Basic paleomagnetism: some old and
new lessons from Icelandic lava flows
Leo Kristjansson
University of Iceland
(Slightly revised version of an invited talk given at the 2008 Joint Assembly
of the American Geophysical Union, Fort Lauderdale, Florida, 27-30 May.
A paper discussing some aspects mentioned here is in press in Jökull v. 58)
Some advantages of Iceland (65°N, 103,000
sq.km.) for paleomagnetic fieldwork
• Basaltic volcanism was essentially continuous from 15
Ma ago to the present, generating millions of lava flows
• Tens of successive lava flows are accessible in very
many mountainside profiles, with geomagnetic reversals
occurring every 15-20 flows on average
• In many instances, several such profiles can be
combined into a composite section of hundreds of flows
• Areas with minimal secondary hydrothermal (< 100°C)
and tectonic (< 6-8° tilt) disturbances may be chosen
• Dikes and other intrusions can be easily avoided
• Continued...
NW-Iceland: eroded plateau 600 m high. Age 12-13 Ma, tectonic tilt 2 - 4°
The lavas are generally separated by scoria zones and thin
sedimentary beds; thicker sediments appear occasionally
2.9 km section studied by Kristjansson
et al. (2003), ages c. 9 to 5 Ma
--------------One more advantage of Iceland:
cooling water is usually not far away
Advantages, continued:
• Good road connections, communications, services, etc.
• No danger from landowners, armed insurgents, heat,
grizzlies, snakes, poisonous plants, insects, microbes,...
• Little problems with vegetation cover, chemical
weathering, lightning-strike remagnetization, etc.
• No complications from continental drift, area rotations
about the vertical, polar wandering, anisotropy,
inclination errors, variable lock-in depth, folding,....
• Viscous remanence is often present but easily removed;
rotational remanence is very seldom noted
• The primary remanence direction is usually very stable
and internally consistent
• In short : Iceland is a good place for studying some basic
aspects of paleomagnetism
In paleomagnetic work at the
University of Iceland and by foreign
expeditions, emphasis has been on
using remanence directions to help in
stratigraphic mapping projects.
Collecting four samples and AFdemagnetizing at 4-5 peak fields, is
more than adequate for that purpose:
average a95-values of 5° or so are
typical, even in the oldest formations.
Accuracy will in many cases
deteriorate with AF treatment beyond
30 mT.
Typical
Zijderveld
plot (BeskeDiehl & Li,
1993)
It has been verified that lavas exposed across several km yield consistent results.
In the lava near the top here, samples were collected at 8 sites spread over 1.5 km.
The between - site a.s.d. of the remanence directions was a little over 2°.
A few disadvantages of Icelandic lava series:
• Stratigraphy of volcanics from the last 2 Ma tends to be
jumbled due to various effects of glaciation and erosion
• The time interval between successive lavas (av. thickness
10 m, extent ~ few km) averages 5-10 ka; I’ve estimated
that a reversal takes on average around 4 ka
• Some locations are too steep for sampling
• The volcanism was quite sporadic in
time and space, with hiatuses occurring
• Stratigraphic mapping is still only fragmentary
• Short field season: just July & August in many areas
• The number of available radiometric dates is insufficient
• The lavas are not very suitable for paleointensity; for
instance, their susceptibility often changes irreversibly on
laboratory heating to even 150°C or less
Total sampling: of order 8000 lavas. Data from some not published in detail.
Circles: 1960’s (n = 2); triangles: 1972-78 (n = 3); squares: since 1979 (n = 4)
Each sign stands for 60-100 lava flows. Crosses: smaller isolated surveys.
Some instances in the past where studies on Icelandic lavas
contributed to progress in paleomagnetic methods and concepts
1950’s
-First application of polarity reversals in local stratigraphy
-First use of Fisher’s directional statistics
-Development of the idea that the mean field is a geocentric axial dipole one
-Successful application of AF demagnetization in removal of viscous remanence
-First observation of weak “intermediate” directions at polarity-zone boundaries
1960’s
-One of the first records of short geomagnetic reversal “events”
-Long composite profile in East Iceland demonstrated the occurrence of at least
60 reversals in the period from c. 13 to 2 Ma ago
1970’s and early 1980’s
-K-Ar dating of some polarity zone boundaries and of two events in the Gilbert
-Demonstration that N and R fields are similar in terms of intensities, a.s.d., etc.
-Further indications that many short reversal events are missing from time scales
We will now look at a collection of over 5000 Icelandic lava flows with 3 or 4
samples per flow, and a mean a95-value of around 7°
Example: Do 20 directions give an adequate representation of the paleo-field ?
Let us arbitrarily take about 20 successive lavas from this 114-lava profile in
W- Iceland, of c. 9 Ma age (Kristjansson and Johannesson, 1999)
45°N
Questions: Should we reject the pole in SE-Africa on the grounds that it
is a “transitional” one and thus possibly a type different from the others ?
Should we then also reject the pole in the Great Lakes ?
Should we try to generate a few hundred more poles by bootstrapping ?
This is the total collection from the study by Kristjansson and Johannesson
(1999) in W-Iceland, spanning approximately 7 to 11 Ma old lavas.
Among new questions coming up: Is there a Pacific “pole hole” ?
Here we have a new compilation of most of the available poles from
Iceland, of about 1 to 15 Ma age. The Pacific hole has disappeared.
Some simple conclusions from the above
• 20 spot readings seem much too few for obtaining
reasonably reliable estimates of mean poles and scatter
• Even with 300-odd lavas, the computed a95 is over 3°.
This probably increases to 4-5° when one allows for
serial correlation and non-Fisherian distribution
• The distribution of VGPs is essentially continuous in
latitude, which among other things indicates that the
application of a 40°- or 45°- latitude cutoff before
averaging may not be very appropriate.
Let us now look at the distribution of low- and mid-latitude VGPs in longitude....
The 940 VGPs seem to
be quite evenly spread
The main bump in the histogram lies to the west of
one “preferred” longitude interval suggested originally
in 1991.
A bump does not appear at that longitude interval in
these lava data for VGP latitudes > 50°.
Let us also look at the latitude distribution of VGPs...
We note that there is a fairly flat “tail” from 20-30° towards the Equator,
which cannot be accounted for by Fisherian statistics.
The suggestion by C.G.A. Harrison in 1980 that this tail is part of a population
of “random poles” forming about 10% of the whole data set, is still realistic.
If you measure samples from a large enough number of lava flows, you
have a chance of obtaining for instance the relative mean intensity of
the field as function of VGP latitude, as other factors cancel out.
This has been attempted with data from Icelandic lavas, first by P.
Dagley and R.L. Wilson in 1971-72 (1300 flows x 2 cores).
Assuming a linear
relationship between
geomagnetic dipole
moment and VGP
latitude, a factor of
around 4-5 between
the magnitudes of
mean axial and
mean equatorial
VDMs may be
deduced.
VGP co-latitude
I did similar compilations in 1982 and again in 1995, using
the then available data sets of lavas with 3 or 4 samples
from each, after 10 mT demagnetization to remove VRM.
I confirmed the estimate that mean VDMs in axial and
equatorial positions of the VGP differed by a factor of about
4-5, assuming a linear decrease of VDM with VGP latitude.
However, I noted that that the VDM vs. latitude relationship
showed signs of levelling off at both ends.
Let us now look at my current compilation of over 5000
flows with 3-4 samples from each.
The central set of data is a geometric average of lava
intensities, each having been transformed to its VGP.
Bars: typical s.e. values
The flat right-hand portion of my suggested relationship between VDM and latitude
lies between 25° and the Equator, cf. the flat portion of the VGP frequency distribution.
The ratio between near-axial and near-equatorial virtual dipole moments is then c. 2.5;
remember however, that some high-latitude VGPs belong to the random-pole set.
Returning to the plot by Dagley and Wilson (1972), let us estimate
roughly their mean VDMs for near-axial and near-equatorial VGP’s
A ratio of 2.5
between these
seems to be
not far off
Hopefully, results of this sort will aid people in modelling the paleomagnetic field and its sources, along with “absolute” paleointensity data
Let us now look at another property, viz. the angular (or
circular) standard deviation of virtual-pole positions from
Icelandic lavas, and its possible change since 15 Ma ago.
Paleomagnetists accept the possibility that e.g. polarityreversal rates and the mean field intensity may vary on time
scales of tens of Ma, but little evidence has been gathered so
far on possible long-term variations in the a.s.d.
I presented evidence for these with H. Johannesson in 1989.
We used data from collections with n = 2 (U. of Liverpool),
n = 3 (U. of Rhode Island) and n = 4 (mostly U. of Iceland),....
Data from surveys in 10 different areas of Iceland, approx. 4000 lavas (1989)
Broken lines: n = 2 samples/flow; thin lines: n = 3; thicker lines: n = 4
Let us now plot angular standard deviations of VGPs, after adding many
subsequent collections and grouping the lavas by age. Each point is
based on 510 to 950 lavas with n = 3 or 4, total 5030. In the top graph,
rejection criteria have been tightened from a95 < 21° in 1989 to a95 < 14°.
The error bar indicates 95% confidence limits for 500 lavas (Cox 1971)
The elevated a.s.d.-values in the older rocks may be partly due to a high
frequency, long duration, and/or large amplitude of excursions, cf. example
34-lava profile of c.13
Ma age, NW-Iceland
In some cases the geomagnetic pole may even have been
moving around erratically, e.g. in c. 12 Ma old lavas in NWIceland described by Kristjansson & Johannesson (1996)
Some materials, or aspects related to paleomagnetism,
which have not been studied much in Iceland so far:
•
•
•
•
•
Interbasaltic and other sediments, tuffs
Subaqueous/subglacial material such as pillow basalts
Historical and other Holocene lava flows
Intrusions, acidic rocks
Thick series of thin pahoehoe “flow units”
• AMS
• Rock magnetism
• Regional heating and alteration effects
Few foreign groups have carried out paleomagnetic fieldwork in
Iceland in recent years, and the level of our own efforts is decreasing.
The most valuable potential of Iceland for paleomagnetism may
lie in accurate radiometric dating of reversals and excursions.
Finally, it may be mentioned that paleomagnetic research in Iceland has
aided in the interpretation of the local aeromagnetic anomalies
Thank you