Machine learning for monitoring ground deformation with InSAR data Nantheera Anantrasirichai
University of Bristol, UK
Dr Nantheera Anantrasirichai is a Senior Lecturer (Assoc. Prof.) in the Department of Computer Science at University of Bristol. Previously she was a Senior Research Fellow associated with the Bristol and Bath Creative Industry Cluster working on AI-enable production, and a Fellow of the Centre for Observation and Modelling of Earthquakes, Volcanoes and Tectonics (COMET). Her research interests include AI-based production, image analysis and enhancement, medical imaging, and remote sensing. She has contributed to many projects across multiple disciplines, including engineering, psychology, biology and the creative arts. She has developed state-of-the-art methods for denoising long-range imagery distorted by atmospheric turbulence and for detecting ground deformation globally in noisy satellite imagery.
Introduction of the Lecture
Interferometric synthetic aperture radar (InSAR) can be used to measure ground displacement over large geographic areas. However, while detecting deformation using InSAR images is conceptually straightforward, it is difficult to automate, particularly in dense vegetation areas and where the signals are sparse. In this talk, a simple but efficient machine learning framework based on modern deep learning to detect ground deformation will be present. The model can be applied to individual interferograms for rapid deformations and can be tested on time-series for slow and sustained ones. The image processing techniques to deal with difficult, noisy, and sparse signals will also be addressed. This talk will analyse based on two used cases: i) the system for globally monitoring volcanic unrest and ii) the system for detecting ground deformation in the built environment.