Since Conrad (1925) first identified a midcrustal refraction event in central Europe seismologists have developed layered models of the continental crust with velocity steps at various levels. Modern analysis of crustal refraction data, including automated inversion, relies on ray theoretical techniques or wave theory for plane layered media, which generally parameterize the Earth as a series of grossly horizontal layers. The bias toward layered models is in part based on physical and geologic intuition, but it is to a large measure a result of available interpretation methods, for example layer-based reflectivity and ray tracing methods. In contrast, reflection seismology has produced a picture of a highly heterogeneous crust, with wavelength-scale variations in velocity existing at different levels of the crust in different tectonic regimes. Finite difference modeling shows that zones of small amplitude random velocity fluctuations not only produce short discontinuous reflection segments as observed on deep seismic reflection data but also produce strong wide-angle reflections. Some of the events identified as crustal wide-angle reflections in field data may in fact be the result of scattering from wavelength-scale heterogeneities and need not result from reflection from a first- or second-order velocity discontinuity. Significant changes in the large-scale velocity structure associated with zones of random velocity fluctuations produce relatively small changes in the dynamic properties of the backscattered wave field.

This observation is in disagreement with the common credo that the seismic refraction method is primarily sensitive to changes in the large-scale velocity structure. As a consequence, the ray theoretical interpretation of such wide-angle reflections as first-order discontinuities leads to crustal velocity structures that are nonunique and may be erroneous in detail, while correctly predicting average crustal velocity and thickness. We are not suggesting that crustal velocity does not change with depth on the large-scale; rather, we are proposing that crustal seismic events observed at intermediate and large offsets may be the result of scattering from wavelength-scale velocity fabric rather than specular reflection or refraction from first-order discontinuities. ¿ American Geophysical Union 1992