Ambient Seismic Noise Tools for Geothermal Exploration
-
Venue:
Geb. 06.42 - Room 001 (Seminar room) / Online
-
Date:
June 17 2025
-
Speaker:
Claudia Finger (Fraunhofer IEG)
Katrin Löer (TU Delft)
-
Time:
3:30 p.m.
Abstract
The prediction of seismic wave fields between stations using machine learning offers great potential for geophysical monitoring, particularly in remote areas or in regions with sparse sensor coverage. We introduce a novel encoder-decoder deep learning architecture that successfully learns the transfer function between seismic stations. By learning the complex signal transformations, this method enables accurate predictions of how seismic signals alter as they travel from one station to another. Notably, high quality predictions are achieved using only two days of data consisting solely of ambient seismic noise. Extending the approach from individual station pairs to entire seismic arrays, Virtual Seismic Arrays are introduced as a powerful proof of concept. By training the algorithm on all station pairs within an array, a set of predictive models is obtained that collectively form the Virtual Seismic Array. This enables the reconstruction of full-array recordings from a single reference station, even after physical sensors are no longer present. In the secondary microseism frequency band, beamforming analysis validates the effectiveness of Virtual Seismic Arrays by showing a high degree of agreement between the original and predicted waveforms. This novel application of encoder-decoder networks for modelling transfer functions has the potential to enhance seismic monitoring, while reducing the need for continuous sensor coverage. By reconstructing signals at multiple stations from a single reference station, the approach enables ongoing array functionality in remote regions while reducing costs and maintaining array capabilities.
Abstract
Accurate subsurface seismic velocities are crucial for drilling exploration wells, exploring geothermal resources, or locating seismic events. Due to their dispersive nature and prevalence in ambient seismic noise, surface wave velocities can be used to obtain shear velocities beneath seismic arrays. Localized shear velocity anomalies indicate the presence or absence of fluids; Temporal variations in shear velocities can indicate changes in fluid content or poisson ratio over time, i.e. during geothermal operations.
The toolbox B3AM (B3AMpy for Python) for three-component beamforming of ambient noise data provides a means to characterise the seismic (noise) wavefield and image near-surface seismic properties quickly and cheaply. Provided with three-component array data, B3AM outputs dispersion curves for pro-/retrograde Rayleigh and Love waves, estimates of wavefield composition, polarization and propagation direction as a function of frequency, and can be extended for surface wave anisotropy analysis. We will present recent examples of how these information can aid in exploring geothermal ressources.
For the GeoHEAT project, which explores a joint analysis of passive seismic and borehole geo-radar data for characterising and monitoring fractured geothermal systems, we implemented and tested the beamforming workflow for a novel nodal data set from the Kanton of Thurgau (CH). Besides dispersion analysis and source directionality, we consider wavefield composition and classify time windows with respect to their dominant wave type to inform and improve Green’s function recovery for ambient noise cross-correlation tomography.