PhD: Early warning of landslide events using computer vision and geophysical image analysis
My background is in Physics, having obtained my undergraduate MSci in Physics from the University of Nottingham. During my degree I spent a summer at the Diamond Light Source synchrotron, researching how coherent infrared light emitted by bunches of electrons can give information about the shape of the electron bunch. During my Masters project I used a magnetoencephalography (MEG) sensor array to construct images of a magnetised brain from measurements of the magnetic field. This is where I first became interested in inverse theory, ultimately leading to my pursuing a PhD in ERT.
My project combines computational imaging techniques with electrical resistivity tomography (ERT), a geophysical technique used to image the Earth’s subsurface. The idea is to develop automatic image analysis techniques which will allow 3D ERT arrays to remotely monitor slopes for landslides, providing a warning if a landslide seems likely. I am interested in developing the inversion (image formation) process so that it is better adapted for remote monitoring. The best current methods involve inefficiently inverting the entire time series of measurements at once. Ideally each measurement should be inverted as soon as it has been received, whilst still using the information about the past of the system contained in the previous measurements. This would result in a system able to measure and invert ERT data automatically, on short timescales, bringing the prospect of automatic landslide detection much closer to realisation.