Towards temporal monitoring using coda correlations of icequakes on Erebus volcano, Antarctica

Abstract:

Recent theoretical advances pertaining to the properties of multiply scattered wavefields have yielded a plethora of numerical and controlled source studies aiming to better understand what information may be derived from these otherwise chaotic signals. Where temporal monitoring is concerned, recent advances in data mining have allowed the recovery of remarkably time-coherent correlation functions from ambient noise, and have furthermore permitted the direct mapping of velocity changes due to large earthquakes. Though not directly representative of the Green’s function between stations, these time-coherent correlation functions have even been used to predict eruptive sequences on the Piton de la Fournaise volcano by detecting injection events. However, the exact nature of the correlation function in this case, particularly its coda, is unclear.Here, we seek to expand the concept of temporal monitoring to a more deterministic case, where we do in fact have a practical understanding of the correlation functions. In this study, we use a large network of short period and broadband instruments on Erebus volcano, Antarctica, to reconstruct inter-station correlation functions that converge towards a high degree of symmetry. We subsequently use a Markov Chain Monte Carlo algorithm to iteratively resample the time windows on which are built the correlation functions for maximum symmetry and coherence in time. These resampled correlation functions are then scanned for changes in decoherence, and these changes are then mapped in time and space to the volcanic edifice. We compare these changes to other evidence of temporal variation on Erebus volcano.