An Update on the 0--5 Ma Geomagnetic Field recorded by Lava Flows

Lava flows have been used since early last century to study secular variation of the ancient magnetic field, and over the past decade there has been a concerted effort to augment the global data set recording the field for 0-5~Ma. We present a synthesis of recently-collected paleomagnetic directional data and others drawn from earlier publications. The combined data set has significantly better geographical coverage, particularly in the southern hemisphere and at high latitudes, and is used to study variations in inclination anomaly and VGP dispersion with latitude. Typically data quality is assessed via the 95% confidence cone about the mean direction, or an estimate, k, of the Fisherian precision parameter. The new data set, although not yet comprehensive in spatial coverage, comprises over 3100 flows with k greater than 100, almost an order of magnitude more than data of similar quality from previous global compilations. These lava flows preferentially sample the Brunhes and Matuyama epochs. Our analyses exclude transitional data, defined as sites with VGP latitudes less than 45 degrees. The common signature in previous models for the non-axial-dipole contributions to the average field, is that of an axial quadrupole term with a magnitude 2% -- 5% of the axial dipole. Such models predict negative inclination anomalies at all latitudes, with the largest signal at low latitudes. Our data set exhibits this signature at low latitudes, but more complicated signals elsewhere, notably small positive inclination anomalies from 40--50 degrees latitude. Matuyama epoch data, in particular, are better fit by both axial quadrupole and octupole contributions to the non-dipole time-averaged field. The strong latitudinal increase in VGP dispersion predicted by most secular variation models is not seen in our data set. However the small sample size for near-equatorial data sets means that the uncertainties in these VGP dispersion estimates are large, and suggests that alternative measures of secular variation are needed to assess and compare plausible secular variation models. Regional differences in both inclination anomaly and VGP dispersion are observed at some latitudes.