HR: 0800h
AN: GP11D-0856    [Abstracts]
TI: Evaluation of Agreement Between Paleosecular Variation Models and Empirical Distributions at 20 Degrees Latitude
AU: * Lawrence, K P
EM: klawrence@ucsd.edu
AF: Institute of Geophysics and Planetary Physics, Scripps Institution of Ocenaography, 9500 Gilman Drive Mail Code: 0225, La Jolla, CA 92093-0225 United States
AU: Constable, C
EM: cconstable@ucsd.edu
AF: Institute of Geophysics and Planetary Physics, Scripps Institution of Ocenaography, 9500 Gilman Drive Mail Code: 0225, La Jolla, CA 92093-0225 United States
AU: Johnson, C L
EM: cljohnson@popmail.ucsd.edu
AF: Institute of Geophysics and Planetary Physics, Scripps Institution of Ocenaography, 9500 Gilman Drive Mail Code: 0225, La Jolla, CA 92093-0225 United States
AB: Several approaches have been employed to derive paleosecular variation (PSV) models of the magnetic field, one of which supposes that PSV can be represented as a Giant Gaussian Process (GGP). We examine the accuracy of models based on the GGP and the validity of their statistical and symmetry assumptions by direct comparison with paleomagnetic directional data drawn from latitudes around 20 degrees. We compiled paleomagnetic directional data from multiple published studies of volcanic provinces in four regions, Hawaii, Mexico, Reunion and French Polynesia. Sites were selected according to the following criteria: ages younger than 5 Ma, sampling latitudes approximately 20 degrees from the equator, and uncontaminated by tectonic rotation. If multiple directions were reported for the same lava flow we averaged those results. The resulting data set is large enough to justify more extended analyses than previously possible: namely statistical analyses of the distributions of magnetic directions and quantitative comparisons with proposed PSV models. For each region we calculate virtual magnetic poles (VGPs), paleomagnetic directions projected along the expected axial dipole field, and perform a principal component analysis, calculating shape and strength parameters for the data distributions. We determine the total angular dispersion of the calculated VGPs for each region. Data from all regions are merged to generate the largest possible data set for studying zonal (no longitudinal dependence) models. Sub-sets of the zonal data set are used to evaluate the influence of temporal and geographic variations in the PSV. We explore mechanisms for quantitatively evaluating agreement among empirical distributions of these data sets and distributions predicted from a variety of secular variation models. Using models proposed by Johnson and Constable (1999), CJ98, and Tauxe and Kent (2004), TK03, we examine the consequences of axial symmetry, or lack thereof, for the secular variation. We compare the observable geographic dependence of magnetic directional variation to a range of models with different variances in the axial-dipole and non-axial-quadrapole Gauss coefficients, and investigate whether the disparity between the modeled and observed distributions places limits on the values these parameters may take. Preliminary results suggest that the model parameters from CJ98 and TK03 cannot simultaneously satisfy data from all four regions. The tests outlined above will be used to determine whether this results from sampling bias in the data or an inadequacy of the model.
DE: 1532 Reference fields (regional, global)
DE: 1560 Time variations--secular and long term
SC: Geomagnetism and Paleomagnetism [GP]
MN: 2004 AGU Fall Meeting