UC San Diego

Paleosecular variation (PSV) models require high-quality, globally distributed paleomagnetic data sets in order to accurately predict observed geomagnetic field behavior. Previous paleomagnetic studies have lacked modern high- quality directional data, especially at equatorial and high northern latitudes. As a result of these incomplete data sets, PSV models, notably Model G [McFadden et al., 2008] and TK03 [Tauxe and Kent, 2004], may not accurately predict PSV behavior. We present 65 high-quality directional paleomagnetic sites from low (Costa Rica) and high northern (Jan Mayen and Spitsbergen) latitudes. We also compile a new paleomagnetic data set, PSV10, and investigate VGP dispersion with latitude. PSV10 spans the last 10 Ma and includes all published studies that meet our selection criteria. We do not incorporate a latitudinal VGP cutoff and include all sites regardless of VGP latitude, so long as the published study does not intentionally sample transitional directions. Transitional directions are a natural aspect of secular variation, and removing them biases PSV behavior, especially at high latitudes where scatter is greater. Model G and TK03 underestimate dispersion of virtual geomagnetic poles (VGPs), of PSV10, at high and low latitudes. Attempts to modify the model parameters of Model G and TK03 to fit PSV10 were unsuccessful. PSV10 shows significant latitudinal variation between equatorial and high latitude sites, while dispersion across low to mid-latitudes is fairly consistent New PSV models must be developed to explain the observed behavior of the magnetic field