GNGTS 2014 - Atti del 33° Convegno Nazionale

In the north area we also made a magnetic survey with a GSM-19GF Overhauser magnetometer in walk-grad mode, setting an acquisition frequency of 2 readings/s. Both the total magnetic field (TMF) and the vertical gradient of the TMF were acquired with lower and upper sensors at 0.2m and 0.8 m above the ground respectively. On the whole we acquired 5093 raw readings along 51 profiles with a profile interval of 0.2 m. A further interpolation showed that, along any profile, on average, we had a 0.3 m reading interval. After outliers rejection and bad quality data removal we worked on 4647 readings. Data were then destriped and the gradient and TMF maps were obtained with a triangular interpolation. In the south area we used a Radar K2 from IDS with a 200 MHz GSSI antenna. On the whole we made 41 profiles, each one about 25 m long. The profile interval was 0.25 m while the trace interval was 0.075 m. The trace’s length was 100 ns and they were digitized with 512 samples. All the profiles show, on average, a spectrum with a -3dB band ranging from 100 to 250 MHz. Raw radargrams showed again a lot of clutters and there were no or little evidences of clear coherent reflections in the x-twt domain. The processing was reduced and it consisted of: move start time, to set properly the zero time; dewow to remove low frequency effects; a Butterworth 4 poles bandpass filter (100-500 MHz); a non linear manual gain, increasing with time, to recover energy in the deeper radar section and a time cut to 50 ns. The choice of the filter bandwidth, larger than the -3dB band, has been done to preserve shorter wavelets. Indeed in the average spectrum there is still a platform of energy at -9dB up to about 600 MHz, Time slices, on the envelope of the Hilbert transform of the radargrams, have then been made summing up the samples over 2 ns window length. Results. Results are shown in Figs. 3a and 3b. The vertical gradient map, as the total field one, is noisy and it is likely pervasively influenced by the serpentine pebbles and blocks. A simulation run with the software POTENSOFT (Ozgu Arısoy and Dikmen, 2011) showed that a cube having 0.2 m side with a magnetic susceptibility of 0.05 (SI), 0.2 m below the lower sensor, gives a total field and an vertical gradient anomaly compatible, both in amplitude and wavelength, with some of the high spots found in the field. Such a spread noise hinder the recognition of any alignment. The magnetic map does not seem to have any relation with the underlying structures revealed by the GPR despite the noticeable size of the structures themselves. It should however be observed that the walls are built randomly assembling blocks of serpentine and other lithologies in such a way that they might also give a magnetic signal similar to an incoherent noise. In every GPR time slice clutters are thoroughly in evidence. All the top of the walls are very irregular and there are not flat surfaces in agreement with the ones already dug up. Clutters concentration between and aside the walls let us suppose that there could be a remarkable amount of ruined material. The tops of reflecting walls in the North area are more shallow than the one in the South area. A long diagonal reflecting body, with an incoherent direction with respect to the others, appears in the western part of the South area. Looking at raw data both magnetic and GPR surveys failed to find any structure. Only time slice processing of GPR data revealed a frame of buried walls limiting rooms with dimensions comparable to the ones already dug up. Comments and conclusions. Once again this case history pointed out the importance of the information on the target and surrounding environment materials in the early stage of the design of a geophysical prospection (Sambuelli and Strobbia, 2002). The results of the prospection allow for important preliminary evaluations concerning an unexplored area and are crucial in order to plan further excavations. In this area, anomalies showing straight lines have emerged, which confidently belong to two different buildings. The North area, in particular, shows the south continuation of two parallel structures already unearthed in trench F for a very short part. They respectively correspond to the outer west wall and to a inner wall of the same building and they both can be traced for more than 12 m. Other structures, approximately perpendicular to these ones, apparently delimitate two rows of quadrangular rooms showing recurrent dimensions (4-5 m). Further anomalies in the south-east corner of the North area and 194 GNGTS 2014 S essione 3.2