Advanced Time-Series InSAR Analysis using ICOPS for Monitoring Surface Deformation of Fagradalsfjall Volcano during the 2021 Eruption

The Fagradalsfjall volcano erupted on March 19, 2021, marking its first eruption in nearly 781 years. During the six-month eruption period (March–September 2021), a total of 90 Sentinel-1 synthetic aperture radar (SAR) images were collected between January and December 2021. These images consisted of 60 frames from the Sentinel-1A satellite and 30 frames from the Sentinel-1B satellite, providing a short temporal baseline of approximately 6 days between interferogram pairs. The dataset was processed using an advanced time-series InSAR technique based on the Improved Combined Scatterers Interferometry with Optimized Point Scatterers (ICOPS) algorithm, which analyzed surface deformation through a combination of Persistent Scatterer (PS) and Distributed Scatterer (DS) points, collectively referred to as Combined Scatterer (CS) points. To refine the analysis, a convolutional neural network (CNN) was applied to evaluate the temporal patterns of the CS points, and the Optimized Hot Spot Analysis (OHSA) method was employed to spatially optimize these points by identifying statistically significant patterns between hot and cold points. In detail, PS points were identified using an amplitude dispersion index threshold of 0.4, in line with standard StaMPS processing procedures. For DS points, a combination of amplitude and phase information was used: the amplitude data helped detect statistically homogeneous pixels (SHPs) through a Generalized Likelihood Ratio (GLR) test, while phase information enabled analysis of spatial and temporal coherence between each interferogram pair. By combining SHP detection with spatial and temporal coherence, the DS points were selected for further analysis as part of the CS point combination. To enhance displacement pattern reliability, a CNN was employed to assess consistency based on correlation coefficients. OHSA, using Getis-Ord Gi* statistics, was then applied to identify statistically significant hot spots by clustering data according to z-scores and p-values, determining the spatial significance of the deformation patterns. Finally, validation of ICOPS results against GNSS measurements around the deformation area demonstrated consistency in observed deformation patterns. The analysis revealed that deformation around the Fagradalsfjall volcano was primarily driven by magma reservoir activity associated with dike intrusion beneath the surface, which was accompanied by increased earthquake events. Seismic activity in the region was visualized through cross-sections of earthquake distributions during the deformation period, providing deeper insights into the volcanic activity.

Acknowledgment: This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (No. NRF–2023R1A2C1007742).