Abigail Robinson

My PhD research explores the use of machine learning and statistical methods to enhance and automate seismic and acoustic monitoring at volcanoes. 

PhD project title:

Advancing Real-Time Volcano-Seismic Monitoring through Machine Learning

Project description:

Abigail Robinson PhD profile

Seismic and acoustic monitoring is often essential to understanding the internal dynamics of a volcanic system and assessing the likelihood of hazardous activity. In recent decades, technological advancements have made it increasingly feasible to collect larger seismic and acoustic datasets at volcanoes. However, much of the data processing is still done manually by observatory analysts, creating new challenges as data volumes grow. At the same time, larger data volumes and the existence of decades of hand labelled data present great opportunities to harness new data-driven methods, such as deep learning, for more efficient monitoring.
In recent years, deep learning methods have generated highly accurate results in seismic processing tasks such as phase picking, earthquake detection, and seismic source characterization. However, the uptake of these methods into volcano monitoring workflows has been limited, in part, due to a lack of research developing models specifically for volcanic seismic data that are suited to real-time monitoring.
In my PhD, I am working to explore and overcome limitations in existing automated seismic monitoring methods, with a particular focus on deep learning, with the overall aim of developing new tools and models which can realistically enhance human monitoring capabilities at volcano observatories. I am working with seismic datasets from Nabro volcano, Eritrea and Tungurahua volcano, Ecuador, and collaborating with the Instituto Geofísico, Ecuador, to ensure that methods developed consider real-world volcano monitoring demands and limitations. 

Resources: