The young and rapidly developing discipline of cryoseismology uses icequakes to monitor the effects of global warming on ice sheets. The geophysical observatory at Neumayer Station in Antarctica has recorded a time series of more than 20 years of seismological data that includes not only earthquakes but also icequakes caused by surrounding ice dynamics. Routine processing by the observatory has not previously analyzed cryogenic events because they are far too numerous for manual analysis. To tap this valuable seismological data reservoir and reveal changes in the nearshore ice shelf environment over the past two decades, we need efficient algorithms that can automatically detect and classify icequakes. Modern machine learning methods are in principle capable of distinguishing between earthquakes and icequakes (Hammer et al. 2015), but they were developed for computer vision tasks with millions of datasets to train. In almost all other applications - including this project - well-annotated training is rare. Therefore, we need to address a "Small Data" problem in a Big Data environment.
To circumvent this, this PhD project applies techniques from transfer learning and so-called active learning strategies, where the algorithm determines during training those datasets where detailed annotation is required. For the task of distinguishing between different seismic events, the project will build on recent developments for Deep Learning applied to inverse problems (Arridge et al. 2019). Using this approach, the PhD student will develop a fast and efficient algorithm to reanalyze the lifetime dataset from the Neumayer Geophysical Observatory and operate on the real-time data stream. The resulting consistent catalog of cryogenic events will then allow monitoring of changes in the stress state of the coastal ice shelf over time.
Objectives in this project are: (1) development of an algorithm for automatic detection and discrimination of icequakes, (2) application to the Neumayer seismological archive, (3) statistical analysis of temporal variations in icequake occurrence.
More information at: https://www.mardata.de/