Two‐dimensional and three‐dimensional resistivity imaging in archaeological site investigation

Abstract In this work, the application and the effectiveness of two‐ and three‐dimensional non‐linear inversion algorithms in processing and interpretation of electrical resistivity tomography (ERT) data collected from archaeological areas are investigated in the framework of a new field technique for gathering three‐dimensional pole–pole tomographic data in a relatively small amount of time using standard archaeological prospection equipment. The inversion routine, for both the two‐dimensional and the three‐dimensional case, is based on a smoothness constrained algorithm and the forward modelling calculations are carried out using two‐dimensional and three‐dimensional finite element solvers respectively. Results of combined two‐dimensional inversions (quasi‐three‐dimensional) are compared with the full three‐dimensional inversions. Comparisons are carried out in relation to the optimum survey direction of gathering the tomographic data using the pole–pole array for synthetic data arising from three‐dimensional structures commonly encountered in archaeological sites. The response of the algorithms in the presence of noisy data was also tested. The algorithms were also used in the processing of real data collected from the archaeological sites of Sikyon and Europos in Greece. The results from the synthetic and the real data indicate the superiority of the three‐dimensional inversion algorithms in processing tomographic data. The reconstructed three‐dimensional images do not suffer from the artefacts encountered in the quasi‐three‐dimensional approach, owing to the three‐dimensional nature of the archaeological features. Most importantly, both synthetic and real data results indicate that a single survey direction is adequate to produce a valid three‐dimensional subsurface image when full three‐dimensional inversion is used in contrast to the quasi‐three‐dimensional approach, which would require that two survey directions be used to obtain satisfactory results. Copyright © 2006 John Wiley & Sons, Ltd.