake rates were averaged from two identical experiments of same
sampling periods.
2.
Supporting text
2.1. Selection of end-member DIC and TA values
Average values in 200-m slope waters at the north edge of the Bering Sea (59.0 to 60.4oN) are
used as the source seawater end-members. They are: S = 33.218±0.046, TA =
2257.9±16.2 μmol/kg, and DIC = 2161.4±3.4 μmol/kg. The river water end-member, TA = 1100
2 μmol/kg, is taken from the PARTNERS project (5). We set river DIC = 1150 μmol/kg, using the
TA/DIC ratio of 0.956 for river waters (6). Ice meltwater S, TA and DIC values are set as 5,
450 μmol/kg, and 400 μmol/kg respectively according to Rysgaard et al. (7), which are
equivalent to TA = 104.0 μmol/kg and DIC = 59.8 μmol/kg at 0 salinity. Any reasonable errors
associated with these end-member values won’t affect our discussion. The modified seawater,
where ice melt starts, is set at 30‰, with its TA and DIC values calculated from the seawaterriver mixing line.
2.2. Use of thermodynamic constants
CO2SYS (8) was used to carry out various CO2 system property simulations. Carbonic acid
dissociation constants K1 and K2 of Millero (9) and the solubility constant from Weiss (10) were
used in the program to simulate the mixing processes (Fig. 4 in the main text). A MatLab
version of the CO2SYS was used to simulate the dynamic process of CO2 invasion from the
atmosphere and the pCO2 increase in the surface mixed layer with time. In the this simulation,
carbonic acid dissociation constants of Mehrbach et al. (11) as refitted by Dickson and Millero
(12) were used to calculate pCO2 from DIC and TA.
2.3. Water column stratification
In the basin areas, from north to south along the two lines (roughly 148oW and 172oW, see
Fig.S1 for station locations), stratification increases with more meltwater in the south (see Fig.
S3). Along the 148 oW line (B85 to B23), limited by sample depth resolution, we can only
determine that the mixed layer depth was less than 30 m. From the 172 oW line (N01 to R15), we
determined that the mixed layer depth was roughly 20 m or slightly shallower. These results are
consistent with other observations of very a shallow mixed layer depth in the Canada Basin (13).
2.4. Rate of CO2 gas invasion during the ice-free period
We selected the mixed layer depth as 20 m according the measured density profiles (Fig. S3).
Measured salinity and TA data are used for the summer 2008 simulation. Then, corresponding to
the initial pCO2 of 225 μatm at -1.5oC, an initial DIC value at the early spring condition is
calculated using CO2SYS (the MatLab version). Next, the temperature during which the gas
exchange occurs is raised to 0 and 4oC respectively, and a new pCO2 at that temperature is
calculated. For the respective ΔpCO2 (air-sea), a CO2 invasion flux is also calculated using the
equation of Wanninkhof (14). Wind speed (7 m/s) observed in 1999 is used for the calculation
(15). This amount of CO2 for a time step is then added to the DIC inventory in the mixed layer.
Using the new DIC and TA, a new pCO2 is calculated for the next time step, and so on. One
could carry out a better simulation by defining or assuming how temperature would change over
time, but the above two simulation curves should bracket the CO2 dynamics in the mixed layer
reasonably well. The bottom line is that the pCO2 curve will approach the atmospheric pCO2
value after 90-100 days. Similarly, the 1999 conditions are simulated at 0oC and -1.5oC. For the
summer 1999 conditions, we use the salinity value to calculate an initial TA (scaled from 2008
TA). The 1999 initial DIC is calculated from this TA and an initial pCO2 of 225 μatm at -1.5oC.
We also calculated the total amount of CO2 uptake over a 100-day period from the DIC
increase in the mixed layer. DIC increases from an initial value of 1750.3 μmol/kg to 1790.3 and
1774.1 μmol/kg respectively at 0 and 4oC. The average of these two values (DIC increase = 31.9
μmol/kg) is presented in the main text. This value is converted to a water column total uptake by
multiplying the water the mixed layer depth (i.e., 31.9 mmol/m3*20 m = 637.3 mmol/m2).
3 Multiplying this flux by the total extra open water area of 0.6x1012 m2 in summer 2007, we
derived the total extra CO2 that can be taken up in that area as 4.6 x1012 gC/yr.
Ice-free period for 2008 was estimated from AMSR-E at the
http://www.iup.physik.uni-bremen.de:8084/amsredata/asi_daygrid_swath/l1a/n6250/
The ice-free condition started around July 12 at the 75oN/150oW area. However, early open areas
such as leads and polynyas at lower temperature and lower pCO2 must also permit effective CO2
gas exchange. Thus we assumed an effective ice-free condition for the summer of 2008 started
from early July. By the time of our survey from middle August to early September, an effective
ice-free period of 1.5- to 2-month is thus assumed. The ice-free condition at 75oN lasted until
October 20. Thus a total ice-free period of 90-100 days is assumed for the summer of 2008.
For the summer of 1999, the ice edge (ice concentration = 15%) was on the 74oN latitude
in the first week of September. Thus we assume a maximum ice-free condition for 1-month at
the survey time in the early and middle September of 1999 for the southern Canada Basin.
3. Supporting figures
Canada Basin
Chukchi Sea
Beaufort Sea
Figure S1. Cruise tracks of summer 2008 Chinese National Arctic Research Expedition
(CHINARE) and two earlier surveys. During the summer of 2008, I/B Xuelong arrived at 66oN
on August 1, 75oN on August 12, and 85oN on August 27.
4 Figure S2. Contrast of surface water pCO2 in 1994, 1999 and 2008. They are all normalized to
the 1994 average temperature (-1.6oC) using the equation of Takahashi et al. (16). Note that in
the margin areas, low pCO2 was mainly a result of strong biological uptake. Peter Jones was the
lead scientist responsible for the 1994 DIC and TA data from which pCO2 values are calculated.
Sigma_theta
Sigma_Theta
21.00
23.00
25.00
27.00
19.00
0
20
20
40
40
60
60
Depth (m)
Depth (m)
19.00
0
80
100
160
23.00
25.00
27.00
80
100
120
120
140
21.00
B85
B82
B77
B23
140
160
180
180
N01
P31
M07
R15
200
200
Figure S3. Density (Sigma_theta) - depth profiles. Sigmas_theta was calculated from salinity and
temperature data.
5 4. Supporting tables
Table S1. Water column primary production rates measured in summer 2008 (mmol m-2day-1).
Station locations are given in Fig. S1.
Canada Basin ice-free areas
B33
0.96
P25
0.86
B22
1.50
B13
2.38
S14
3.70
M05
2.02
S23
1.72
marginal sea areas
R03
393.2
R07
143.2
R09A
12.9
R13
35.2
C15
76.9
C17
115.8
C25
21.9
mean
SD
n
mean
SD
n
1.88
0.97
7
114
132
6
Table S2. Initial conditions used for simulations of atmospheric CO2 invasion.
mixing depth
salinity
water temp.
TA
DIC
p CO2
air p CO 2
wind speed
2008
20
24.78
‐1.5
1853.7
1750.3
225
2008
20
24.78
0
1853.7
1750.3
241.57
2008
20
24.78
4
1853.7
1750.3
290.61
1999
20
26.5
‐1.5
1982
1860.3
225
1999
20
26.5
0
1982
1860.3
241.62
375
7.3
375
7.3
375
7.3
360.1
7.3
360.1
7.3
5. Supporting references
S1.
D. Pierrot et al., Deep-Sea Research, Part II: Topical Studies in Oceanography 56, 512
(2009).
S2.
W.-J. Cai, Y. Wang, Limnology and Oceanography 43, 657 (1998).
S3.
Z. A. Wang, W.-J. Cai, Limnology and Oceanography 49, 341 (2004).
S4.
T. Hama et al., Marine Biology 73, 31 (1983).
S5.
L. W. Cooper et al., Geophysical Research Letters 35, (2008).
S6.
X. Guo et al., Continental Shelf Research 28, 1424 (2008).
6 S7.
S8.
S9.
S10.
S11.
S12.
S13.
S14.
S15.
S16.
S. Rysgaard, R. N. Glud, M. K. Sejr, J. Bendtsen, P. B. Christensen, Journal of
Geophysical Research 112, C03016 (2007).
D. L. Pierrot, E., D. W. R. Wallace, MS Excel Program Developed for CO2 System
Calculations. (Oak Ridge National Laboratory, 2006).
F. J. Millero, T. B. Graham, F. Huang, H. Bustos-Serrano, D. Pierrot, Marine Chemistry
100, 80 (2006).
R. F. Weiss, Marine Chemistry 2, 203 (1974).
C. Mehrbach, C. H. Culberson, J. E. Hawley, R. M. Pytkowicz, Limnology and
Oceanography 18, 897 (1973).
A. G. Dickson, F. J. Millero, Deep-Sea Research 34, 1733 (1987).
L. A. Codispoti, C. N. Flagg, J. H. Swift, Deep-Sea Research, Part II: Topical Studies in
Oceanography 56, 1144 (2009).
R. Wanninkhof, Journal of Geophysical Research 97, 7373 (1992).
A. Murata, T. Takizawa, Continental Shelf Research 23, 753 (2003).
T. Takahashi et al., Deep-Sea Research, Part II: Topical Studies in Oceanography 49,
1601 (2002).
7