Understanding fiber optic sensitivity to a wavefield: A framework to separate site amplification from orientation effects

Interpreting amplitudes in Distributed Acoustic Sensing (DAS) data is challenging because the recorded signal is influenced by multiple factors.

To differentiate the impact of fiber orientation from site effects, we develop expressions of axial strain for different body wave polarizations. These expressions consider a linear fiber segment with any orientation in space. From these we explore array geometry properties and the potential of the DAS transfer function as a polarization filter. This last property arises from the polarity inversion characteristic of shear waves and the averaging nature of the gauge length. If the gauge length is set to be a loop instead of a linear segment then the DAS will average all azimuth for a horizontal loop, canceling SH waves. For a vertical loop, all dips are averaged canceling SV waves traveling within the loop plane. These results could reflect a link between DAS and rotational seismology. 

From these transfers functions, we develop a low-cost forward model based on ray theory that predicts amplitude recorded in a DAS array. Differences in amplitude between the modeled and observed wavefields relate to local site amplification from which, we create an amplitude correction factor. We evaluated this method using active seismic experiments from the PoroTomo dataset, successfully identifying regions with anomalous high amplitude responses consistent with the recordings following a magnitude 4.3. 

The results, together with the main elements of our approach, are transferable in many new sensing strategies, including optimization of fiber deployment geometry, generations of synthetic data and the acceleration and improvement of existing location methods through DAS-specific amplitude and phase corrections.
In summary, by exploiting the known directional sensitivity of DAS, we draw new insights from amplitude variations along the fiber array, treating energy loss as equally informative as energy gain in interpreting the wavefield.