Graduation Year


Document Type




Degree Granting Department


Major Professor

Sarah Kruse, Ph.D.

Committee Member

Mark Stewart, Ph.D.

Committee Member

Mark Rains, Ph.D.


3D migration, FDTD modeling, covered-karst terrain, subvertical reflectors, GPRMAX


Ground penetrating radar (GPR) is widely used to identify locations of sinkholes in covered karst terrain in Florida. Some sinkholes serve as hydraulic conduits between the surficial and underlying aquifers. Their role is critical in determining the surficial aquifer response to pumping in deeper aquifers. Improved methods for discriminating between hydraulically active sinkholes and plugged sinkholes could help regional water management. In the covered karst of west-central Florida a clay-rich weathering horizon forms over the limestone. The clay-rich layer is in turn overlain by surficial sands. Ground penetrating radar profiles typically show a strong reflector from the top of clay-rich horizon as well as internal layering within sands. Active sinkholes are expected to have sandy conduits that broach the clay layer, and perhaps layering in the overlying sand indicative of ongoing subsidence. Three dimensional simulations of GPR profiles over sinkhole with and without conduits were run with the finite-difference time-domain (FDTD) program GPRMAX. Results from the synthetic surveys were then processed with standard techniques, including migration. The modeling confirms that conduits appear in GPR records primarily as gaps in the return from the clay layer. The modeling also shows that non-traditional survey geometries (varying antenna spacing and orientation) are unlikely to recover more information than traditional proximal transmitter-receiver separation. Also examined are GPR profiles and 3D grids over a set of active and inactive sinkholes in Tampa, Florida. Results from these surveys showed decent structural recovery of a small sinkhole similar in structure to that of the modeled ones. Indications of active subsidence and possible conduit structure were apparent from this data. Finally, the dense surveys served as a benchmark to compare interpretations taken with the same surveys at lower spatial resolutions and profiles with 2D-only processing methods in order to understand errors in analysis and interpretation that are possible from 2D surveys. Two-dimensional surveys, 2D processed and migrated, showed some similarity to the 3D results previously mentioned but contained more complexities and artifacts, which led to poorer interpretation ability.