REFRACTION SEISMIC MODELING AND INVERSION FOR THE DETECTION OF FRACTURE ZONES IN BEDROCK

We have investigated the response of several synthetic models of variable complexity to tomographic inversion using Rayfract. This software is fairly advanced and complex and offers many different options when inverting refraction seismic data. Using this program to investigate the detection of fracture zones in bedrock, its parameters may be roughly grouped in three categories: the inversion and weighting method used, whether single or multi-run will be employed and the intensity of smoothing. We have discerned that multi-run Conjugate Gradient inversion method, with CosineSquared weighting and a 2D Plus-Minus starting model can give fairly good results. Minimal smoothing is also essential for the quantitative characteristics of the detected zones to be accurately calculated, but this is a hyper sensitive procedure which may result in over or underestimations of zone velocity values. Generally, we have concluded that it possible to locate and characterize fractured zones in bedrock albeit with some limitations. It has been found that the imaging of the position and inclination of zones can be problematic especially when the zones are neighboring bedrock areas with small velocity contrast. The detectable depth extent of fracture zones can be followed to a certain depth, but deep zones give the same response as shallow zones due to geological noise. The width of the zone is almost always very accurate and overburden layers can be precisely defined when the interactive picking of branch points prior to inversion is carefully done. The velocity of a zone can be calculated with a good combination of inversion parameters. Moreover, as seen after reprocessing some of the Knappe tunnel data, tomographic inversion can pick up zones that cannot be interpreted traditionally. Finally, denser shot point spacing can bring about a noticeable improvement on the inversion results.