Plate tectonics shapes the face of the earth and subduction and collision zones are among the most important features on Earth. Here, crustal material is recycled into the mantle or integrated into growing orogens. However, the processes active at depth cannot be studied directly and we thus rely on geophysical imaging methods to visualize the geometries that result from the ongoing processes. Additionally, these processes can be studied in fossil subduction and collision zones. However, the scales at which observations from geophysical imaging are made are orders of magnitude larger than those made in field-based studies of fossil subduction and collision zones. This thesis provides insight into how eclogitization modifies the physical properties of deeply buried rocks and what influence the resulting lithologies and their geometrical configuration have on geophysical imaging. In an interdisciplinary approach, I show how structures that are likely representative for those present at depth in subduction and collision zones develop and what their geometries at depth will be. I then derive their petrophysical properties and show how these are modified on various scales, and how this influences the detectability of such associations using geophysical imaging techniques. To do so, the island of Holsnøy in western Norway serves as a natural laboratory that is ideal to study eclogitization of crustal material. Geological mapping on Holsnøy constrains the geometric framework of the constituting lithologies and the scales at which such structures could be expected to establish. Previously, several authors have shown that many of the eclogite occurrences on Holsnøy are produced contemporaneously with ductile deformation forming shear zones at various scales. Our geological mapping aided by photogrammetry using drone images reveals that large parts of this exposed continental sliver were eclogitized statically without associated ductile deformation. This shows that even in domains with ongoing regional deformation, low-strain domains develop within the descending crustal material. Nevertheless, even the major shear zones that are exposed are only a few hundred meters thick, and thus far below the scale that is detectable by geophysical imaging techniques. However, geological mapping of the area suggests that the exposed structures are, at least in a qualitative sense, scale independent, suggesting that the same structural framework could be present at a larger scale in active subduction and collision zones. Measurements of P and S wave velocities of the exposed granulitic protolith and eclogites suggest that eclogitization of the lower crust causes three major changes of the petrophysical properties: (1) increased P and S wave velocities, (2) an increase of the seismic anisotropy, and (3) a decrease of the VP/VS ratio, suggesting distinct variations in the geophysical signal when the descending material is partially eclogitized. Additionally, testing the signal that the exposed shear zones would give in reflection seismic and receiver function studies reveals that the variations in shear zone structure indeed produces variations in the retrieved waveforms. Nevertheless, as the exposed structures are too small for geophysical imaging, the finite element method is used to calculate the effective properties of representative structures acting as an effective medium. The results show that the geometrical configuration of the constituting lithologies only has a minor impact on the P wave velocities and anisotropies of the resulting effective medium. Furthermore, our effective medium calculations on the kilometer scale show that eclogitization of crustal material can indeed produce significant seismic anisotropy. In this case, the calculated anisotropy reaches ~5%, which would produce a dependence of the retrieved signal in, for example, receiver function studies on the backazimuth of the sampled rays. Such backazimuthal dependence is indeed observed in active collision zones such as the Himalaya-Tibet collision system and the results presented here can thus be used to constrain the lithologies at depth, suggesting that the lower crust of India below the Himalaya is partially eclogitized along shear zones similar to those exposed on Holsnøy.