Geomagnetic secular variation and the statistics of palaeomagnetic directions

In this study, we examine the role of palaeosecular variation (PSV) in the use of statistics for palaeomagnetic studies, and we provide new reliability criteria for palaeomagnetic poles or directions. We first conclude that Fisher statistics should not be applied to average palaeomagnetic directions but to virtual geomagnetic pole (VGP) distributions instead. Secondly, we strongly advocate that typical properties of geomagnetic field behaviour are taken into account in the assessment of palaeomagnetic data sets. The latitude-dependent properties (E, S, k) provide useful guidelines for the reliability of a palaeomagnetic data set. A reliable assessment of these properties depends on the (sufficient) number of palaeomagnetic samples being taken. Therefore, as an additional instrument of assessing data sets, we provide a N-dependent A95 envelope, bounded by an upper limit A95max, and a lower limit A95min that helps to ascertain whether or not a distribution has sufficiently well-sampled PSV and therefore geomagnetic field behaviour. Applying these criteria is indispensable for studies of geomagnetic behaviour, or for studies aiming at using TK03.GAD for inclination error correction through the elongation/inclination (E/I) method. For palaeomagnetic studies aimed at geological reconstructions, they form helpful guidelines and increase the confidence in the rocks having faithfully recorded the field. An analysis of published Eastern Mediterranean data shows that the vast majority of studies do not conform to the Van der Voo criteria, in particular with respect to N and A95. We have provided criteria that are on the one hand more lenient (lower N may still provide relevant information), and on the other hand more strict (for high N the criterion of A95 < 16° should be adapted to a requirement of lower A95, e.g. A95 < 5° for N > 80).