Consideration Of Automatic Waveform Interpretation Of Gpr Data To Detect Void Space Beneath A Concrete Highway

For waveform interpretation of ground penetrating radar (GPR) data, one must compute<br>the response of the earth to a specific source wavelet. This is not commonly done, because it<br>requires either prior knowledge of the source wavelet shape, or a good understanding of the local<br>earth model. When waveform interpretation is possible, then certain steps in the interpretation<br>process can be automated. This paper describes a successful case history of GPR waveform<br>interpretation used to solve a problem of locating voids under a concrete highway. It worked in<br>this case because the earth model was well constrained according to design specifications and<br>core information. Automatic interpretation is really just the creation and use of a computer<br>algorithm to recognize specific features in the data. The human interpreter must still do the work<br>of determining what the diagnostic features are, and then do a good job of writing the program to<br>recognize them. This process is only cost-effective for large surveys or those where the<br>distinguishing data features are already known.