Details

• Thesis title: Monitoring subsurface elastic properties using ambient seismic noise: 2011 El Hierro eruption and Reykjanes geothermal reservoir
• Author: Pilar sánchez Sánchez-Pastor
• Date: June, 21st, 2019 10:30 am
• Place: Aula Magna Carmina Virgili de la Facultat de Ciències de la Terra

Thesis supervisors

• Dr. Martín Schimmel (ICTJA-CSIC)
• Dr. Jordi Díaz(ICTJA-CSIC)

Thesis tutor

• Dr. Alex Marcuello (UB)

Abstract

Monitoring the elastic properties of the subsurface is of a special interest to mitigate the associated risk with natural and artificial hazards. In the last decade, coda-wave interferometry has become an excellent tool to characterize the subsurface in a large variety of environments. Addition- ally, coda-wave interferometry applied to the ever-present seismic ambient noise enables a continuous retrieval of virtual-source responses that allows monitoring structural and mechanical changes of media.

Throughout this thesis, we perform a comprehensive study of seismic noise interferometry employing all the currently used methodologies to observe lag-time changes: Time evolution of waveform similarity, waveform stretching and moving window cross-spectrum technique. Furthermore, we intro- duce some improvements in order to increase the temporal accuracy and sensitivity of said methodologies to detect tiny medium changes.

In particular, we carry out the study in two scenarios with very different tectonic settings: the 2011 El Hierro eruption and the Reykjanes geothermal system (RGS). We compute the phase auto- and cross-correlation (PCC) of 1.5 years of continuous seismic noise records of all available seismic stations in both cases. The PCC provides an accurate and amplitude-unbiased mea- sure of coherence between two seismic traces and allows obtaining detailed daily seismic response of media.

In the case of the 2011 El Hierro volcanic eruption, through the analysis of waveform similarity time series of auto-correlations we clearly identify dif- ferent pre-eruptive phases, as well as the end of the main magmatic emission and three magmatic intrusions that occurred in 2012. We use probabilistic sensitivity kernels to locate the places of the highest dynamics, providing that the most a↵ected areas correspond to the magmatic accumulation zone and the extinct volcanic area of Tin ̃or. In this study, we also introduce the change point analysis approach in order to automatically detect significant changes in time series.

The second scenario consists in studying stress changes and potential de- formations of the subsurface caused by geothermal well operations at RGS. For that purpose, we retrieve and analyse time series of waveform similarity values and seismic velocity variations. However, the continuous production over a large area and various injection wells make challenging the detection of time-lag changes in the coda. To tackle this issue, we decompose the similarity time series into the time-frequency domain through the S- transform, which allows us to discriminate fluctuations associated to injection and production rate drops. Furthermore, we observe a slow seismic velocity decrease in the reservoir due to the water deficit as well as seasonal variations associated with the energy production demand.