Despite numerous techniques being investigated, detecting changes in ground deformation patterns derived from InSAR time series remains a complex and ill-defined problem. This presentation showcases the implementation and the performance of the BEAST (Bayesian Estimator of Abrupt Change, Seasonal Change and Trend) algorithm in decomposing the InSAR time series into three main components (trend, seasonality and noise), with the aim of identifying the time intervals during which changes occurred. The initial step of the analysis involves a synthetic dataset, designed to incorporate specific characteristics of real time series (e.g. slope changes, discontinuities, harmonic components), in order to assess the algorithm’s efficiency against ground truth. Following this validation step, the workflow is applied on real InSAR time series recorded in Slănic town, Romania. The purpose of the presented analysis is to be further integrated into a processing workflow for early risk assessment of ground movement.
Short Bio:
Ioana-Cristina Igret is currently enrolled in the first year of PhD at the West University Timișoara. Ioana holds a Bachelor’s Degree in Computer Science and a Master’s Degree in Software Engineering. Her research focuses on the analysis of satellite-derived time series and the detection of changes in ground deformation patterns associated with land subsidence.
