Development of joint moving-interface inversion methodology for geophysical DC resistivity and time-domain EM data

点击翻译按钮获取中文摘要

Uranium mining is important to Canada's economy, and finding more uranium ore deposits to mine is criticalto the future of this industry. Most uranium ore deposits in Canada are to be found in the Athabasca Basin in northern Saskatchewan. The uranium ore-bodies are located hundreds of metres beneath the Earth's surface at the junction between fault zones in the deep basement rocks and the sedimentary rocks which lie on top of the basement rocks. These fault zones in the basement rocks often contain graphite, which makes them relatively good at conducting electricity compared to the surrounding rocks. Electrical and electromagnetic surveys are sensitive to the presence of such electrically conductive features in the subsurface, and so play an important role in the exploration for new uranium ore-bodies in the Athabasca Basin. The data from such surveys contain information about the location, depth, size and electrical conductivity of the graphitic fault zones. The better the location, etc., of the fault zones can be determined from these surveys the less drilling is required, which can significantly reduce the expense of the exploration process. To derive this information from the survey data requires the use of computer modelling and interpretation techniques. However, current techniques are not suited to the sharp, distinct nature of the graphitic fault zones, producing instead fuzzy, indistinct representations of the faults. This project will extend a new computer interpretation approach, previously developed for gravity and magnetic data, to include electrical and electromagnetic data. This new approach specifies the subsurface features by discretizing the interfaces between neighbouring rock types, and then moving and adapting these interfaces so that geophysical survey data computed for this model of the subsurface match the data measured in a survey. This is in contrast to the current approach, which is to divide the subsurface into many little cells and to find the values of electrical conductivity in all the cells such that the match between computed and measured data is good and that the variation of the physical property from one cell to the next is smooth.