Detecting Non-Geocentric Axial Dipole Structure in the Time-Averaged Field

The approximation of Earth's magnetic field by a geocentric axial dipole (GAD) is central to applications of paleomagnetism in global tectonics. Quantifying non-GAD contributions to the geomagnetic field is not only important because of the consequences for tectonic studies, but is essential to understanding the role of inner core growth and core-mantle boundary influences on field generation. Significant departures from GAD, in particular those that can be represented by a zonal octupole field, have been suggested over long time periods (10⁸ -- 10⁹ yr) prior to 250 Myr ago. However, biased estimates of the time-averaged field direction are obtained from unit vectors (i.e., in the absence of paleointensity data); deviations of inclination from that predicted by a GAD field are well approximated by a zonal octupole contribution. Simulations from statistical models for paleosecular variation that match 0--5 Myr paleodirection and paleointensity data indicate that the biases in predicted inclinations are on the order of 2^o. Increased secular variation would result in larger biases, so this effect may be important during periods of low paleointensity and/or earlier in Earth's history when the inner core was smaller. Over shorter time intervals (10⁶ yr), the time-averaged field shows smaller, but observable departures from GAD. Although a zonal quadrupole contribution has been considered robust, proposed longitudinal (non-zonal) structure in the time-averaged field has been challenged on the grounds of inadequate spatial and temporal data coverage, data quality, and contamination by local tectonic effects. It is now possible to compile regional data sets comprising paleodirection and occasionally paleointensity data from tens to hundreds of sites, and spanning the period 0--5 Ma. These include Hawaii, Reunion, Japan, French Polynesia, New Zealand, and North America. In addition, a recent sampling program has focused on obtaining paleomagnetic data from 0--5 Ma lava flows from previously under-sampled high latitude and the southern hemisphere regions. New time-averaged field models constructed from 0--5 Ma normal polarity data have improved data coverage in the southern hemisphere and suggest the presence of southern hemisphere flux lobes, undetectable with previous data sets. We use 0--5 Ma data sets, new time-averaged field models, and statistical models to examine the conditions required regionally and globally to detect non-GAD field structure.