588
Frederichs et al.
terms of a variable sediment source, transport or
alteration.
Magnetic volume susceptibility is the most
widely used parameter to quantify concentration
changes of the magnetic mineral fraction in marine
sediments. Susceptibility records are generally
interpreted as proxies for the ratio of terrigenous
(magnetite bearing) to biogenic (diamagnetic)
deposition. In this respect not only continental
sources must be considered. Outcropping weathered ocean floor basalt may also release considerable quantities oftitanomagnetite, e.g., in regions
of tectonic activity or high bottom current erosion.
Magnetite biomineralization and diagenesis may
further complicate the situation. With these possibilities in mind, the K log of core GeoB 1505-2 reads
as a basically climate induced signal which carries
additional features (pronounced minima) in some
of the sub-stages mentioned before.
To unravel such a composite record the following approach has been developed (von Dobeneck
1998). The observed susceptibility Kobs is regarded
as the sum of partial susceptibilities corresponding
to discrete particle or mineral fractions. These partial susceptibilities are determined by calibrating the
concentration related (linear) parameters (Table 1).
The scaling coefficients, which link the basically
uncalibrated individual parameter to the respective
partial susceptibility, can be found by multiple linear regression provided the various signal patterns
are sufficiently dissimilar. With indices corresponding to the fractions defined in Table 1 and to paramagnetism and diamagnetism, the observed susceptibility
is approximated by a predicted susceptibility
Kpre = ~o+ ~l·Kfd + ~2· Mar + ~3· Mjr+ ~4· Mhir
It is obvious that Ksp and ~1 - Kfd, Kmd and
~3 -MiT> as well as Khe and ~4· Mhir are corresponding terms, whereas Ksd contributes to both
~2 -Mar and 133 -Mir. Using a more MD particle
selective parameter such as low field remanence
(e.g., M ir of -0.01 T) would avoid this ambiguity.
For the sake of simplicity the original parameter
Mir of 0.1 Tisretainedhere, to some extent overestimating Kmd at the expense of Ksd. The paramagnetic susceptibility Kpara carmot be estimated
from the considered remanence parameters, but
defmitely contributes to Kobs and must have an
expression in Kpre.
If the fluxes of paramagnetic minerals are
linked to one or several magnetic fractions (e.g.,
weathering products such as paramagnetic clays
and antiferromagnetic hematite are often associated and have proportional concentrations), the
regression calculation will include their susceptibility by increasing the coefficient ~i of the respective magnetic fraction. Otherwise Kpara will contribute to the constant ~o. The diamagnetic susceptibility Kdia is largely independent of the terrigenous
fraction and can therefore be associated with ~o.
The regression results emphasize the sensitivity
of this approach. Goodness-of-fit R equals 0.986
and the standard error of the estimate is 8.7-10- 6 SI.
Predicted and observed susceptibilities (Fig. lla)
match very well, largest discrepancies occur at
sharp maxima. The constant ~o = -12.4-10- 6 SI
convincingly coincides with Kdia for a mixture of
CaC0 3 and H 2 0. Also the other coefficients ~1
to ~4 (Fig. 10) appear altogether plausible from a
rock magnetic point of view. The relation
Ksp = 3.97 Kfd implies that for purely superparamagnetic particles we should have Kfd IKsp = 113.97
= Kfd/K = 0.25, i.e., Kfd% = 25%. Dearing et a1.
(1996) give a theoretical value ofl4-16%for Kfd%,
suggesting that our regression moderately underestimates Ksp. The relationship Ksd = 35.10- 5 Marl
Mo (unit magnetization Mo = 1 Aim) results in a
SD specific Karm IKsd ratio of 94 which is distinctly
higher than the expected value ('" 60; Oldfield
1994). As mentioned before, a too small Ksd is
determined here, because the SD fraction also contributes to M ir and therefore to Kmd. The hematite
susceptibility Khe = 16-10- 5 Mh;,lMo compares
fairly well to a Khe = 3-10- 5 Msir IMo (Thompson
and Oldfield 1986) as the Mhir acquired between
0.3 and 0.8 T represents only a portion of the total
Msir· Finally, Kmd = 9.10- 5 Mir IMo corresponds to
a reasonable mean MD magnetite grain-size of
1 0 ~m (Thompson and Oldfield 1986).
With the concept of partial susceptibilities, the
K record of core GeoB 1505-2 can now be
Frederichs et al.
terms of a variable sediment source, transport or
alteration.
Magnetic volume susceptibility is the most
widely used parameter to quantify concentration
changes of the magnetic mineral fraction in marine
sediments. Susceptibility records are generally
interpreted as proxies for the ratio of terrigenous
(magnetite bearing) to biogenic (diamagnetic)
deposition. In this respect not only continental
sources must be considered. Outcropping weathered ocean floor basalt may also release considerable quantities oftitanomagnetite, e.g., in regions
of tectonic activity or high bottom current erosion.
Magnetite biomineralization and diagenesis may
further complicate the situation. With these possibilities in mind, the K log of core GeoB 1505-2 reads
as a basically climate induced signal which carries
additional features (pronounced minima) in some
of the sub-stages mentioned before.
To unravel such a composite record the following approach has been developed (von Dobeneck
1998). The observed susceptibility Kobs is regarded
as the sum of partial susceptibilities corresponding
to discrete particle or mineral fractions. These partial susceptibilities are determined by calibrating the
concentration related (linear) parameters (Table 1).
The scaling coefficients, which link the basically
uncalibrated individual parameter to the respective
partial susceptibility, can be found by multiple linear regression provided the various signal patterns
are sufficiently dissimilar. With indices corresponding to the fractions defined in Table 1 and to paramagnetism and diamagnetism, the observed susceptibility
is approximated by a predicted susceptibility
Kpre = ~o+ ~l·Kfd + ~2· Mar + ~3· Mjr+ ~4· Mhir
It is obvious that Ksp and ~1 - Kfd, Kmd and
~3 -MiT> as well as Khe and ~4· Mhir are corresponding terms, whereas Ksd contributes to both
~2 -Mar and 133 -Mir. Using a more MD particle
selective parameter such as low field remanence
(e.g., M ir of -0.01 T) would avoid this ambiguity.
For the sake of simplicity the original parameter
Mir of 0.1 Tisretainedhere, to some extent overestimating Kmd at the expense of Ksd. The paramagnetic susceptibility Kpara carmot be estimated
from the considered remanence parameters, but
defmitely contributes to Kobs and must have an
expression in Kpre.
If the fluxes of paramagnetic minerals are
linked to one or several magnetic fractions (e.g.,
weathering products such as paramagnetic clays
and antiferromagnetic hematite are often associated and have proportional concentrations), the
regression calculation will include their susceptibility by increasing the coefficient ~i of the respective magnetic fraction. Otherwise Kpara will contribute to the constant ~o. The diamagnetic susceptibility Kdia is largely independent of the terrigenous
fraction and can therefore be associated with ~o.
The regression results emphasize the sensitivity
of this approach. Goodness-of-fit R equals 0.986
and the standard error of the estimate is 8.7-10- 6 SI.
Predicted and observed susceptibilities (Fig. lla)
match very well, largest discrepancies occur at
sharp maxima. The constant ~o = -12.4-10- 6 SI
convincingly coincides with Kdia for a mixture of
CaC0 3 and H 2 0. Also the other coefficients ~1
to ~4 (Fig. 10) appear altogether plausible from a
rock magnetic point of view. The relation
Ksp = 3.97 Kfd implies that for purely superparamagnetic particles we should have Kfd IKsp = 113.97
= Kfd/K = 0.25, i.e., Kfd% = 25%. Dearing et a1.
(1996) give a theoretical value ofl4-16%for Kfd%,
suggesting that our regression moderately underestimates Ksp. The relationship Ksd = 35.10- 5 Marl
Mo (unit magnetization Mo = 1 Aim) results in a
SD specific Karm IKsd ratio of 94 which is distinctly
higher than the expected value ('" 60; Oldfield
1994). As mentioned before, a too small Ksd is
determined here, because the SD fraction also contributes to M ir and therefore to Kmd. The hematite
susceptibility Khe = 16-10- 5 Mh;,lMo compares
fairly well to a Khe = 3-10- 5 Msir IMo (Thompson
and Oldfield 1986) as the Mhir acquired between
0.3 and 0.8 T represents only a portion of the total
Msir· Finally, Kmd = 9.10- 5 Mir IMo corresponds to
a reasonable mean MD magnetite grain-size of
1 0 ~m (Thompson and Oldfield 1986).
With the concept of partial susceptibilities, the
K record of core GeoB 1505-2 can now be