476
Andersen et al.
mined by calibration with core-top samples from
the South Atlantic (Fig. 5). To calculate ancient bvalues from sedimentary I)ISN, both correlations
shown in Figs. Sa and b were combined. The bvalues for the samples from core GeoB 1016-3
were derived from the measurements of bulk sediment 1)15N in the same core (Holmes et al. 1997).
By using eq. 3 with an E f = 25 %0, the CO 2 (aq)
concentrations obtained in this way were then converted to CO 2 partial pressure values (PC0 2 ) via
Henry's Law using the alkenone-derived sea surface temperatures (SST) and a constant salinity of
35 %0 (see eq. 1). The solubility coefficients were
calculated after Weiss (1974).
According to a propagation of errors, all calculation steps together result in an absolute error of
the paleo-PC0 2 estimates in total of about ± 36
Ilatrn (see Table 2). To determine the impact of
each variable, the PC0 2 was recalculated adding
one absolute error to each variable without changing the other parameters. The difference between
the two PC0 2 estimates is proportional to the importance ofthat variable on thePC0 2 calculation.
Parameter
Initial value
Absolute error
Unit
Impact on PC02 [liatm 1
T
298.15
1.0
K
11.5
Eb(a)
-8.97
0.11
%0
-11.3
o13Crp
1.50
0.05
%0
1.4
Vital effect
0.50
0.3
%0
8.6
Ob
2.00
0.3
%0
8.6
Od
-6.99
0.32
%0
9.3
I
Ol~
8.76
0.2
%0
-4.0
J
P0 4 (OI5N)
0.445
0.025
limol/l
4.0
PO. (eq.)
0.445
0.16
limolll
26.1
I
PO. (total)
0.445
0.162
limol/l
26.4
I
b-value (PO 4)
110
8.98
%0 limol
26.4
b-value (eq.)
110
5.90
%0 limol
17.3
b-value (total)
110
10.74
%0 !lmol
31.5
O13Calk
-24.1
0.2
%0
-4.8
Ealkenone
4.2
0.2
%0
-5.5
OP
-20
0.28
%0
-7.9
Ef
25
0.5
%0
-13.2
EP
13.28
0.43
%0
12.4
ce
9.39
0.86
limol/l
36.4
Salinity
35
1
%0
1.4
Ina
-3.54
0.02593
Mol 1. 1 atm· 1
-8.3
a
0.03
0.00075
Mol 1. 1 atm· 1
-8.1
PCO,
323
36.4
)1atm
Table 2. Error estimation of the paleo-PCO, reconstructions done by propagating the errors of each parameter
used. Equivalent intermediate steps are surrounded with a box. The impact on the caIculatedPCO" determined for
each parameter by adding its absolute error without changing the other parameters, corresponds to the importance
of each parameter on the final paleo-PCO, estimate.
Andersen et al.
mined by calibration with core-top samples from
the South Atlantic (Fig. 5). To calculate ancient bvalues from sedimentary I)ISN, both correlations
shown in Figs. Sa and b were combined. The bvalues for the samples from core GeoB 1016-3
were derived from the measurements of bulk sediment 1)15N in the same core (Holmes et al. 1997).
By using eq. 3 with an E f = 25 %0, the CO 2 (aq)
concentrations obtained in this way were then converted to CO 2 partial pressure values (PC0 2 ) via
Henry's Law using the alkenone-derived sea surface temperatures (SST) and a constant salinity of
35 %0 (see eq. 1). The solubility coefficients were
calculated after Weiss (1974).
According to a propagation of errors, all calculation steps together result in an absolute error of
the paleo-PC0 2 estimates in total of about ± 36
Ilatrn (see Table 2). To determine the impact of
each variable, the PC0 2 was recalculated adding
one absolute error to each variable without changing the other parameters. The difference between
the two PC0 2 estimates is proportional to the importance ofthat variable on thePC0 2 calculation.
Parameter
Initial value
Absolute error
Unit
Impact on PC02 [liatm 1
T
298.15
1.0
K
11.5
Eb(a)
-8.97
0.11
%0
-11.3
o13Crp
1.50
0.05
%0
1.4
Vital effect
0.50
0.3
%0
8.6
Ob
2.00
0.3
%0
8.6
Od
-6.99
0.32
%0
9.3
I
Ol~
8.76
0.2
%0
-4.0
J
P0 4 (OI5N)
0.445
0.025
limol/l
4.0
PO. (eq.)
0.445
0.16
limolll
26.1
I
PO. (total)
0.445
0.162
limol/l
26.4
I
b-value (PO 4)
110
8.98
%0 limol
26.4
b-value (eq.)
110
5.90
%0 limol
17.3
b-value (total)
110
10.74
%0 !lmol
31.5
O13Calk
-24.1
0.2
%0
-4.8
Ealkenone
4.2
0.2
%0
-5.5
OP
-20
0.28
%0
-7.9
Ef
25
0.5
%0
-13.2
EP
13.28
0.43
%0
12.4
ce
9.39
0.86
limol/l
36.4
Salinity
35
1
%0
1.4
Ina
-3.54
0.02593
Mol 1. 1 atm· 1
-8.3
a
0.03
0.00075
Mol 1. 1 atm· 1
-8.1
PCO,
323
36.4
)1atm
Table 2. Error estimation of the paleo-PCO, reconstructions done by propagating the errors of each parameter
used. Equivalent intermediate steps are surrounded with a box. The impact on the caIculatedPCO" determined for
each parameter by adding its absolute error without changing the other parameters, corresponds to the importance
of each parameter on the final paleo-PCO, estimate.