f Delineating karstic cavities with groundpenetrating radar and microgravimetry (Jura Mountains, Western Switzerland)

Ground-penetrating radar (GPR) and microgravimetric surveys have been conducted in the southern Jura mountains of western Switzerland in order to map subsurface karstic features. Many environmental and engineering projects require such mapping and often demand high accuracy. The study site, La Grande Rolaz cave (Fig. 1), is an extensive system in which many portions have been topographically surveyed. By using small station spacing and careful processing/modeling for the GPR and microgravimetric data, and by combining these data with topographic data from within the cave, accurate interpretations have been achieved. Four GPR profiles, 50-80 m long, were recorded across an area with thin topsoil and above the known extent of the cave (Fig. 1). The contour in Figure 1 represents the extent of the cave at eye-level height from the cave bottom, which corresponds to a depth below ground surface of 10-11 m throughout the cave. A center frequency of 100 MHz and a 0.2-m trace spacing were employed. A common midpoint profile was also recorded and subsequent velocity analysis yielded velocities of 0.10-0.13 m/ns. Processing consisted of applying a time-varying gain, time delay removal, bandpass filter, migration and topographic correction. Microgravimetric measurements were made every 2 m along the GPR profile locations. By using multiple measurements, hourly loops to avoid excessive drift, and techniques to minimize wind and sunlight effects, a precision of 0.025 mgals was achieved. Basic gravity corrections were applied to the data. The GPR profiles show continuous horizontal reflections with some disruptions (interpreted as fractured limestone beds) and several areas of dense diffraction patterns to depths of 250 ns two-way traveltime. Several diffraction areas on profiles 2 and 4 correlate well with known cave sections (compare A1-2, B in Figs. 2a-d with corresponding labels in Fig. 1). Other diffractions on these and on profiles 1 and 3 are interpreted as previously unmapped and apparently isolated cavities. Migrating with a constant velocity of 0.11 m/ns collapses most diffractions, allowing cavity geometries to be delineated and steep fractures to be interpreted (Cl-3 in Fig. 2d). Incomplete collapse of the deeper A2 diffraction is probably due to the velocity effect of the cavity that is indicated by the shallower A2 diffraction. Correlations between GPR profiles and microgravimetric/topographic cross-sections are good. For example, a -0.15-mgal microgravimetric anomaly (Fig. 2e) corresponds with the large cave section B in the migrated GPR profile 4 (Fig. 2d). After topographic correction and depth conversion, the top and bottom of cave section B are interpreted to be 3-4 m and 1213 m deep, respectively, which correspond with the subsurface topographic data (Fig. 2f). Interpretation of the GPR profiles is more confident after forward modeling and analysis of reflection strengths. The latter shows that cavities produce the strongest reflections and is, therefore, helpful in determining the shapes of the cavities. By matching synthetic hyperbolas with diffraction patterns, above-surface reflections from nearby trees (line D in Fig. 2b) can be identified. A synthetic radargram of profile 4, created with the subsurface topographic model (lower part, Fig. 2f), helps distinguish images of cave section B from those of small off line and unmapped in-line cavities north of B (E1-2 in Figs. 1, 2d).