Three‐dimensional inversion of automatic resistivity profiling data

Geophysical investigations through mobile multi‐electrode systems, such as the automatic resistivity profiling (ARP) method, can increase the size of the surveyed areas without jeopardizing the spatial resolution of the survey. The representation of the apparent resistivity data in maps corresponding to the different measuring dipoles is sufficient in most routine applications for outlining the buried archaeological structures. In specific cases where a more quantitative interpretation of the apparent resistivity data is demanded, a three‐dimensional resistivity inversion can provide the necessary tool for this purpose. This work investigates the possibilities and limitations of the three‐dimensional resistivity inversion in processing the ARP data. A three‐dimensional finite element smoothness‐constrained inversion algorithm was used. The active constraint balancing (ACB) method was also applied in order to enhance the stability and the resolving power of the inversion procedure. Resistivity models that are commonly encountered in archaeological exploration were used to generate synthetic apparent resistivity data using a three‐dimensional finite element forward modelling program. Inversion of the synthetic data showed that the maximum investigation depth of the ARP method is comparable to the length of the larger receiving dipole and cannot exceed the 2–2.5 m for the particular ARP device tested in this work. Archaeological structures buried within this depth range can be effectively mapped, while the resolution of the subsurface structures is related to the data acquisition parameters. The inversion algorithm was also used to reconstruct the three‐dimensional resistivity distribution from the ARP data set collected from the Andilly archaeological site in France. The results effectively showed that the three‐dimensional inversion can act as a complementary tool in acquiring a more quantitative interpretation model of the buried archaeological features.