P wave anisotropy caused by partial eclogitization of descending crust demonstrated by modeling effective petrophysical properties

Eclogitization occurs deep in subduction and collision zones inaccessible to direct observation. Field-based studies dealing with crustal material previously transformed at eclogite-facies conditions and exhumed to the surface provide information from the micro scale up to a few kilometers. On the other hand, geophysical methods aim at imaging the ongoing processes in-situ. However, these methods are limited by the achievable resolution and typically only sensitive to structures a few kilometers in size, leaving a large gap between the scales at which observations are interpreted. In this study we try to discern the implications of structures mapped in field-based studies to interpretations of geophysical imaging. We therefore calculated effective anisotropic P wave velocities for a suite of representative structural associations using the finite element method. The structural associations are directly extracted from observations of partially eclogitized assemblages on the island of Holsnøy in the Bergen Arcs of western Norway. Physical properties of the constituting lithologies are taken from laboratory measurements of the same rocks and the calculations are performed on a variety of scales, from the 20-m scale up to the kilometer scale to be able to predict how the effective seismic properties change with varying scale. Our results show that the P wave velocity of the effective medium is solely controlled by the volumetric fraction of the constituting lithologies and their elastic properties. We find that the structural relationship of the different lithologies has no significant influence on the resulting seismic velocities. P wave anisotropy, however, is controlled by the constituting lithology with the highest initial anisotropy and to a lesser extent by the modal abundance of the different lithologies. Further, our results show that seismic anisotropy is largely transferable across scales validating the assumptions often made when measuring seismic velocities on centimeter-sized sample volumes. On the kilometer scale, a scale that is potentially resolvable by geophysical methods, our results show that an eclogite-facies shear zone network such as the one exposed on Holsnøy would indeed produce a significant P wave anisotropy on a crustal scale. This anisotropy is produced by the eclogite-facies shear zones themselves even though eclogites are typically considered to be low-anisotropy rocks. Comparison of our results with active settings of continental collision and subduction zones reveals that eclogite-facies shear zones have the potential to produce a significant backazimuthal bias of the retrieved signal in geophysical imaging and underline the significance of seismic anisotropy as a tool to further increase the sensitivity of seismological methods to lithological variations.