GRSG 35th Conference 2024 Abstract

Title: Using AI for the automatic interpretation of InSAR data

Author: Alessandro Novellino

Organisation: British Geological Survey

To extract the most information from large-scale InSAR dataset it is imperative to understand the mechanisms leading to motion and how this manifest in an InSAR dataset such as the EGMS. The British Geological Survey (BGS) have been at the forefront of InSAR ground motion interpretation for the past 2 decades; in projects such as ESA’s Terrafirma, thematic FP7 projects e.g. PanGeo and SubCoast, and provide expert advice to the UK government surrounding potential fracking sites and CCS sites along with a sustained research programme. BGS also produce national hazard susceptibility mapping, known as GeoSure, which is routinely used by insurance companies to assess ground motion hazards.

The study of many epochs of InSAR data for many study sites provide the opportunity to examine how patterns of ground instability have evolved with time and how this relates to the processes taking place. This presentation will illustrate the evolution of techniques and approaches for analysing and interpreting ground motions through different case studies.

Through these applications, BGS has developed automatic AI and ML tools which examine not only the InSAR average velocity but also the time series to group areas of similar motion characteristics and to then detect when changes occur. The application of such tools to the EGMS time series and integration of results with BGS GeoSure national hazard susceptibility datasets provide a pathway to the ongoing interpretation of national GB wide InSAR datasets.