Quality improvement of crosshole georadar tomography: pre- and post-inversion data analysis strategies

摘要

Input data of georadar tomographic inversion can be contaminated by systematic errors involved in data acquisition and preparation. Such errors are caused by mislocations of the antennas (e.g., unrecognised borehole deviation), traveltime delay problems or unconsidered anisotropy of the medium. If they are not recognised and not eliminated they may produce severe artifacts in the tomograms leading to misinterpretation of the results. However, specific errors produce significant anomalies in certain plotting schemes and certain artificial structures in the reconstructed geophysical parameter distributions. This is demonstrated using purposely contaminated synthetic data sets. These synthetic data examples illustrate how certain systematic errors, e.g., unrecognised borehole deviation, can be identified and distinguished from the expected real subsurface structure. Furthermore, a field experiment is presented carried out for fracture and void detection in a limestone formation, where such consequent data analysis helped to identify and remove discrepancies in the data. Tomographic results were clearly improved showing a more reliable and consistent behaviour. The field and synthetic data examples and calculations show that careful pre- and post-inversion analysis and handling of tomographic georadar data lead to a quality increase and to more reliable results.