Different ground-penetrating radar (GFR) techniques have been tested on the
same site in order to establish the performance and reliability of this me
thod when applied to civil-engineering problems. The Laboratoire Central de
s Fonts et Chaussees (LCPC) test site at Nantes, France, was selected becau
se it includes most of the underground heterogeneities commonly found in ur
ban contexts, such as pipes, small voids, etc. The GPR survey consisted in
recording measurements in tomographic (surface to horizontal borehole measu
rements), monostatic (2D surface profiling) and bistatic (Common Mid Point
[CMP]) modes above various buried heterogeneities. Different processing tec
hniques were also performed, such as tomographic inversion, 2D and 3D migra
tion, velocity analysis, as well as numerical simulations, the results of w
hich can be summarized in three points. (1) Although the different filling
materials of the site can be distinguished by velocity and attenuation tomo
graphy, the buried heterogeneities are more difficult to identify because o
f limited resolution related to angular aperture and Fresnel zone. (2) 2D s
urface profiling can detect the different shallow heterogeneities, such as
pipes and voids, down to a depth of several meters. Additional processing.
such as forward modeling and attenuation curve analysis, provides more quan
titative information related to the medium. A comparison between 2D and 3D
migrated data highlights the error introduced when the structures are consi
dered to be perfectly cylindrical. (3) CIC LP analysis gives relatively goo
d estimations of vertical velocity contrasts when the medium is layered. A
lithologic log can be derived assuming that the velocity changes are relate
d to material variations. (C) 2000 Elsevier Science B.V. All rights: reserv
ed.