Quantifying the Reliability of Ground Penetrating Radar at Archaeological Site

This project aimed to develop a method for quantifying the reliability of ground penetrating radar (GPR) at archaeological sites in terms of location and to account for type II errors. Seven GPR surveys and nine excavation units at the Burial Hill site in Plymouth, Massachusetts were used in creating this methodology. Archaeological entities had their location compared to their associated reflections in radargrams. These entities were split into feature classes based on their geometry; feature classes were then ranked in terms of discrepancy in location between GPR and archaeological data. Five excavation units also had all of their entities compared to GPR data in order to form hypotheses as to why some did not generate recorded reflections. The ranking method was also used to create a margin of error for the depth at which archaeological entities will be found when ground truthed relative to GPR data. Finally, the migration filter proved to be crucial in radargram processing for removing obscuring hyperbola tails as well as confirming the validity of assumed velocities.