The complexity of the crust from refraction/wide-angle reflection data

ThePmP wide-angle reflected signal from the Moho shows a wide variation in its characteristics when cornpared from region to region. If the earth's crust is simple and homogeneous and the Moho is a sharp discontinuity, supercriticalPmP wavelets are large and isolated in the 100 km to 300 km distance range when compared to the preceding signals on the seismic traces. If the crust itself has numerous short reflectors and the Moho transition zone is thick or badly disrupted from past tectonic disturbances, thenPmP is often poorly defined and difficult to identify in the coda of earlier arriving signals. These signals are poorly correlated because the reflectors themselves are in general not only discontinuous but tend to be distributed at various depths within the crust. The effect of the vertical velocity gradient in the crust is to make the reflective field for downgoing waves much greater than for upgoing waves. The large variations in reflection coefficients wiht angle of incidence has the effect of making the coda generated by a reflective machanism distance dependent. The reflected signals are also contaminated with scattered signals from smaller scale heterogeneities which may be distributed more uniformly within the crust.In conventional seismic inversion methods applied to crustal refraction experiments, emphasis is usually placed on obtaining a velocity model of the subsurface structure. These models are relatively simple and are limited in the amount of complexity which can be uniquely inferred from the data. The main problem arises because of the difficulties in the identification of the origin of the signals which are themselves often incoherent from trace to trace.In this paper it is shown how the conventional record section may be complemented with a normal moveout corrected intensity section which emphasizes areas of large signal complexities. Data from this intensity section is then used as input to obtain a quantitative measurement of a complexity parameter. These measurements may be used to infer or compare differences in crustal heterogeneity from one region to another. The discussion is illustrated with both numerical modeling data as well as data from recent crustal experiments which were conducted over the Canadian Shield.