Reliability of geomagnetic paleointensity data: the effects of the NRM fraction and concave-up behavior on paleointensity determinations by the Thellier method

Abstract

To test the reliability of the Thellier method for paleointensity determinations, we studied six historic lavas from Hawaii and two Gauss-age lava flows from Raiatea Island (French Polynesia). Our aim is to investigate the effects of the NRM fraction and concave-up behavior of NRM-thermal remanent magnetization (TRM) diagrams on paleointensity determinations. For the Hawaiian samples, the paleointensity results were investigated at both sample and site levels. For consistency and confidence in the paleointensity results, it is important to measure multiple samples from each cooling unit. The results from the Raiatea Island samples confirm that reliable paleointensities can be obtained from NRM-TRM diagrams with concave-up curvature, provided the data are accompanied by successful partial TRM (pTRM) checks and no significant chemical remanent magnetization (CRM) production. We conclude that reliable determinations of the paleofield strength require analyses of linear segments representing at least 40-50% of the total NRM. This new criterion has to be considered for future studies and for evaluating published paleointensities for calculating average geomagnetic field models. Using this condition together with other commonly employed selection criteria, the observed mean site paleointensities are typically within 10% of the Definitive Geomagnetic Reference Field (DGRF). Our new results for the Hawaii 1960 lava flow are in excellent agreement with the expected value, in contrast to significant discrepancies observed in some earlier studies.

Overestimates of paleointensity determinations can arise from cooling-rate dependence of TRM acquisition, viscous remanent magnetization (VRM) at elevated temperatures, and TRM properties of multidomain (MD) particles. These outcomes are exaggerated at lower temperature ranges. Therefore, we suggest that, provided the pTRM checks are successful and there is no significant CRM production, it is better to increase the NRM fraction used in paleointensity analyses rather than to maximize correlation coefficients of line segments on the NRM-TRM diagrams.

We introduce the factor, Q = N < q >, to assess the quality of the weighted mean paleointensity, H-w, for each cooling unit.