Thesis title: The coherency of seismic surface waves before, during and after the 2011 Tohoku tsunami

Research interests: Interferometry, Seismology, Tomography, Machine Learning

The Makgadikgadi Basin (MB) is a region in eastern Botswana that hosts a large palaeo-lake and modern salt pans. Within the area, a series of potential fault scarps have been identified. The MB is one of three rift belts that lie at the southern end of the East African Rift System (EARS) and is believed to either be adjacent to or a potential extension of the EARS. The regional significance of the rift system in the MB and the series of normal faults remains unclear, but it is believed that this rift forms part of the southwestern branch of the EARS. The MB’s geomorphology, preserved quite well due to the arid climate of the region, was studied extensively, and subtle horst and graben topography was identified, with fault scarps under 40m identified. The fault structures were also identified to be more complex that initially thought, continuing south across the lake basin and through the Zambezi valley to the north.

For the initial dataset, a team from the University of Cape Town collected a seismic dataset from the MB area using an array of geophones. The data obtained from this will be analysed using ObsPy and MSNoise, which are both Python packages used to analyse seismic data, with the latter being used for ambient seismic noise specifically.

Machine-learning will also be used in order to detect microseismicity in these faults. This allows the statistical properties of large datasets to be inferred, which is often a significant improvement over traditional methods when the amount of data is very large. SeisBench, an open-source machine-learning deploying framework in seismology, has numerous models which will be used to perform the tasks for the large datasets.

Supervisor/s: Dr Diego Quiros & A/Prof Alastair Sloan

Degrees: 

1. BSc Geological Sciences and Environmental & Geographical Sciences, UCT