Geology Reference
In-Depth Information
Fig. 6.11
Triassic - early
Cretaceous APW path for
N. America, based on a
compilation of 15
high-quality paleopoles
(May and Butler
1986
).
Circles
are 95 %
confidence cones
Kent and Van der Voo
1990
; May and Butler
1986
; van Fossen and Kent
1992
; Kent and Witte
1993
). According to May and Butler (
1986
), the
most important factor controlling the accuracy of
APW paths is the reliability of the selected data
base, thereby, these authors propose to select
only high-quality paleopoles and to evaluate
their time sequence. Clearly, this technique relies
on an accurate selection of the data, which
will be paleopoles determined using the most
severe demagnetization analyses and from a large
number of samples and sites. A good example of
this class of APW paths is shown in Fig.
6.11
.
Although the approach of constructing APW
paths from few reliable paleopoles does not suffer
the problems of the sliding-window method, it
relies too strongly on the more or less subjective
process of paleopole selection, which does not
guarantee the correctness of each selected datum.
For example, only two of the three early Triassic
(246 Ma) paleopoles in Fig.
6.11
may have been
drawn from the same Fisher distribution. There-
fore, the third paleopole could be either correct
(in fact, it is located close to the 240 Ma result)
or the other two paleopoles are correct and this
result should be discarded. Actually, in the selec-
tion of reliable paleopoles for the construction of
APW paths, it is necessary taking into account
that even the “best” paleomagnetic direction can
produce a
wrong
paleopole if it is attributed an
incorrect age.
The third approach to the construction of
APW paths tries to overcome the limitations of
the methods described above through statistical
regression techniques. In this instance, APW
paths are built by fitting smoothed regression
curves on the sphere through swaths of
paleomagnetic data. Parker and Denham (
1979
)
were the first to propose an interpolation method
based on cubic splines. Similarly, Thompson and
Clark (
1981
) used weighted, least-squares cubic
splines to fit smoothed curves to the colatitudes
and longitudes of paleopoles from North America
and Europe. Musgrave (
1989
) applied a modified
version of the weighted least-squares regression
method to a study of Cretaceous and Cenozoic
Australian paleomagnetic data, while Jupp and
Kent (
1987
) developed a sophisticated fitting
algorithm based on spherical smoothing splines.
Examples of APW paths based on the method of
