HR: 14:30h
AN: V53D-04    [Abstracts]
TI: A Statistical and Wavelet Analysis of Physical Property Data From the 2950 m Deep Bellevue Borehole, Bushveld Complex, South Africa
AU: * Webb, S J
EM: susan.webb@wits.ac.za
AF: University of the Witwatersrand, School of Geosciences Private Bag 3, WITS, 2050, South Africa
AU: Ashwal, L D
EM: lewis.ashwal@wits.ac.za
AF: University of the Witwatersrand, School of Geosciences Private Bag 3, WITS, 2050, South Africa
AU: Cooper, G R
EM: gordon.cooper@wits.ac.za
AF: University of the Witwatersrand, School of Geosciences Private Bag 3, WITS, 2050, South Africa
AB: Susceptibility (n=~110,000) and density (n=~~2500) measurements on core samples have been collected in a stratigraphic context from the Bellevue (BV-1) 2950 m deep borehole in the Northern Lobe of the Bushveld Complex. This drill core starts in the granitoid roof rocks, extends through the entire Upper Zone, and ends approximately in the middle of the Main Zone. These physical property measurements now provide an extensive database useful for geophysical modeling and stratigraphic studies. In an effort to quantify the periodicity of the layering we have applied various statistical and wavelet methods to analyze the susceptibility and density data. The density data have revealed a strong periodic layering with a scale of ~~80 m that extends through the Main and Upper Zones. In the Main Zone the layering is unusual in that the density values increase upwards by as much as 10%. This is due to systematic variation in the modal abundance of mafic silicates and appears to be related to separate pulses during emplacement. The magnetic susceptibility data in the Upper Zone also show a strong cyclicity of similar scale. The discrete wavelet transform, using the real Haar wavelet, has been applied to help discretise the susceptibility data and clarifies the geological boundaries without blurring them, which is a common problem with multipoint moving averages. As expected, the histogram of the entire data set is non-Gaussian, with a long tail for high values. We can roughly fit a power law to the log histogram plot indicating a probable fractal distribution of susceptibilities. However if we window the data in the range 750-1000 m the histogram is very different. This region shows a strong peak and no power law relationship. This dramatic change in statistical properties prompted us to investigate these properties more thoroughly. To complement the wavelet analysis we have calculated various statistical measures (mean, standard deviation, skew, and kurtosis) of a window of data 21 points wide for the length of the drill core. In order to graphically image how the 21 point windowed histogram varies with depth, we have developed a "Windowed Histogram Image" (WHI) which graphically demonstrates how much the histogram for a particular window size varies downward. This highlights the changing statistical pattern in the drill core highlighting pattern breaks that are likely to be indicative of emplacement pulses.
DE: 1036 Magma chamber processes (3618)
DE: 8145 Physics of magma and magma bodies
DE: 8400 VOLCANOLOGY
SC: Volcanology, Geochemistry, Petrology [V]
MN: 2007 Fall Meeting