Fine Detection of Transient Signals in GNSS Time Series Using Multi-Station Hankel Spectrum Analysis

Abstract

Fine detection of micro-transient deformations exhibiting diverse spatiotemporal patterns is crucial for unraveling geophysical processes in geodetic data sets, despite the challenges posed by high noise levels and correlated periodic signals. In this study, we extend the application of Hankel Spectrum Analysis (HSA, proposed by Shi and Ding (2023), https://doi.org/10.1029/2023jb026438) to record common modes of transient variability. This method employs a versatile multi-transient function to model various transient motions and decompose signals into three components: periodic oscillation, transient deformation, and background noise in GNSS time series. In multivariate data sets, we upgrade HSA to multi-station HSA that can better exploit the spatiotemporal correlations of geophysical fields, allowing for a more comprehensive extraction of transient motions shared across each series. This research first applies multi-station HSA to GPS observation arrays to reanalyze the history of micro-deformation events at the Akutan volcano. After meticulously filtering out non-tectonic motions (e.g., sea-level and loading effects) across semi-annual, annual, and decadal variability, our method successfully identifies repetitive exponent transient growth corresponding to the radial extension inflation of each cycle. Further numerical modeling of these continuous transient deformations enables us to track the migration of the geophysical magma source (shifted to the northwest at a depth of ∼4 km) and the pressure response of the storage system (residuals of ∼5.7 × 10⁶ m³). Since 2014 (cycle 3), there has been an abnormal occurrence of high correlation in transient earthquake swarms, indicating magma intrusion into the brittle rock.