High-resolution electrical-resistivity, seismic-refraction, and seismic-reflection surveys were performed at three locations in the Inner Bluegrass Region of Kentucky along coincident survey lines in order to correlate results and determine which method is most effective at locating karst features in this area. The first two survey locations at Slack’s Cave and the Kentucky Horse Park were chosen in order to investigate known karst features. High and low electrical-resistivity anomalies were correlated to air- and water-filled karst voids, respectively. Seismic velocity anomalies, including parabolic time suppressions, amplitude terminations, and surface-wave backscatters, were also observed and correlated to these karst voids. These findings were applied to a third location along Berea Road in order to investigate undiscovered karst voids. Three seismic targets were selected based on backscatter anomaly locations and were aligned in a northwest trend following the general bedrock dip, joint orientations, and suspected conduit orientation. Overall, the seismic-reflection method provided the highest resolution and least ambiguous results; however, integration of multiple methods was determined to help decrease ambiguities in interpretation created by the inherent non-uniqueness found in the results of each method.

Although the petroleum industry has imaged reflections from the sides of salt domes, steeply dipping structures have not been imaged as reflectors outside of sedimentary basins; to do so requires appropriate data acquisition, prestack depth migration, and an excellent seismic velocity model. Poststack time migrated seismic images, normal moveout velocity analysis, well logs, and other geologic information are used to build the velocity model. In regions of interest outside of sedimentary basins, such as major strike-slip faults, seismic reflectivity is often sparse and little is known of detailed subsurface geology. Alternate methods of velocity model construction must be used. First arrival (refraction and turning ray) traveltime tomography is proposed to construct the preliminary velocity model for steep-dip prestack depth migration in settings with little a priori subsurface information. A densely spaced synthetic seismic data set with long-offset recording, modeled after a real survey across the San Andreas Fault, was constructed using a finite-difference algorithm. First arrival traveltimes were picked from the data and a velocity model was constructed using tomography. The velocity model was used to perform a Kirchhoff prestack depth migration of the synthetic shot gathers. The subsurface structure was sufficiently reconstructed that the velocity model could be refined through migration velocity analysis. A series of tomography tests was used to determine the spatial resolution limits of the velocity model. Isolated erroneous anomalies with sizes near the resolution limits were added to the velocity model derived from tomography and used as input for migration. This pessimistic test provided an adequate image and identifiable arrivals in migrated common image gathers, allowing the velocity model to be improved through migration moveout analysis. Data acquisition requirements for tomography include long recording offsets and times, larger sources, and dense spacings, very similar to the requirements for steep-dip reflection imaging. / Master of Science