GROUND-PROBING RADAR AS A TOOL FOR HETEROGENEITY ESTIMATION IN GRAVELDEPOSITS - ADVANCES IN DATA-PROCESSING AND FACIES ANALYSIS

Pleistocene gravely braided-river deposits in river valleys constitute a large fraction of the natural ground-water reservoirs in Switzerlan d. The knowledge of the distribution and variability of hydraulic cond uctivity within these deposits are key factors for the estimation of w ater residence times and of description of large-scale mixing processe s in aquifers such as macrodispersion. It has been shown elsewhere tha t the spatial variability of hydraulic conductivity is related to the composition and the characteristic dimensions of sedimentary structure s, which are themselves related to the dynamics of ancient braided-riv er systems. In many contamination problems, sedimentological informati on is sparse and drill-core descriptions and pumping-tests only give a limited picture of the geometry of inhomogeneities. The ground-probin g radar (GPR) method is a promising tool for resolving changes of phys ical properties in gravel deposits at the scale of natural inhomogenei ties arising from changing sedimentary composition. However, the main limitation of GPR is the rapid attenuation of electromagnetic waves in subsurface sediments such as gravels, which leads to a limited penetr ation of the order of 10 to 15 m for a 2 50 MHz antenna. The objective s of our present work are: (1) To show how digital processing methods similar to reflection seismics may be applied for velocity and profile processing. These methods can improve both the resolution of radar pr ofiles, in particular at greater depths, and the determination of velo city distributions from CDP experiments. (2) To examine whether and to what extent the characteristic lithofacies of Pleistocene gravel depo sits can be recognized as mappable reflection patterns on ground-probi ng radar (GPR) reflection profiles in order to gain information about the geometry of inhomogeneities. Using modem digital data processing m ethods, such as band pass, high- or low-cut filtering, deconvolution a nd velocity analysis, much more significant information can be obtaine d from the recorded GPR field data-sets. Our results demonstrate that on GPR reflection images the basic fluvial forms such as (1) pool depo sits generated at the junction of two channels, and (2) channel deposi ts may be distinguished. Their shape and characteristic spatial dimens ions may be recognized from a series of profiles in different directio ns. Because the method can detect changes in the water content, the re flection image may be related even to small changes in the degree of s aturation of the sediments. Thus reflectors can indicate the changing composition of sediments.