Jökull - 01.01.2009, Page 54

J. T. Andrews and J. Harðardóttir
val of ca. 6 cm at the beginning and end of each u-
channel. Hence, for a 3 m JM96 core we have deleted
36/300 readings per core, whereas we only deleted
24/300 readings on the 150 cm sections of B997.
The magnetic measurements performed on the
u-channels involved natural remanent magnetization
(NRM) measurements, including several demagne-
tization steps (Appendix I). Volume susceptibility
(k"10"5 SI units) was measured (not always success-
fully) using a Bartington loop separate from the mag-
netometer (Thomas et al., 2003). Anhysteretic rema-
nence (ARM) was imparted with a 100 mT AF field
and 0.5 mT DC field. The ARM was measured af-
ter stepwise AF demagnetization (Appendix 1). Fi-
nally, isothermal remanent magnetization (IRM) was
induced with a 1.0 T strong magnetic field. The u-
channels were then subject to several demagnetization
steps. However, the strength of the signal frequently
overloaded the system; for this reason we have ex-
cluded evaluation of the IRM data.
We focus on parameters that describe various as-
pects of the sediment- and paleo-magnetic character
of these shelf sediments. For example, if the miner-
alogy is dominantly magnetite then the susceptibil-
ity of ARM (kARM) when normalized with volume
magnetic susceptibility (k) is a measure of the domain
grain size of magnetic minerals, and is often referred
to as “magnetic grain size” (higher kARM/k values in-
dicate finer magnetic grain size (King and Channell,
1991)). ARM(J0)/ARM(J20) has also been shown to
be closely correlated with grain-size of the bulk sedi-
ment in a core from the North Iceland shelf (Andrews
et al., 2003). The NRM data gives information on the
declination, inclination, and intensity of the magnetic
field at the time of deposition (or shortly thereafter).
The characteristic remanence intensity of the sedi-
ment is, however, also affected by the concentration
and type of available ferrimagnetic minerals; thus the
intensity measurements are normalized with parame-
ters such as MS, ARM, or IRM (Lund and Schwartz,
1999).
Quantitative X-ray diffraction (XRD) of the <2
mm sediment fraction has been carried out on 3 of the
East Greenland cores and 5 from Iceland (Andrews
and Eberl, 2007). Weight % estimates of quartz, mag-
netite, hematite, and volcanic glass have been deter-
mined (Appendix 2). Magnetite, plus its close relative
maghemite, varied on average between 2.5 and 7 wt%
whereas hematite is much lower. The higher values
of quartz in East Greenland sediments compared to
Iceland (Appendix 2) represent proximal glacial ero-
sion of Precambrian Shield rocks of Greenland and/or
transport of quartz in sea ice from the Arctic Basin,
compared to the more variable drift ice history of
Iceland.
DATA PROCESSING AND
ILLUSTRATIONS
Our primary objective is a regional comparison of the
data and we do not present the down core data (the
raw data will be deposited in the NOAA Paleoclimate
databases – www.ngdc.noaa.gov/paleo/data.html).
Downcore data along a fjord/shelf transect has been
published for the Vestfirðir area (sites B997-339 to
-336, Figure 1 and Table 1) (Andrews et al., 2008).
We have chosen the Exploratory Data Analysis ap-
proach for measures of central tendency and disper-
sion, namely the median and pseudo-sigma (Hamil-
ton, 1990); this follows previous arguments in deal-
ing with the sediment properties from many of the
same cores (Andrews et al., 2002a). The median
is a more conservative measure of central tendency
than the mean, as skewness or outliers do not influ-
ence it. Pseudo-sigma is defined as the interquartile
range/1.34 (Hamilton, 1990). Finally, in order to com-
pare variability amongst the parameters we calculate
a coefficient of variability (CV%) as: (median/pseudo
sigma)*100. CV% is a statistic designed to normalize
the usually positive correlation between means and
standard deviations because the latter is frequently
correlated with the average (Hamilton, 1990). The
tabulated data (medians, CV% etc.) can be obtained
from the first author.
We used an Excel®program (Mazaud, 2005) to
calculate the maximum angular deviation (MAD) re-
sulting from the AF demagnetization steps, the me-
dian destructive field (MDF mT) for demagnetization,
and to derive an estimate of the characteristic declina-
tion and inclination (Kirschvink, 1980) (Appendix 2).
54 JÖKULL No. 59