Volcano deformation monitoring using a Sequential Monte Carlo approach, applied to the 2020 unrest episode at Reykjanes, Iceland.

Abstract:

Observations of surface deformation using satellite based geodetic techniques such as InSAR and GNSS are inherently useful for volcano unrest monitoring. GNSS stations, once deployed, are able to transmit data continuously and are routinely used for positioning in 8- or 24-hour intervals. Freely available Sentinel-1 SAR data, can provide spatially dense deformation estimates with repeat times of up to 6 days. Our interest in this study is in using the incoming data streams of GNSS and InSAR to obtain and update deformation source model estimates, of analytical deformation models, in near real time. To this end we employ a Sequential Monte Carlo approach (the Auxiliary Particle Filter algorithm), which is a Bayesian method that allows for propagation of the full posterior probability distribution of model parameters through time, updated each time new data is acquired. To investigate the performance of the algorithm in an ongoing volcanic crisis, we test it on an episode of deformation during volcanic unrest in 2020 on the Reykjanes Peninsula, Iceland, and present the results. We aim to develop a freely available software based on this work for the purpose of volcano deformation monitoring.