Travel time and amplitude inversion applied to the 1986 PASSCAL Ouachita experiment and the 1986 GLIMPCE Lake Superior experiment

An inversion method has been tested and applied to travel-time and amplitude data from the 1986 PASSCAL Ouachita seismic experiment. A spline parameterization of velocity and attenuation structure allows lateral and vertical gradients within layers as well as velocity discontinuities. A first arrival travel-time inversion applied to the PASSCAL Ouachita data has been used to image upper crustal structure. Twenty-one shotpoints at 10 km intervals were recorded along two 100 km deployments with 400 seismographs on each segment. The inversion results correlate well with a Triassic fault structure and with subsurface extent of geologic units as interpreted from a COCORP reflection profile and well data. An interface inversion has been tested and applied to reflected data from the experiment. Inverted interface depths are in agreement with those derived from a one-dimensional velocity model over the southern half of the PASSCAL profile. The inversion images a depth of 10-12 km for a mid-crustal layer which thins southward from a thickness of 10 km to 4 km. A lower crustal layer with an average thickness of 12 km and a Moho depth of approximately 29.5 km are also determined. The shallowing of the Moho over the northern 50 km of the profile matches the inferred location of a Paleozoic continental margin. Geophysical studies of the modern Atlantic continental margin provide the simplest comparison to the crustal structure derived here. An inversion for specific attenuation based on differences between observed and calculated amplitudes from the inverted velocity model images low attenuation values north of the Triassic graben structure. Independent travel-time and amplitude inversions allow the determination of relationships between velocity and attenuation. Application of the travel-time inversion to ocean bottom seismometer data collected adjacent to Line A from the 1986 GLIMPCE Lake Superior experiment confirms general trends of an independently derived forward model including lateral velocity variations across fault structures associated with the Midcontinent rift system. The RMS error of the 250 node inversion model is 0.045 s with velocities constrained to less than 0.1 km/s. Shallow crustal features of faults imaged by forward modeling are poorly constrained as indicated by low resolution values from the inversion model. Independent estimates of velocity, geologic thickness, and fault offset determined from converting two-way travel-times to depth indirectly confirm details of the coincident reflection profile interpretation