The vertical radar profiling (VRP) technique uses surface-to-borehole
acquisition geometries comparable to vertical seismic profiling (VSP).
Major differences between the two methods do arise due to the fundamentally
differing nature of the velocity-depth gradients and transmitter/receiver
directivities. Largely for this reason, VRP studies have so far essentially
been limited to the reconstruction of velocity-depth profiles by
inverting direct arrival times from single-offset VRP surveys. In
this study, we investigate the potential to produce high-resolution
subsurface reflection images from multi-offset VRP data. Two synthetic
data sets are used to evaluate a processing strategy suitably adapted
from VSP processing. Despite the fundamental differences between
VRP and VSP data, we found that our processing approach is capable
of reconstructing subsurface structures of comparable complexity
to those routinely imaged by VSP data. Finally, we apply our processing
flow to two multi-offset VRP data sets recorded at a well constrained
hydrogeophysical test site in SW-Germany. The inferred VRP images
are compared with high-quality surface georadar reflection images
and lithological logs available at the borehole locations. We find
that the VRP images are in good agreement with the surface georadar
data and reliably detect the major lithological boundaries. Due to
the significantly shorter ray-paths, the depth penetration of the
VRP data is, however, considerably higher than that of the surface
georadar data. VRP reflection images thus provide an effective means
for the depth-calibration and extension of conventional surface georadar
data in the vicinity of boreholes.