Speaker
Description
Newtonian noise from ambient seismic fields limits the low-frequency sensitivity of the Einstein Telescope. Coherent noise cancellation based on Wiener filters constructed from seismic correlations is a widely used mitigation strategy, but its performance depends on accurate knowledge of the cross-spectral densities between seismometers and the gravitational perturbation at the test mass location. Previous studies have relied on analytical models assuming homogeneous half-spaces, or on adjoint-based spectral element method (SEM) simulations that require an independent run for each reference receiver.
We present an ambient-noise simulation approach based on randomly distributed surface sources using SEM. In contrast to the adjoint method, this approach yields cross-correlations between all receiver pairs in a single simulation, resulting in substantial computational savings when the number of receivers is large.
Our results show that even in a homogeneous half-space, cross-correlations involving underground receivers contain body wave contributions alongside Rayleigh waves, which cannot be unambiguously separated in the ambient noise field. In the presence of realistic geological layering and topography, wave scattering further complicates the wavefield. Consequently, array optimization cannot rely on analytical coherence models based solely on Rayleigh or body waves, and instead requires accurate geological models combined with numerical simulations. We further compare simulated and observed seismic spectra from co-located surface and borehole seismometers at the Sos Enattos P2 site.