Transcription

1 Authors versión of the paper: V. Pérez-Gracia, J.O. Caselles, J. Clapés, G. Martinez, R. Osorio, 2013, Non-destructive analysis in cultural heritage buildings: Evaluating the Mallorca cathedral supporting structures, NDT & E International, Volume 59, October 2013, Pages DOI: /j.ndteint Non-Destructive Analysis In Cultural Heritage Buildings: Evaluating The Mallorca Cathedral Supporting Structures V. Pérez-Gracia (1), J.O. Caselles (2), J. Clapés (2), G. Martinez (3), R. Osorio (2), (1) Universidad Politécnica de Cataluña (BarcelonaTech), EUETIB/CEIB, Dpt. Resistencia de Materiales y Estructuras en la Ingeniería, C/Urgell 187, Barcelona, Spain. vega.perez@upc.edu (2) Universidad Politecnica de Cataluña (Barcelona Tech), Dpt. Ingeniería del Terreno, Cartográfica y Geofísica, C/Jordi Girona 1, Barcelona, Spain. oriol.caselles@upc.edu (3) Universidad Michoacana de San Nicolás Hidalgo. University City, Av. Fco. J. Múgica S/N., Morelia, México. gmruiz@umich.mx Abstract Geophysical prospecting surveys are being increasingly used in non-destructive evaluations of structures, and several methods can be applied in the evaluation of cultural heritage buildings. However, accurate studies of cultural heritage structures usually need the application of combined techniques, being also historic and structural knowledge necessary. The present paper describes the application of two non-destructive testing techniques: ground-penetrating radar and seismic tomography, in the analysis of some structural elements inner geometries and physical properties. This job is part of a more complete project developed to define the

2 Mallorca Cathedral structural behaviour. Both geophysical methods are used in a complementary way. GPR allows to detect small anomalies (changes of about centimetres), and the results are used to select the most appropriate seismic tomography initial model. The aim of the study is to define the internal structural configuration as well as the stone quality. Results reveal the internal structure of columns, walls and buttresses, showing different structural elements. Also, even the visual inspection point to external damages, the detailed NDT evaluation indicates that the inner structure is in good condition and the ashlars are of good quality. Keywords: NDT, Mallorca Cathedral, Seismic tomography, GPR, cultural heritage. 1. Introduction The construction of Cathedral of Saint Mary, Mallorca (Spain), started in the XIV Century. Actually it is a representative Catalan Gothic Style building (figure 1a and b). This style, similar to European Gothic, is characterized by the airy distribution within the building space. The slender columns avoid the visual separation between the three naves of the church. The structure was designed to transmit mainly loads to buttresses. However, this structural solution implies bigger buttresses, reducing natural light. The columns are slim octagonal structural elements, built with limestone ashlars (see figure 1b and c). Three pairs of columns, close the high altar, are older and thinner than the other columns of the church. They are 1.59 m circumscribed diameter, while the last four pairs are 1.80 m (Domenge, 1999). Historical documentation describes structural problems during the Cathedral construction and use, most likely due to a non-appropriate loads distribution design. Nowadays, Visual inspections of this building evidence structural problems in most or the arches (Domenge, 1999), and cracks are visible in columns, buttresses and walls (González and Roca, 2003).

3 Preservation of this historical building is the objective of several projects, involving exhaustive non-destructive testing (NDT) studies, some of them by means of geophysical surveys. The global objective of these studies is to define accurately the dynamic structural behaviour of the monument. Geophysical prospection and borehole data point to the changes in the ground quality as a possible cause to the columns bending (Pérez-Gracia et al., 2009). Attending this conclusion, the supporting structures conservation state is evaluated with GPR and seismic tomography. This paper describes these surveys and discusses the main results. Figure 1. a) Mallorca Cathedral. b) Lateral nave. c) Detail of one octagonal column. d) Ashlars distribution in the column (dashed lines indicates ashlars junctions). e) Octagonal plant and inferred ashlars distribution in plant after NDT evaluation: dashed lines indicate the second ashlars row; the inner white square space was unknown material. 2. Methodology The application of more than one geophysical method has been demonstrated to be a powerful technique to solve or diminish uncertainties associated to all indirect evaluations (e.g. Hildebrand et el., 2002; Forte and Pipan, 2008; Pérez-Gracia et al., 2009). In this study, two techniques were selected attending the size of the columns and the known damages. High

4 frequency GPR evaluation provides enough resolution to define discontinuities or changes in materials, as well as the structure shape, but results cannot be easily associated to mechanical material properties. However, high frequency seismic techniques provide less resolution, but results can reveal materials information related with mechanical properties. Moreover, seismic tomography needs an initial model. In many cases, uniform initial models are used. Notwithstanding, convergence of the problem is not always possible by applying those simplified models and, sometimes unreal solutions are achieved. In this way, GPR data could provide enough and valuable information of the inner medium, allowing to define a more accurate first model. As a result, seismic tomography iterations could converge to a more realistic final model Ground Penetrating Radar (GPR) GPR is a well-known survey, widely applied to archaeology and cultural heritage (Goodman, 1994; Carcione, 1996; Pérez-Gracia et al., 2000; Pieraccini et al., 2005; Pérez-Gracia et al., 2008a; Solla et al., 2011). GPR antennas emit electromagnetic short pulses (1 to 60 nanoseconds) near the VHF/UHF band ( MHz). The pulses are transmitted towards the studied medium. Reflection of these pulses is produced in interfaces between zones with different electromagnetic properties. There, part of the energy is returned to the surface, arriving to a receiver antenna, and part is propagated through the discontinuity. Receiver antenna incorporates an electronic circuit, demodulator, connected to the amplifier and receiver circuit; then, electromagnetic arrivals cause the generation of an audio frequency band pulse that is sent through a highly screened cable to the central unit, where the signal is reconstructed, processed and stored. Each received pulse is show up as a track. So, moving the antenna on the medium surface, an image record is obtained revealing the existence of anomalies due to inner electromagnetic changes. Horizontal axis represents the antenna

5 position, while vertical axis corresponds to the two-way travel time (TWT). The conversion TWT into depths depends on the wave velocity knowledge. In the study of the cathedral columns, an average wave velocity could be properly estimated since dimensions can be measured (Pérez-Gracia et al, 2000). Notwithstanding, this value could be variable due to changes in the materials. Variations are expected in fracture or damaged zones because of the contrast between the air dielectric constant and the healthy limestone. Air in cracks produces most likely the wave average velocity increment. The average velocity of the wave in the medium depends on its different phases (solid or mineral components, liquid usually waterand gaseous or air). The combination of phases determines the electromagnetic behavior of the soil, including the wave velocity. Due the velocity in air is higher than in minerals, it could be expected that the presence of high number of cracks increases the velocity. In this way, damaged zones could be defined by evident changes on the wave velocity and appreciable reflections in discontinuities. Resolution depends mainly on the material characteristics and on the wave frequency band. As higher the frequency, greater the resolution, even penetration depth diminishes (Annan, 2003; Pérez-Gracia et al., 2008b). Dimensions of constructive elements analysed in this work allow the application of 900 MHz and 1.5 GHz centre frequency antennas obtaining centimetric resolution. Attending those previous considerations, the main objectives of the GPR study are three: 1) To define the inner structure of columns, buttresses and walls. 2) To detect, if exist, damaged zones in columns due to internal cracks. 3) To definea proper first medium model using all this information. This model will be used in the seismic evaluation.

6 Walls inspection was carried out in three selected zones of the main nave wall (P1 to P3 profiles in figure 2), in other three zones in the north nave (P4 to P6 profiles in figure 2), and another in the north front (P10 in figure 2). In each one of these zones GPR data was obtained with a 900 MHz centre frequency antenna. In the study of the buttresses, radar data was acquired in two profiles at different heights in three zones with a 900 MHz centre frequency antenna (P7 to P9 profiles in figure 2). Three columns (C3, C4 and C5 in figure 2) were intensively studied with the 1.5 GHz antenna. Data was collected from five profiles in each one of the eight sides (columns have an octagonal plant), in order to define properly the internal structure. Other columns were extensively studied, obtaining a single vertical profile in each one, with the 900 MHz antenna (C1 to C14 in figure 2). Figure 2. Cathedral plant and situation of the GPR profiles and seismic tomography: P1 to P10: GPR profiles in walls; C1 to C14: single GPR vertical profiles columns. C3, C4 and C10: columns analysed with GPR profiles in each side and with seismic tomography.

7 2.2. Seismic tomography Seismic tomography consists of the 2D or 3D reconstruction based on the travel wave velocity and/or amplitude. Depending on the transmission and reception device, it can be defined as reflection, diffraction, transmission and refraction tomography (e.g., Youn and Zhou, 2001, Cardarelli and de Nardis, 2001). Tomography images are usually shown in false colours, indicating the characteristics of each pixel in which the space is divided. The dimension of the pixel, that must be higher than the wave resolution limit, is related to the coverage provided by the wave trajectories and to the wave resolution. Pixel characteristics could be defined when one wave path crosses at least this pixel, having the equations system no null determinant. When the pixel size is defined smaller than the wave resolution limit, the defined property is not real, even the average of the nearby pixels will present the property value. In seismic prospecting, wave resolution is closely related to the error in the arrival time estimation, which depends on the wave frequency and velocity. In general, as greater is the frequency, better is the resolution. Then, generation and pick up of high frequency signals can significantly improve the results. Nowadays a great number of high frequency portable acquisition systems exist, allowing detailed evaluations in difficult access areas. In the study of the Cathedral columns, as in other high frequencies seismic studies, good contact between sensors and surfaces is required. However, the study of monuments must be done without damage the different elements. Both requirements were applied in the Cathedral columns study, where a very low weight sensor was fixed by means of wax, adding an elastic tape for instrument security. This solution guarantees proper signal acquisition in the case of frequencies lower than 10 KHz. Seismic tomography of transmitted is obtained exclusively from the first arrival of the waves, obtaining probably the best resolution, because first arrival is always detectable. The delay

8 between the first arrival and the other phases arriving from multiple reflections and refractions in the internal heterogeneities of the studied element, allows the clear identification of this arrival. Also, the first wave arrival time offers a precision of about 1/10 of its period, whereas for the rest of phases precision is estimated to be 1/4 of the period (Dobrin, 1976). The analysis of structure and quality of the materials was carried out by means of 2D transmitted wave tomography in three columns. A small section column (circumscribed circle diameter 1.59 meters) and two great section ones (circumscribed circle diameter 1.80 meters) were selected due to the external observed damages (cracks). The studied columns were also evaluated by means of intensive GPR survey (C3, C4 and C10 in figure 2). The source was an instrumented hammer. Data was acquired with a high frequency and sensitivity accelerometer, scanned at samples per second. The configuration of the device and the coverage of transmitted rays, in the hypothetic case of a homogenous column, are shown in figure 3. Figure 3. a) Measurement points on the column surface. b) Piezoelectric accelerometers used in seismic tomography; c) Source and accelerometers position on the surface column; d) seismic configuration and coverage in the hypothetical homogeneous column.

9 Non refracted rays travel in straight trajectories, being the equations system to solve this problem: xij = ti v j j Where t i is the measured travel time wave between the emitter (instrumented hammer) and the receiver (accelerometer), ν j is the velocity and x ij is the trajectory. The medium is divided in elements or pixels. The results is obtained as the sum of the values in each one of the pixels. In non-homogeneous columns, waves suffer refractions due to the changes in the velocities in the close pixels. The problem is solved with the iterative computation process Simultaneous Interactive Reconstruction Technique (SIRT), until convergence of the solution. The SIRT algorithm considers the rays curvature due to internal refractions (Sandmeier, 2003; Cardarelli and de Nardis, 2001; Michelena et al., 1993). 3. GPR survey results Twelve radargrams were obtained in the main and lateral naves walls (figures 2 and 4a). GPR imaging seems to indicate that a single 45 cm row of block stones layers constitutes most likely the structure, being possible two ashlars rows in some cases. Small irregularities (few centimetres) are detected, probably associated to the internal blocks sides (Figure 4b and c). However, there is not any important reflector inside the block stones. Then, it is possible to conclude that no significant internal cracks affect walls in the studied areas. Six radargrams were obtained to study the buttresses (P7 to P9 profiles in figure 2). GPR data seem to indicate that buttresses are constructed using block stones in both sides as an exterior structural covering, with irregular filling materials between both stone layers. In the edges, this structure seems to be changed by two block stones rows, being probably a heavier and

10 stronger structure. Reflection in the inner part of the buttresses is clearly marked. It probably indicates that they could be stuffed (Figure 4d and e). However, reflections on the junctions between ashlars present low intensity. This kind of reflection could be caused on junctions with thin mortar and lack of voids. Figure 4. a) Radar profile on a wall. b) GPR image obtained from a wall. c) Possible radar data interpretation, showing the wall formed with stone blocks without voids between them. d) Radar data obtained in buttresses. e) Radar image interpretation showing two different zones detected in buttresses. The most homogeneous area corresponds to one extreme of this structure. The non-homogeneous zone is probably due to filling materials between ashlars. Some picked columns were studied by means of an intensive NDT evaluation obtaining several 2D radar images from each one of the eight columns sides. The other cathedral columns were extensively analysed with a single profile. The extensive study of all the Cathedral columns suggests similar constructive characteristics in all them. Externally, each row is built with four external block stones, being each block a whole side of the octagon and half of the two adjacent sides (see figure 1d and e). Next row is rotated 45 degrees (see figure 1c, d and e).

11 GPR data acquisition is done by means vertical profiles in the centre of each column face, crossing single ashlars in a row, and passing along junctions in the next row (figures 5a and b). GPR images indicate that columns are built with five stone blocks, four external stones and one internal block. Discontinuities were detected in the evaluation at different depths (figure 5e), associated to the different stone blocks allocation. Notwithstanding, no important anomalies could be associated to voids of important cracks, and only possible discontinuities and fissures could be related to stone blocks junctions. On the other hand, significant changes in wave velocities are not observed. Then, probably the columns are entirely formed with limestone blocks, with no voids or filling unconsolidated materials in the inner structure. Figure 5. Radar profile on the columns (a) and distribution of the block stones (b). Different ashlars rows correspond to different ashlars allocation (c and d). Radar data (e) detects differences depending on the ashlars allocation. Possible fissures in junctions between external blocks and the central one are observed in some cases. Contact between ashlars seems to be shallower when only one outer block holds all the side of the column (Figures 5c, d and e). One possible interpretation of the radar images is resumed in figure 5e. This interpretation describes the columns built with massif ashlars, with no important voids or irregular materials in the column centre. The contacts between ashlars

12 seem to be clearly defined in many cases. Results from the most intensive GPR analysis and from seismic tomography are in agreement with this interpretation. The intensive GPR study was done in three selected columns, obtaining similar results than from the extensive evaluation. In general, GPR data from these profiles shows a quite constant TWT to the deeper reflection (reflection in the central section area). It could be interpreted as a constant depth to the inner stone. Also, velocity is probably rather similar in all cases, indicating homogeneous constructive materials. A typical velocity is measured taking into account the columns and walls thickness, being 10 cm/ns the average value. This velocity corresponds to dry limestone or sandstone. Different values of speed ranges can be found, for example, in Jol (2009), Davis (2004) or Hanninen (1997). The observed changes in TWT associated to the depth to the reflective surface could be associated to the changes in the stones position, because it depends on the distance to the centre of the columns side (figure 6). Radar data obtained at, approximately, 1/3 of the column edge (P1/3 in figure 6) show the anomaly at a constant depth in all the profile. However, radar data obtained approximately at, 1/10 from the edge (P1/10 in figure 6) show a symmetrical image that the central one (PC in the figure 6). Images obtained in those profiles show that TWT to the internal anomaly changes in PC profile depending on the row: the inner stone contact is close to the surface when the profile is on a stones junction, and TWT increases when profile crosses a whole stone. However, TWT in profile P1/10 is not dependent on the row. These results could point to define the internal shape of the stones: the inner surfaces are not parallel to the column surfaces. This geometry is most likely the cause of the anomalies due to diffractions in sharp edges. GPR data obtained from these five vertical profiles in each column side were also interpolated in order to obtain a synthetic cross section radargram (figure 6).

13 Figure 6. Synthetic profile obtained interpolating GPR data from vertical profiles in the intensive evaluation. 4. Seismic tomography results Seismic tomography images point to an inner sector built with a stone similar to the outer limestone ashlars. For further structural evaluations it is possible to conclude that columns probably are entirely raised with limestone blocks, and no voids of unconsolidated filling materials are used. As a result from the seismic evaluation, an average velocity was computed in each column. Results in columns C3, C4 and C10 (figure 2) are, respectively, 5600 m/s (figures 7a and 7b), 6450 m/s (figures 7c and 7d) and 6250 m/s (figures 7e and 7f). These velocity values are really high. Limestone presents high seismic waves velocity variability, between 1700 m/s and 7000 m/s, depending on three main factors: (1) The rock cohesion. Limestone quality is better as greater propagation velocity is. (2) Pressure. A compression growth produces wave velocity augment. Then, non-uniform velocity distribution could indicate possible non-uniform load distributions (Knight and Endres, 2005, Dobrin, 1976).

14 (3) Cement composition. Limestone seismic velocity is highly influenced by their cement. The high average velocity measured in columns C3, C4 and C10 probably denotes the great quality of the columns stones. In these columns, low wave velocity values (lower than 1500 m/s) are only observed in the 1%, 9% and 5% of them, respectively. Whereas the surface with very low velocity (smaller than 300 m/s) corresponding to bad quality or cracked areas, is a 1% in column C3 and C10, and a 3% in column C4. In column C3, sectors with very low velocity correspond to the junctions between stones. Therefore, it could be deduced that no deteriorated internal areas exist in the stones. In junctions, narrow bands of medium velocity (about 4000 m/s) probably indicate microcracking or wide mortar zones. Column C10 presents an outer very low velocity zone at about 10 centimetres deep, and two small low velocity areas near stones joints. Column C4 has several low velocity areas. Five of them are related to outer damaged zones. The sixth zone is detected in the interior and could be associated to a junction between three stones. Variations in high velocities are not homogeneous. Stones in column C3 exhibit the less variability: velocities between 5000 and 6000 m/s are measured in the 84% of the studied section. Column C4 has an intermediate variability, measuring velocities higher than 6000 m/s in the 70% of the section. Column C10 has the maximum variability, corresponding the 65% of the section to velocities higher than 6000 m/s. Figure 7 resumes the main results, showing the different velocities section and the separation between high, medium and low velocity areas.

15 Figure 7. Seismic tomography results. Wave velocity distribution in the section in columns C3 (a), C4 (c) and C10 (e). Separation between low, medium and high velocity areas corresponding to the columns C3 (b), C4 (d) and C10 (f). 5. Discussion and conclusions Damaged elements (columns, walls and buttresses) detected in visual and previous geophysical inspections were evaluated carefully with GPR and seismic tomography. As a result of the whole NDT evaluation, the inner structural shape of several structural cathedral elements was inferred. Also, the possible quality of the stones was also defined. GPR data provides a valuable initial model to obtain a seismic tomographic final model, reaching a quick convergence, because this method offers an accurate preliminary assessment of the internal structure. Seismic measures can provide information about material quality, based on the wave velocity values. The columns study indicates that these elements are probably solid (without stuffed materials) and built with regular stones, arranged in rows, and rotated 45º from the adjacent row. Each row is constituted by four outer prismatic ashlars, shaping a hexagon base. The centre is filling with a single and squared stone. Seismic wave velocities indicate that probably stones are of good quality, and no voids are inside the columns. The rotated ashlars distribution is

16 probably design in order to distribute properly the loads, because the central square stones hold up on the four outer stones of the row below. The zones affected by fissures, detected with the intensive studied seem to be insignificant, and damages are probably smaller than few centimetres. Column C3 shows quite homogeneous seismic velocity distribution, slightly increased in the North-East direction. The stone joints in column C3 present narrow bands of medium velocity (about 4000 m/s), probably indicating a micro-cracking zone or a wide zone of mortar. Medium velocity zones in the surface of the column correlate accurately with observed damages. Column C4 is characterized by important seismic velocity contrasts, showing low velocity zones in five external sides, corresponding properly to some visible damages. However, results indicate that external damaged zones penetrate less than 7 centimetres in the stones. The stone joints are most likely in good conditions, even a low velocity sector is observed in an inner stone corner, showing a possible internal damaged zone. Although column C10 is the smaller of the studied ones and have repairs in the South and South-West sides, seismic tomography indicates only two small damages areas close the stones junctions in the South and West side, and a large zone in the South-East side. In these three irregular sectors velocities are about 2500 m/s. On the other hand, the average velocity in column C10 is higher than in Column C3, probably indicating greater compression stress. GPR was also used to evaluate walls and buttresses. This part of the study also indicates that the naves outer walls are made by 45 cm rows of stones, without inner filling material. The inner sides of these walls probably present irregularities of few centimetres. These anomalies cause slight reflections on the radar signal.

17 The buttresses are made by two stones walls confining the inner filling formed probably with irregular materials and lime-based mortar. The edges of the buttresses have not these filling materials and they are built with double rows of block stones. Junctions seem to be in good conditions in the buttresses, and probably have not significant cracks, being the mortar layer very thin. Acknowledgements This work has been partially funded by the project New Integrated Knowledge based approaches to the protection of cultural heritage from Earthquake-induced Risk-NIKER funded by the European Commission (Grant Agreement n ), and by the Spanish Government, by the European Commission and with FEDER funds, through the research projects: CGL /BTE and CGL References. Annan, A.P., GPR for infrastructure imaging, In: International Symposium (NDT-CE 2003), Non-Destructive Testing in Civil Engineering 2003, September 16-19, 2003 in Berlin, Germany, Proceedings BB 85-CD, 12 pgs. Carcione, J.M., Ground radar simulation for archaeological applications. Geophysical Prospecting, 44: Cardarelli, E. and de Nardis, R., Seismic refraction, isotropic and anisotropic tomography on an ancient monument (Antonio and Faustina Temple AD 141). Geophysical Prospecting. 49(2), Daniels, D. (editor), 2004, Ground Penetrating Radar (IEE Radar, Sonar, Navigation and Avionics Series), The Institution of Electrical Engineers, London, U.K. Dobrin, M.B., Introduction to geophysical prospecting. MacGraw-Hill, Auckland. Domenge, J., L obra de la Seu. El procés de construcció de la Catedral de Mallorca en el tres-cents. Institu d Estudis Balears, Palma de Mallorca, Spain (in Catalan). Goodman, D., Ground-penetrating radar simulation in engineering and archaeology. Geophysics, 59(2): González, J.L. and Roca, P., Estudio, diagnóstico, peritación y en su caso planteamiento de actuaciones, sobre el comportamiento constructivo estructural de la Catedral de Santa María, en la Ciudad de Palma, Isla de Mallorca (Baleares). UPC, Barcelona, Spain (in Spanish).

18 Forte, E. and Pipan, M., Integrated seismic tomography and ground-penetrating radar (GPR) for the high-resolution study ofburial mounds (tumuli). Journal of Archaeological Science, 35: Hänninen, P., 1997, Dielectric coefficient surveying for overburden classification, Geological survey of Finland 396: Hildebrand, J.A., Wiggins, S.M., Henkart, P.C., Conyers, L.B., Comparison of Seismic Reflection and Ground-penetrating Radar Imaging at the Controlled Archaeological Test Site, Champaign, Illinois. Archaeological Prospection, 9: Jol, H.M. (editor), 2009, Ground Penetrating Radar Theory and Applications, Elsevier, Amsterdam, Netherlands ; Oxford, UK Knight, R.J. and Endres, A.L., An introduction to rock physics principles for nearsurface geophysics, in: Butler, D.K. (Ed.), Near surface geophysics. Society of exploration geophysicists, Tulsa, OK, USA, pp Michelena, R.J., Muir, F. and Harris, J.M., Anisotropic travel time tomography. Geophysical Prospecting. 41, Pérez Gracia, V., Canas, J.A., Pujades, Ll.G., Clapés, J., Caselles, O., García, F., Osorio, R., GPR survey to confirm the location of ancient structures under the Valencia Cathedral (Spain). Applied Geophysics. 43, Pérez-Gracia, V., García, F., Pujades, L.G., González-Drigo, R., Di Capua, D., 2008a. GPR survey to study the restoration of a Roman monument. Journal of Cultural Heritage, 9: Pérez-Gracia, V., R. González-Drigo, D. Di Capua, 2008b, Horizontal resolution in a nondestructive shallow GPR survey: An experimental evaluation, NDT&E International, 41: Pérez Gracia, V., Caselles, J.O., Clapés, J., Osorio, R., Martínez, G. and Canas, J.A., Integrated near-surface geophysical survey of the Cathedral of Mallorca. J. of Archaeological Science, 36: Pieraccini, M., Mecatti, D., Luzi, G., Seracini, M., Pinelli, G., Atzeni, C., Non-contact intrawall penetrating radar for heritage survey: the search of the Battle of Anghiari by Leonardo da Vinci. NDT&E International, 38: Sandmeier, K.J., REFLEX. Program for prospecting and interpretation of reflection and transmission data. Karlsruhe: Sandmeier ed. Solla, M., Lorenzo, H., Rial, F.I., Novo, A., GPR evaluation of the Roman masonry arch bridge of Lugo (Spain). NDT&E International, 44: Youn, O.K., Zhou, H.-W., 2001, Depth imaging with multiples, Geophysics 66(1):