We adapt and test seismic attributes techniques on a 2-D and 3-D multi-frequency GPR dataset recorded in
an abandoned limestone quarry, analogous to a speci!c set of hydrocarbon reservoirs. Our main objective
was to image the vertical and lateral lithological variations, the network of stratigraphic joints and fractures
and to characterize the rock mass based on the radar response. We apply semi-automatic horizon mapping
techniques using manually picked seeds (control points) on selected attributes, and automatic extrapolation
both on in-line and cross-line, starting from seed positions. We also apply Principal Component Analysis
(PCA) and cluster analysis on group data with similar multi-attribute response to reduce the total number
of calculated attributes by minimizing the interpreter's bias. We compare and validate the results with direct
outcrop measures, imaging a hydrocarbon reservoir analogue in 3-D to over 10 m beneath the topographic
surface.
