Lisa Tauxe, & Kenneth P. Kodama

Paleosecular variation models for ancient times: Clues from
Keweenawan lava flows

Published in: Physics of the Earth and Planetary Interiors 177 (2009) 31–45.

Abstract: Statistical paleosecular variation models predict distributions of paleomagnetic vectors as a function
of geographic position. Such models have been used in a variety of applications that test whether a
given data set fairly represents the variability and average properties of the geomagnetic field. The simple
relationship between inclination of the geomagnetic field and latitude predicted by geocentric axial dipole
(GAD) models has been a cornerstone for plate reconstructions for decades, yet many data sets exhibit
a tendency to be shallower than expected for a dominantly axial geocentric magnetic field. Too shallow
inclinations have variously been interpreted as platemotion, permanent non-dipole field components or
bias in inclination fromsedimentary processes. Statistical PSVmodels could in principle be used to resolve
the cause of inclination anomalies because there is a simple relationship between the elongation of the
distribution of directions in the vertical plane and the average inclination. Shallowing of inclinations
from sedimentary processes results in a progressive transformation of the elongation direction in the
vertical plane containing the average direction into a pronounced elongation in the plane perpendicular
to that. However, the applicability of statistical models based on the last 5 million years for more ancient
times is an open question. Here we present new data from the Keeweenawan North Shore Volcanics
(∼1.1 Ga). These data are consistent with statistical PSV model predictions and are less well fit by models
that include a 20% axial octupole component. We also find evidence for a pervasive overprinting by
hematite in a shallower direction and find support for the contention that the asymmetric reversal(s)
observed in Keweenawan aged rocks along the North shore of Lake Superior can be explained as an age
progression, with the reverse directions being older than the normal directions. Finally, we re-consider
implications froman analysis of inclinations fromthe Global Paleomagnetic Database for the Paleozoic and
Pre-Cambrian.We find that the data are inconsistent with a random sampling of any simple geomagnetic
field model and conclude that the data set under-samples the field in a spatial sense.


Geology of the North Shore of Lake Superior, Minnesota:

Geological map of four sequences of the North Shore Volcanics (based on Miller et al., 2001). U–Pb dates on rhyolite flows (numbers next to yellow squares) are
from Davis and Green (1997) and Paces and Miller (1993)

sampling locations

Sampling sites for this study.

White circles are reversely magnetized while red circles are normal. Circles with an X were excluded from further consideration. Site 17 is in close association with roof rocks of the Beaver Bay Complex where the structure is complicated (see Miller et al., 2001). Other excluded sites had high within site scatter.

Data from all acceptable site mean directions using specimens from the site that met minimum criteria for various magnetic mineralogies. Solid (open) symbols are in the lower (upper) hemisphere, plotted in equal area projection. Data are in tilt corrected coordinates. (a) Sitemeans based solely on “magnetite” remanences.
Light blue dot is the antipode of the sole reverse site (NS026) based on magnetite remanence directions. (b) Site means based solely on “hematite” site means.
(c) Site means for sites not included in (a) or (b) that were based on both hematite and magnetite specimen remanences.

Elongation/inclination curve of the TK03 model (Tauxe et al., 2005) shown
as solid red line. TK03.g30 (20% octupole contribution) is the dashed green line. Data from large igneous provinces compiled by Tauxe et al. (2008) shown as light blue. Y: Yemen (28–30 Ma); D: Deccan (65 Ma); F: Faroe (55–58 Ma); K: Kerguelen
(24–30 Ma). Also shown are the ∼30Madata from the Ethiopian traps (E) of Rochette et al. (1998). The results fromthe Keweenawan lavas of this study are shown in black and are marked “NS”.


1) As in earlier studies of the North Shore Volcanics, (e.g., Books, 1968; Palmer, 1970), we find two main directional groups in the North Shore Volcanics: a normal direction that is shallow and to the west and a reverse direction that is steep and to the southeast. These directions are the same as those found initially by Dubois (1962) in the Pigeon River Dikes and the Logan sills respectively.

2) Scrutinizing the demagnetization data in more detail reveals that the magnetization of the North Shore volcanics can be attributed to both magnetite and hematite directions. We have classified site mean directions into three groups: those that were based on magnetite directions solely, those that were based on hematite directions solely and those that had specimens with both types of remenence (here called "mixed'"). While the "magnetite" and "mixed" site means are indistinguishable, the magnetite and hematite site means do not share a common mean, with the hematite direction being significantly shallower. The magnetite directions are significantly steeper than the hematite directions, and are intermediate between the reversely magnetized directions and the hematite components. The three magnetite directions are fortuitously based on all three magnetizations with the hematite direction being the steepest and the magnetite direction being the shallowest. The antipode of the latter is compatible with the other magnetite directions. We interpret these results as support for the hypothesis that the asymmetry in normal and reverse directions in the Keweenawan data sets stems from overprinting in the shallow westerly direction.

3) While there are substantial differences among the statistical (Giant Gaussian Process) paleosecular variation models, all predict distributions of directions that are circular at the poles where the average directions are near vertical and elongate in the meridian near the equator where the directions are near horizontal. Here we test whether the directional data from the North Shore volcanics are consistent with the PSV model of Tauxe and Kent (2004; see also Tauxe et al. (2008)). Of particular interest is whether a field with a strong non-zero octupolar term can be excluded. Elongation (defined as the ratio of the intermediate and minimum eigenvalues of the orientation matrix of the directions) in the TK03.GAD when plotted against inclination defines a smooth trend, allowing a given data set to be tested against the distribution predicted by the field model. Excluding the sites based solely on hematite remanences, there are 44 site means that we consider to represent "primary"' directions from lava flows in the North Shore volcanic group. From these we calculate the average inclination and elongation; these are compatible with those predicted by the TK03.GAD statistical paleosecular variation model. Moreover, a field model with significant non-zero octupolar contributions provides a much less convincing fit.

4) A re-analysis of the treatment of inclinations from the Global Paleomagnetic Database shows that the data are very poorly fit by a random sampling of a GAD or simple non-GAD field. It is more likely that the existing database significantly undersamples the field in a spatial sense, as discussed by Kent and Smethurst (1998) and Meert et al. (2003).


Books, K., 1968.Magnetization of the lowermost Keweenawan lava flows in the Lake Superior area. USGS Prof. Pap. 600-D, D248–D254.

Davis, D., Paces, J., 1990. Time resolution of geologic events on the KeweenawPeninsula and implications for development of the Midcontinent Rift system. Earth Planet. Sci. Lett. 97, 54–64.

Davis, D.W., Green, J.C., 1997. Geochronology of the North American Midcontinent rift inwestern Lake Superior and implications for its geodynamic evolution. Can. J. Earth Sci. 34 (4), 476–488.

Dubois, P., 1962. Paleomagnetism and correlation of Keweenawan rocks. Geol. Surv. Canada Bull. 71, 75pp.

Kent, D., Smethurst, M., 1998. Shallow bias of paleomagnetic inclinations in the Paleozoic and Precambrian. Earth Planet. Sci. Lett. 160, 391–402.

Meert, J., Tamrat, E., Spearman, J., 2003. Non-dipole fields and inclination bias: insights from a random walk analysis. Earth Planet. Sci. Lett. 214, 395–408.

Miller, J., Green, J., Severson, M., Chandler, V., Peterson, D., 2001. Geologic map of the Duluth Complex and related rocks, northeastern Minnesota.

Paces, J., Miller, J., 1993. Precise U-Pb ages of Duluth Complex and related mafic intrusions, Northeastern Minnesota: Geochronological insights to physical petrogenetic, paleomagnetic and tectonomagmatic processes associated with the 1.1Ga midcontinent rift system. J. Geophy. Res. 98, 13997–14013.

Palmer, H., 1970. Paleomagnetism and correlation of some Middle Keweenawan rocks, Lake Superior. Can. J. Earth Sci. 7, 1410–1436.

Tauxe, L., Kent, D.V., 2004. A simplified statistical model for the geomagnetic field and the detection of shallowbias in paleomagnetic inclinations:was the ancient magnetic field dipolar? In: Channell, J.E.T.e.a. (Ed.), Timescales of the Paleomagnetic Field, vol. 145. American Geophysical Union,Washington, DC, pp. 101–116.

Tauxe, L., Kodama, K., Kent, D.V., 2008. Testing corrections for paleomagnetic inclination error in sedimentary rocks: a comparative approach. Phys. Earth Planet. Int. 169, 152–165.