Ground-penetrating radar (GPR) is a method that can provide detailedinformation about the near subsurface in sedimentary and carbonateenvironments. Classical interpretation of GPR data (e.g., based onmanual feature selection) is often labor-intensive and limited bythe experience of the interpreter. Novel attribute-based classificationapproaches, typically used for seismic interpretation, can providefaster, more repeatable and less biased interpretations. We presenta 3D GPR data set collected across a paleokarst breccia pipe in theBillefjorden area on Spitsbergen, Svalbard. After performing advancedprocessing, we compare the results of a classical GPR interpretationto the results of an attribute-based classification. Our attributeclassification incorporates a selection of dip- and textural attributesas the input for a k-means clustering approach. Similar to the resultsof the classical interpretation, the resulting classes differentiatebetween undisturbed strata and breccias or fault zones. The classesalso reveal details inside the breccia pipe that are not discernedin the classical interpretation. Using nearby outcropping brecciapipes we infer the intra-pipe GPR facies to result from subtle differencessuch as breccia lithology, clast size, or pore-space filling.
- Programområde 3: Energiressourcer