Shallow Structure Characterization at a Groundwater Contamination Site: Seismic Processing and Tomographic Inversion

In order to image the heterogeneities in a shallow (less then 20 meters) buried paleo-channel where groundwater is severely contaminated by trichloroethylene (TCE), two vertical seismic profiles (VSP) and a seismic profile between the two boreholes (VSP-surface) were acquired. The surficial lithology consists of a mix of Quaternary sands and gravels overlying an incised clay layer acting as structural/stratigraphic trap for the polluting DNAPLs (dense nonaqueous phase liquids). The high-resolution data were processed following a standard seismic processing scheme, and inverted using traveltime and waveform tomography methods. The tomographic velocity models were subsequently utilized to depth migrate the VSP-surface data, both post-stack and pre-stack using explicit finite-difference extrapolators. In addition, Q values were estimated for the area of interest, directly from the imaginary part of the tomographic velocities, and using spectral ratio methods on the VSP data. Since the shallow structures can be characterized mechanically to some degree in terms of physical properties such as seismic velocity, the reflection seismic migrated images can be directly compared to the waveform tomography velocity models, given that the two methods produce results with comparable resolution. Generally, a reasonably good agreement exists between the results of the two different methodologies, keeping in mind that the migrated image shows impedance contrasts, whereas the waveform tomography measures velocity fluctuations directly. Both seismic and tomography images suggest that the geological structure in the target area is considerably more complex and heterogeneous than a simple layer of alluvium over an eroded clay layer, as is also suggested by lithology logs. The heterogeneity in the target area suggests that a conventional reflection processing sequence including CMP stacking and post-stack migration may be compromised because of large lateral velocity variations. As we show, pre-stack depth migration produces a better result than post-stack migration. Finally, the obtained Q values are comparable for the two methods and are in good agreement with visco-acoustic modeling and lab experiments existing in literature for similar conditions. As a final step, the best tomographic velocity model and migrated seismic image correlates with the lithologic logs available for the two VSP boreholes, and a lithologic interpretation of the target area is presented. A high-resolution shear wave survey is scheduled for September 2005, which, if successful, will allow us to develop elastic wave velocity models for the site, and further constrain lithologies.