**4. Discussion**

Interpretation of the geophysical results is suggested in Figure 13, where surveyed areas are classified in di fferent color classes by evidence of their archaeological potential:


**Figure 13.** Archaeological interpretation of joint geophysical results in five classes: areas with simultaneous presence of strong ARP and GPR anomalies (**R1** and **R3a**, orange); significant ARP anomalies with locally poor, weak or too noisy, GPR signals (**R2** and **R3b**, yellow); simultaneous presence of strongest ARP and GPR signals (**R4a**, red); areas with important ARP anomalies without GPR comparison (**R4b**, dark gray); areas with relatively weak, sometimes coherent, ARP spatial patterns without clear agreemen<sup>t</sup> or comparison with GPR data (remaining surveyed area, light gray).

After the joint analysis of ARP and GPR data, it was possible to classify some areas in order of potential archaeological importance. In particular, the most significant areas of the ARP survey could be divided into five classes on the basis of the agreemen<sup>t</sup> of apparent resistivity anomalies and the GPR signal behavior or the absence of GPR data. One of the most promising areas is the southern sector internal to the restricted archaeological area, indicated with **R4b** in Figure 13 (dark gray color, 2730 m2). This complex pattern of conductive and resistive anomalies is characterized by linear and sub-rectangular shapes of dimensions which can be attributed to macro-archaeological features. Nevertheless, the absence of higher resolution data, such as GPR, does not allow to fully analyze tridimensional relationships of the different ARP signals and the sector is classified as very promising but not completely explored with complementary geophysical tools.

The other most important group of anomalies, indicated with **R4a** in Figure 13 (red color), is contained in a semi-elliptical shape with the main dimensions of 75 m and 25 m (1460 m2), in spatial continuity with the previous anomaly. The area is characterized by a cluster of very high resistivity values with a main dimension of about 40–50 m at the deeper level of ARP measures (150 Ω·<sup>m</sup> at 1.70 m nominal depth). In this region GPR profiles were able to highlight parts of bigger importance inside the wider anomaly which correspond to the most resistive elements (Figure 12a, b). The area is likely to contain archaeological remains, in particular, because of the complex patterns of GPR signals which are difficult to interpret as geological features. The wide extension of this anomaly and its topographic position at the lower altitude of the agricultural field sugges<sup>t</sup> the interpretation as a possible accumulation of stone pieces, even of possible archaeological importance, due to ancient ploughing.

The third most important anomaly is indicated with **R1** in Figure 13 (orange color). It is an area 390 m<sup>2</sup> in extent, with 40 m by 13 m main axes, with apparent resistivity values up to 120 Ω·<sup>m</sup> at 1.7 m nominal depth. After the integrated analysis of ARP and GPR data (Figure 11b,c), the anomaly could be interpreted, even for this case, as a localized concentration of stone bodies and fragments. In fact, GPR reflections and di ffractions are characterized by high amplitude and spatial concentration without lateral continuity, at least at depths with an acceptable Signal to Noise ratio. Similar properties are recognizable for area **R3a**, which has an extension of 380 m2, 16 m wide and 38 m long. The comparison of ARP and GPR data (right anomaly in Figure 11b,c) emphasized a correspondence, even if weak, between the two datasets. Resistivity values are on the same order as anomaly **R1** (up to 120 Ω·m) and GPR signals present high amplitude in the northern section of the anomaly but are still readable in the southern one. Especially at the northern section, GPR anomalies do not show lateral continuation outside the resistive volumes. Because of these considerations, anomalies in areas **R1** and **R3a** can be potentially considered for fruitful archaeological investigations.

To interpret the other two regions bordered in Figure 13, **R2** (187 m2, 22 m × 12 m) and **R3b** (284 m2, 47 m × 8 m), yellow color, the joint analysis of ARP and GPR data was less e ffective, because in correspondence of the area **R2**, radar signal was present but not intense as in the other cases, while in correspondence of norther section of area **R3b**, it was a ffected by the high noise level. Nevertheless, lateral apparent resistivity contrasts and spatial features sugges<sup>t</sup> the opportunity for further investigations with other complementary geophysical methods and archaeological trials.

The planimetric analysis of apparent resistivity data in the remaining areas (light gray color in Figure 13, 17,830 m2) evidences a complex scattered distribution of spatial patterns with a low resistivity contrast which could be indicative of the presence of in situ archaeological features, less sensitive than possible accumulation of stone fragments to volume analysis, such as ARP survey.

Aiming at the estimation of the e ffective investigation depth of the ARP survey, an inversion procedure, based on a tetrahedral Finite Elements structure, was performed on two apparent resistivity profiles by means of the commercial software ERTLab. Pseudo-sections TI040017 and TI040062 were selected by considering the good concordance between apparent resistivity anomalies and GPR signals. The resistivity models were set up with 3D cells of 20 cm edge dimensions and inverted with smoothness constraints to achieve the final resistivity models. The tomographic results confirm the location of resistive bodies and the presence of an intermediate conductive layer (Figure 14). Some artifacts are present due to the scarcity (only three levels) of data distributed across the pseudo-sections: this is most evident for a shallow layer over the top of the first apparent resistivity level where most resistive values are distributed and characterized by a very high discontinuity along the horizontal profile. The same effect is quite common, even for traditional acquisition ERTs, typically with minor thicknesses [61]. The deepest bodies under the intermediate conductive layer still have a resistive behavior, but values are significantly lower than in the case of pseudo-sections.

**Figure 14.** Electrical resistivity tomographies obtained from ARP data: TI040017 and TI040062 pseudo-sections.

The spatial distribution of inverted data substantially illustrates the same patterns of pseudo-sections but most resistive values were obtained on the top soils instead of the bottom ones where the most intense resistive anomalies were located in the pseudo-sections: this opposing behavior was probably caused by data density and distribution which bring ERTs to very smooth variations at middle-bottom depths and to apparently noisy variations on the top depths. Comparing the ERT results and pseudo-sections, it is possible to observe a good agreemen<sup>t</sup> of depths for apparent interfaces and spatial resistivity transients, with a slight overestimation of depths under conductive volumes. A similar behavior is also described by Papadopoulos et al. [39,40]. Combining ARP, 0.50 m depth, with DTM data (Figure 15), it is possible to evaluate the independence and spatial relations of the widest ARP anomalies to each other. No clear spatial link is evidenced between the most promising areas identified in Figure 13. The only weak relationship which could be notice is on the eastern bottom areas where the strong linear anomalies are in connection with a northern wide area characterized by less intense resistive anomalies (D1 and E1 in Figure 6). The same connection is no longer present at deeper depths where the northern area (E2 in Figure 7 and E3 in Figure 8) gradually becomes less resistive and disconnects from the strongest anomalies to the south. This feature, although the low contrast of apparent resistivities, could be a marker of a geological transition in the investigated volumes, rather than archaeological remains. From the same three-dimensional comparison, no connection is recognizable along the eastward direction between resistive bodies under areas located at di fferent altitudes.

**Figure 15.** Tridimensional visualization of 0.5 m depth ARP data integrated with digital terrain model information (three times vertical exaggeration).
