09:00 〜 09:15
[SSS03-01] A workflow for modeling and analysis of Distributed Acoustic Sensing (DAS) data.
キーワード：Distributed Acoustic Sensing, Enhanced Geothermal System, Semblance
DAS technology is particularly suitable for microseismic monitoring application in geothermal environments. This instrumentation can resist to high temperatures (up to about 100°C or more) higher than the operational temperature of standard acquisition instruments (e.g., geophones), allowing the fiber to be located very close to the reservoir. For this reason, DAS is particularly useful for induced seismicity monitoring of Enhanced Geothermal System (EGS). Being of recent development, this acquisition technology still lacks appropriate modeling and analysis tools able to handle such a large amount of data without losing efficiency. Furthermore, open-access DAS datasets are still a rarity, if compared to other geophysical datasets (e.g., seismological data). Therefore, we aim to generate an open-access synthetic (but realistic) DAS dataset that may help the geophysical community to develop “ad hoc” data analysis methods suitable for this kind of data. In the presented work we make use of the spectral element modeling software 'Salvus', developed by Mondaic, which also allows the simulation of DAS data. In particular, it outputs a strain measurement between all points defined as receivers in the simulation. Using the repositories of DAS data collected at the geothermal test site Frontier Observatory for Research in Geothermal Energy (FORGE) located in Utah (USA), we tried to simulate realistic DAS acquisition conditions of seismic events related to low-magnitude natural seismic activity from the nearby Mineral Mountains and microseismic events related to hydraulic stimulation operations for the generation of an EGS.
In order to obtain realistic synthetic data, we first analyze the spectral properties of real noise waveforms by using the Power Spectral Density (PSD) Analysis. Starting from observed PSDs we model the synthetic noise waveforms using a stochastic approach. Then we add it to the synthetic event traces and compare them with the observed ones. We finally test a semblance-based event detector on a 1-hour continuous waveforms of synthetic data to evaluate the performance of the detector in different operational conditions (e.g., different noise levels and inter-event times).
In order to obtain realistic synthetic data, we first analyze the spectral properties of real noise waveforms by using the Power Spectral Density (PSD) Analysis. Starting from observed PSDs we model the synthetic noise waveforms using a stochastic approach. Then we add it to the synthetic event traces and compare them with the observed ones. We finally test a semblance-based event detector on a 1-hour continuous waveforms of synthetic data to evaluate the performance of the detector in different operational conditions (e.g., different noise levels and inter-event times).