A dissertation submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Master of Science. Johannesburg, 2015 / Three new methods for determining the structural index and source distance for magnetic field data are presented. These methods require only the calculation of the first and second order analytic signal amplitudes of the total field and are applicable to both profile and gridded data. The three methods are first tested on synthetic data and then on two real datasets to test for applicability and repeatability. It was found that each method had different strengths and weaknesses and thus one method cannot be favoured over the others. Cooper (2014) describes how to calculate the distance to source over both profile and gridded data given a user defined structural index. Often however, particularly in the case of real data, the structural index is not known or varies over the surveyed area. These three new methods however do not require any user input since the structural index is calculated thus making them more applicable to regions of unknown geology. It was found that the first of the three new methods, the multi-distance inversion method, was best used as an edge-detection filter, since the use of higher order derivatives resulted in increased noise levels in the distance to source calculation. The third of these new methods, the unconstrained inversion method, discussed in Chapter 7, not only solves for the structural index but also determines the depth of the source. In that particular case, the structural index is used as a rejection filter, whereby, depth solutions associated with structural index values outside of the expected range are deemed to be invalid. Unlike the third new method, the first two methods require the distance to source to be calculated via the approach described by Cooper (2014) (which requires the user to define the structural index), the results of which are later rescaled by the calculated structural index to yield what is termed a rescaled distance to source. All three of the new methods are fully automatic and require no user control.
The techniques were first tested on both profile and gridded theoretical data over sources with known structural index values. All of the methods were able to estimate the structural index of each of the particular sources and give depth estimates that varied from the true depth by less than 20 percent (with deeper sources being more inaccurate). Noise was also added to the theoretical data in an attempt to assess how the methods can be expected to perform with real data. It was found that when applied to noisy data, these methods performed equally well to slightly worse, than when the method developed by Cooper (2014) was used.
As a real world case study these three new methods were tested on aeromagnetic data collected over the Kuruman Military Area, Northern Cape, South Africa. Regional deformations as well as later intrusive dykes and cross cutting faults were imaged by the chosen depth determination procedures. The dolerite dykes in the area were found to occur between 20 to 60 m deep. While the sand cover was estimated to be between 30 to 40 m thick. Overall, the techniques yield distance to source estimates that differ by less than 15 m, over sources, to the results obtained by using the source distance method (Cooper, 2014). To test for repeatability a second aeromagnetic dataset, collected over a dyke swarm within the Bushveld Complex, South Africa was considered. Again comparable (less than 15 m over sources) depth estimates were made between the unconstrained and constrained inversions. Since the distance to source estimates produced by these new unconstrained inversion methods are comparable to those produced by constrained inversion (Cooper, 2014) the project can be deemed successful.
Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:wits/oai:wiredspace.wits.ac.za:10539/19305 |
Date | January 2016 |
Creators | Whitehead, Robert |
Source Sets | South African National ETD Portal |
Language | English |
Detected Language | English |
Type | Thesis |
Format | application/pdf |
Page generated in 0.0023 seconds