%X Modern GPR systems allow for acquiring densely sampled data sets also using different antenna frequencies. Here, we consider such a multi-frequency approach to image near-surface sedimentary structures at different spatial scales. Despite the steady technical development of GPR data acquisition, today’s interpretation techniques largely rely on single-frequency data sets typically interpreted in a manual and, thus, subjective and non-reproducible manner. To pave the way toward a more objective and reproducible interpretation of multi-frequency GPR data sets, we develop an attribute-based multi-scale workflow. We evaluate our flow by integrating information of synthetic 50 and 200 MHz GPR volumes modeled across complex sedimentary structures showing heterogeneities at multiple spatial scales. Our strategy results in a multi-scale facies model comprising major structural variations as characterized by the 50 MHz volume and structural details as resolved by the 200 MHz data. We conclude that this attribute-based workflow poses an efficient and reliable tool to interpret both single- and multi-frequency GPR data and, thus, can either be an alternative or a guide for typical manual interpretation approaches. %U http://resolver.sub.uni-goettingen.de/purl?gldocs-11858/10521 %~ FID GEO-LEO e-docs