FLOW DIRECTIONS IN DIKES FROM ANISOTROPY OF MAGNETIC-SUSCEPTIBILITY DATA - THE BOOTSTRAP WAY

One of the first applications of anisotropy of magnetic susceptibility (AMS) was an attempt to determine flow directions from mafic dikes [K han, 1962]. Since the seminal work of Knight and Walker [1988] definin g the expected behavior of AMS in response to magma flow, there has be en increasing interest in using AMS for this purpose. Here we present a quantitative method for interpretation of AMS data from dikes, using a parametric bootstrap. First, dikes must be sampled with at least fi ve land preferrably more) samples from within 10 cm of the dike margin . The distributions of the eigenvalues and eigenvectors of the AMS ten sor are delineated by calculating eigenparameters of many bootstrapped paradata sets. We generate paradata sets by first selecting a sample at random, then calculating a replacement set of data by drawing tenso r elements from normal distributions with the mean and standard deviat ion of the entire site. The bounds containing 95% of the eigenparamete rs of the bootstrapped data serve as confidence limits for the paramet er of interest. Classification of dikes proceeds as follows: Sites who se maximum and intermediate eigenvalues could not be distinguished are deemed uninterpretable. In addition, sites with principal eigenvector s with angles > 45 degrees away from the dike margin (inverse) or with markedly different directions on either side of the dike (scissored) are excluded. The remaining dikes are classified as having unique flow direction information if the principal eigenvectors from at least one side are distinct from the dike plane based on the distribution of th e bootstrapped principal eigenvectors. If neither side has principal e igenvectors distinct from the dike plane, the dikes are classified as having lineation information only. A study comprising 251 dikes from t he Troodos ophiolite has 151 sites with directional data, 38 sites wit h lineations only, 7 inverse sites, 5 scissored sites, and 55 sites no t fitting into any other category. The flow directions interpreted fro m the data were generally southerly, toward a fossil transform zone.