190 Geophysics in Geothermal Exploration sensors. The coherent components of those EGFs, which can then be submitted to a seismic analysis to infer information about the subsurface, depend on the ambient seismic signal composition, that is, its spatial and temporal characteristics. The reconstructed wavefield is usually dominated by interface waves (Rayleigh or Love waves in onshore context, Scholte waves in offshore context, see Mordret et al., 2020), but can also provide coherent body waves if the ambient seismic signal characteristics allows to (Brenguier et al., 2020). The interferometric process itself is a sequence of signal processing operations (Bensen et al., 2007), which must be carefully parametrized to maximize the signalto-noise ratio (SNR) of the reconstructed coherent waves. After pre-processing two signals of equal durations recorded at two different locations, cross-correlations operations lead to the final EGF. Finally, the SNR of the reconstructed wavefield can be significantly increased by stacking multiple EGFs that have been reconstructed sequentially in time. This operation allowing the extraction of coherent seismic waves form the ambient seismic signal then opens the way to two types of seismic analyses, that can be applied to geothermal context studies (or to other geoscience contexts). The first is Tomography, where seismic properties (usually shear wave velocity, Vs) of the interface waves are analyzed over an array of sensors, to provide as an output 3D models of Vs spatial distribution within the subsurface. The second is Monitoring, where the changes in Vs value are measured in between EGFs reconstructed at different times. The following paragraphs provide a few elements about how the methods are implemented, what inputs are required and what outputs are expected. 6.1.3 Tomography Ambient noise tomography (ANT) aims to resolve a 3D shear wave velocity (Vs) distribution of the investigated area using the dispersion properties of surface wave reconstructed throughout the cross-correlation operation described above. Traditionally, a two-step inversion approach is conducted to map dispersion properties and then define a pseudo-3D Vs velocity model by stitching local 1D velocity models. First, group velocity dispersion curves of surface waves (usually the fundamental mode, but higher modes can also be included) are determined using a FrequencyTime Analysis (FTAN) over a frequency ranges (Levshin et al., 1972). This operation is performed by picking the dispersion curve within the FTAN diagram, as illustrated in Figure 6.3. Recovered dispersion curves are estimated from cross-correlated waveforms, hence averaging the dispersion properties of the medium along the ray between pairs of stations. To resolve spatial seismic velocity variation, inter-station dispersion curves are inverted into group velocity maps (fundamental and higher modes if available) defined over the selected frequency range using a straight ray seismic wave tomography approach (Barmin et al., 2001; Mordret et al., 2013) or Eikonal equation (Lin et al., 2009).
RkJQdWJsaXNoZXIy NjA3NzQ=