Complex stratigraphic relationships, distinct lithologic facies, and variable bed thickness characterize channel-levee elements of turbidite sections examined in outcrop. The 3-D configuration of these features in the subsurface cannot be determined accurately using only logs in sparsely distributed and unevenly spaced boreholes. The missing interborehole information can be filled in by three-dimensional GPR surveys so long as the geological elements can be recognized in the data. This recognition can be aided by 1-D convolutional waveform modeling, which is able to separate multiples and refracted events from primary reflections, to predict severe amplitude loss due to attenuation in shales, to constrain reflection polarities and traveltimes, and to elucidate the role of interference within thin sandstone beds.
Two-dimensional GPR profiles and a 3-D GPR survey image the lowermost and largest of a series of laterally offset, vertically stacked channel sandstones of the Lewis Shale, near Dad, WY. We combine measured sections, borehole logs, and cuttings to form a one-dimensional, blocked framework into which are embedded the laboratory-measured electrical properties of four facies. Zero-offset, non-dispersive impulse responses are computed and convolved with the waveform used in the surveys to produce synthetic radargrams. Modeling applies a frequency-independent attenuation and includes multiples up to second order. These synthetics are compared to nearby GPR sections that have been datum static-corrected using differential GPS elevations. The present modeling makes possible the interpretation of four features of major geological significance—the base of channel, the top of a shale-clast conglomerate, a featureless sandstone, and eroded channel/levee mounds (erosional remnants).
This case history shows that the use of 1-D models can reveal internal consistency among geological mapping, borehole information, laboratory measurements, and radar stratigraphy.