Latitudinal signature of Earth's magnetic field variation over the last 5 million years

Abstract

In order to investigate the latitudinal effect of the geomagnetic field variation, a new data set consisting of virtual geomagnetic poles (VGPs) from all latitudes has been produced. Since the updated data set was limited to data with VGPs within 45° of the geographic poles, data from lava flows with low‐latitude VGPs were added. More rigorous criteria were used to winnow the data. The data were divided into groups from different observation latitudes. In each group it was shown that the distribution of VGP latitudes could be described by a predominance of poles (average 82%) following a Fisher distribution with the rest following a distribution that would produce a uniform number of poles as a function of latitude. A distribution composed of two Fisher distributions also fit the data very well. For the case using a Fisher distribution plus a uniform distribution, the Fisher distribution changed such that the angular standard deviation (ASD) of VGPs from a set of observations taken at the equator is about 10° and the ASD at 60° observation latitude is about 19°. These results are similar to some results seeking to determine the ASD of VGPs as a function of observation latitude using other methods, which have been recently published, but there are also discrepancies. The results allow us to model inclination distributions as a function of observation latitude for comparison with data in which only the inclination is known, such as data from drill holes. It is shown that in order not to have doubt about the polarity of the inclination data, a drill hole has to be located at an absolute latitude greater than 27° for there to be less than a 5% error. This has major importance for the location of the “Mission to the MOHO” of the Integrated Ocean Drilling Program. The results also confirm that the sources of the nondipole field located in the outer core have to be more than twice as strong at high latitudes than at low latitudes so as to produce the observed increase in VGP scatter with observation latitude. The model of the secular variation proposed here is in no way a theory about how the secular variation happens, but it does allow those who wish to develop such a theory to have a model distribution with which to check their predictions.