On the Reliability of 2D Regional-Scale Velocity Model Building in Complex Geological Setting.

Active seismic surveys are routinely employed to study the geological structure of the deep crust and upper mantle. While the seismic waves are recorded at the receiver location, the path that they travelled within the subsurface from a source to a sensor remains unknown - as long as a representative model of physical parameters for a given geological setting is reconstructed. When sparse 2D seismic data acquisition is designed to investigate a geologically complex lithospheric environment, one needs to be aware of the fact that the propagation plane of the wave might be not aligned with the axis of the shot/receiver profile. This means that the seismogram recorded at the receiver position represents information gathered along the wavepath that is three-dimensional. This so-called out-of-plane propagation or 3D-effect is usually not mentioned during most of the 2D case studies, although the problem exists - especially when considering crustal-scale profiles, where the seismic energy propagates over distances of the order of hundreds of kilometres. In this work we bring attention to the 3D-effect during the academic regional-scale velocity model building studies from the wide-angle ocean-bottom seismometer (OBS) data. Using synthetic OBS data generated in 2D and 3D highly complex setting of the subduction zone, we show that the two wavefields differ significantly, due to the different paths travelled by the waves in 2D and 3D cases. This difference is not only visible for the reflected phases but also first-arrivals. To investigate the influence of the 3D-effect on the velocity model building we cast both datasets into 2D full-waveform inversion (FWI). We show that the 2D FWI of the data generated in the 3D model can not only lead to completely meaningless velocity model reconstruction, but can also be hampered by the lack of convergence since the first iteration. With the great number of 2D crustal-scale velocity model building case studies and the associated geological interpretations, it is important to consider that the subtle changes or artefacts in the reconstructed models might result from the 3D-effect and their interpretation as a structure might be vague if not supported by other available data.