Semblance-based imaging of shallow scatterers

The objective of this study was to investigate some of the limitations of a semblance-based technique for imaging of shallow scatterers. We targeted the upper 100m of the subsurface. The study was conducted at two different sites. The first site was an exposed granite outcrop, within the Panola Mountain Research Watershed (PMRW), located 25 km SE of Atlanta, GA. The second site was granite under a few meters thick layer of soil and weathered rock near the city of Lawrenceville, GA. Both sites are within the Georgia Piedmont and are underlain at shallow depths by fractured and unweathered crystalline rock. In the Georgia Piedmont, water resources are limited to surface reservoirs and shallow wells. There is a need for developing non-invasive techniques to detect and characterize fracture systems with water-supply production potential. The analysis consisted of two main steps. The first step was to calculate the semblance coefficient as a function of apparent velocity and azimuth. Semblance is a measure of similarity between multiple channels. It varies between 0 for no coherence and 1 for perfect coherence. For an array with diameter of 15m and frequencies from 100 to 1000 Hz, the precision in azimuth was +/- 10 deg. and about 10 percent in apparent velocity. Semblance coefficient for direct surface waves had maximums at back-azimuths in excellent correlation with the back-azimuths of the source locations (accuracy of 5deg.). The second step was an imaging algorithm. Seismic waves arriving at a given time in the coda are scattered from a point on an ellipse of revolution with its foci determined by the shot location and receiving array, and its size defined by the travel time and velocity of the waves. At both sites we used a set of 16 geophones with 100Hz corner frequency. The source was designed as a simple weight-drop source. At the PMRW the geophones were placed in a near circular array with aperture of 15m. The source was moved around the array at distances of 10 to 50m. The entire experiment was conducted on a granite outcrop. Surface waves dominated the records but dispersion was insignificant. The coupling between geophones and the rock proved to be a critical factor in accurately recording the seismic signal. Because the method had to be developed as non-invasive to protect the natural environment of the site, we could not drill or use devices that could damage the outcrop. Instead we used modeling clay to form a level platform and to attach the geophones to the granite. Sand bags were added to improve coupling to the rock. Two surveys were deployed at the Lawrenceville site on the soil surface. In the first survey, the geophones were placed along a line, 1m apart. The source was moved within 2 by 2m square above the geophone line. In the second survey, the geophone geometry was irregular, while the source was placed along a line. There was no problem with the coupling of the geophones and the soil at this site, but the dispersion of surface waves was significant.