Three-dimensional modeling of the La Pastora Dolmen in Valencina de la Concepción, Seville, using photogrammetric techniques

Three-dimensional modeling of the La Pastora Dolmen in Valencina de la Concepci´ on, Seville, using photogrammetric techniques Rub´ en Martínez-´ Alvarez a , Juan Carlos Mejías-García b , Manuel Jos´ e Le´ on-Bonillo a , Antonio Miguel P´ erez-Romero a , Carmen Marín-Buz´ on a,* a Departamento de Ingeniería Gr´ afica, Universidad de Sevilla, Carretera de Utrera, km 1, 41013, Sevilla, Spain b Departamento de Prehistoria y Arqueología, Universidad de Sevilla, 41003, Sevilla, Spain ARTICLE INFO Keywords: Photogrammetry SfM 3D models Archaeology Prehistoric dolmens ABSTRACT In this study, we present the advances of photogrammetry applied to prehistorical research, focusing on the creation of three-dimensional models of the dolmens of Valencina de la Concepci´ on (Seville). Photogrammetry has proven to be an invaluable tool in the documentation and analysis of archaeological sites, enabling precise and detailed capture of ancient architecture. In this study, we applied photogrammetric techniques to capture high-resolution images of Dolmen de La Pastora, one of the most significant megalithic monuments of the 5th millennium BP in the Iberian Peninsula with great historical and cultural importance. This multidisciplinary focus allows us not only to visually explore the monuments in detail but also to analyse their internal structure, orientation, and possible construction patterns. Additionally, we explore the utility of these 3D models in archaeological education and outreach, offering researchers and the general public the opportunity to virtually interact with these historical monuments. In conclusion, this work manifests the possibilities the power of photogrammetry offers in the field of prehistorical research and its ability to revolutionize the way we study and understand our past. The generated 3D models not only enhance our appreciation of megalithic architecture but also provide an innovative approach to the analysis and preservation of endangered or deteriorating archaeological sites. With this contribution, apart from, we aim to advance the exploration of prehistory and inspire future interdisciplinary research in the fields of archaeology and image technology, providing possible visits to places that are difficult to access or where the constant passage of humans may cause deterioration to the World Heritage Sites. 1. Introduction Archaeological research in the field of prehistory, particularly concerning the documentation and analysis of archaeological sites, increasingly relies on accurate and thorough documentation. This includes not only the material items that identify them but also the spaces and places where their daily activities took place, both in terms of habitat and the funerary world. The discipline has undergone a profound transformation thanks to the implementation of advanced photogrammetry techniques. In this work, significant advances in the field of photogrammetry applied to prehistoric research (Georgopoulos et al., 1999; Cabrelles et al., 2020; Marín-Buz´ on et al., 2021a; Gonz´ alez-Qui˜ nones et al., 2022; Persico et al., 2022) are presented, with a particular focus on the creation of three-dimensional models of prehistoric megalithic monuments (7th to 4th millennia BP). These monuments demand that study and documentation procedures approach absolute accuracy, as only through this aspiration can a properly acceptable result be obtained, facilitating and promoting the analysis, interpretation, dissemination, and protection of this unique historical heritage. Photogrammetry, due to its capacity to obtain precise and detailed documentation of architectural structures from past eras, has become an essential tool in archaeology (Lucci et al., 2019). In the case at hand, we have employed advanced photogrammetric techniques to capture high-resolution images of Dolmen de La Pastora, one of the most prominent megalithic monuments on the Iberian Peninsula, dating back to the 5th millennium BP. This capturing process involved the combination of aerial and terrestrial images, utilizing both drones and conventional * Corresponding author. E-mail addresses: [email protected] (R. Martínez-´ Alvarez), [email protected] (J.C. Mejías-García), [email protected] (M.J. Le´ on-Bonillo), [email protected] (A.M. P´ erez-Romero), [email protected] (C. Marín-Buz´ on). Contents lists available at ScienceDirect Digital Applications in Archaeology and Cultural Heritage journal homepage: www.elsevier.com/locate/daach https://doi.org/10.1016/j.daach.2025.e00417 Received 10 June 2024; Received in revised form 25 February 2025; Accepted 16 April 2025 Digital Applications in Archaeology and Cultural Heritage 37 (2025) e00417 Available online 19 April 2025 2212-0548/© 2025 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC license ( http://creativecommons.org/licenses/bync/4.0/ ). cameras, allowing us to collect data from multiple perspectives (Marín-Buz´ on et al., 2021b; Le´ on-Bonillo et al., 2022). The subsequent stage of this study included the use of specialized software to process the captured images and generate extremely accurate 3D models of the dolmen and its architectural features. The use of this multidisciplinary methodology has allowed us to go beyond the simple objective of enabled a detailed visual exploration of the monument, but has also facilitated a meticulous analysis of its internal structure, orientation, and the identification of possible construction patterns. Furthermore, we have extensively investigated the applicability of these three-dimensional models in the field of archaeological education and outreach, opening the door for both researchers and the general public to interact virtually with these historically significant monuments. Ultimately, the objectives of this work are not only underscores the power of photogrammetry in prehistorical research, but also highlights its transformative capacity in our understanding of the past. The 3D models generated not only enhance our appreciation of megalithic architecture but also provide an innovative approach to the analysis and preservation of archaeological sites in a state of deterioration or at risk of disappearing. With this contribution, we aim to inspire and encourage future interdisciplinary research in the fields of archaeology and image technology, with the goal of deepening the exploration of prehistory and the protection of our valuable prehistorical heritage. 2. Materials and methods 2.1. Dolmen La Pastora The discovery of Dolmen de La Pastora, also known as Tholos de La Pastora, took place in 1860 (Tubino, 1876; Ca˜ nal, 1894), in the municipality of Valencina de la Concepci´ on, Seville, as shown in Fig. 1, in southern Spain. Its coordinates are UTM 29N 759834.47 E, 4144776.10 N ETRS89 (EPSG 25829), situated on the eastern edge of the Seville Aljarafe plateau, just 3 km from the ancient inlet that existed at the foot of this plateau, a result of the Flandrian Transgression around 6500-6000 BP (Arteaga and Roos, 1995; Barrag´ an Mallofret, 2016). This dolmen, along with those of Matarrubilla, Montelirio, and Ontiveros, belongs to the Archaeological Zone of Valencina, encompassing both settlement and its corresponding megalithic necropolis, dating from the late 6th millennium BP to the early 4th millennium BP. It represents the largest archaeological site of this period known to date in the Iberian Peninsula and one of the largest sites in Europe, covering nearly 500 ha (Mejías-García, 2017; Mederos-Martín et al., 2021; Schuhmacher et al., 2021) (Fig. 2). Dolmen de La Pastora is a tholos-type burial monument with a corridor measuring approximately 44.5 m in length that ends in a circular chamber with a diameter of 2.56 m, giving it a total scope of just over 47 m, making it the longest in the Iberian Peninsula (Fig. 3). The corridor is divided into 2 main parts: the first is a small atrium that provides access to the corridor itself, which in turn is divided into three sections of similar dimensions, preserving the original roof the second and third sections whereas the first has completely lost the roof and a great part of lateral walls. The chamber is circular, and its walls are arranged in rows of masonry stones that, using the technique of laying rows, simulate a dome that closes at the roof with a large granite stone serving as a keystone. The tholos is complemented by an earthen mound that completely covers it and highlights the structure in the surrounding landscape (see Fig. 4). Several studies have been conducted (Vargas-Jim´ enez, et al. 2012, 2019; C´ aceres, et al. 2014, 2019; Moyano et al., 2020) regarding its architecture, the lithological materials used in its construction, decoration, symbolism, and even its value, but none of them with the aim of promoting widespread dissemination, targeting both a technical and tourist audience. As a result of the lockdown measures brought about by the COVID-19 pandemic, the regional administration deemed it necessary to promote an activity that would facilitate virtual visits to the dolmen, without the need to be physically present at the site. To achieve this, they commissioned the creation of a photorealistic three-dimensional model that could be viewed on any computing device, both mobile and desktop. Furthermore, it was intended that this three-dimensional model would have the utmost detail so that researchers from various fields could use it, without the need to travel and/or repeat measurements, with the aim of deepening their knowledge and analysis. 2.2. Method and materials The method used in this study is based on the Structure from Motion (SfM) technique, combined with prior georeferencing in the external area and the establishment of Ground Control Points (GCP) in the interior, as depicted in Fig. 5, although they don’t necessarily have to be performed in this specific order. In the event that it is not possible to begin by establishing a Fig. 1. Location of Dolmen de La Pastora in Valencina de la Concepci´ on. R. Martínez-´ Alvarez et al. Digital Applications in Archaeology and Cultural Heritage 37 (2025) e00417 2 georeferenced system due to manifest impossibility, it will be necessary to at least position markers (temporary or permanent) that can be subsequently referenced (becoming Ground Control Points or GCP). 2.2.1. Georeferentiation Today, archaeological excavations tend to be georeferenced, and the official reference framework used in the Iberian Peninsula and the Balearic Islands is UTM ETRS89, (as stipulated by Royal Decree 1071/ 2007 of July 27, which regulates the geodetic reference system for Spain). This georeferencing approach greatly facilitates tasks related to planimetry, delineation, spatial analysis, and the proper protection of historical-archaeological heritage (Verhoeven et al., 2012) can be easily coordinated. In our case, since it is an underground construction that resembles a cave, georeferencing could only be carried out using GNSS technology outside the structure under study. To achieve this, a Leica GS18 Fig. 2. a) Extension of the Valencina’s archaeological site with the location of documented archaeological structures. b) Detail of the southern area of the necropolis with an indication of the main existing dolmens. R. Martínez-´ Alvarez et al. Digital Applications in Archaeology and Cultural Heritage 37 (2025) e00417 3 equipment was employed, which, when connected to the Andalusian Positioning Network (RAP), allowed for georeferencing of the exterior area of the dolmen with centimeter-level precision throughout the data collection process. For the interior part of the dolmen, the use of a total station was necessary. A Leica TCR705 model total station with an angular accuracy of 15 arcseconds and linear accuracy of 2 mm +2 ppm was utilized, with measurements taken using a mini-prism. This total station was also used to measure the exterior points taken with GNSS, enabling the georeferencing of interior points by combining data from different projection systems. First, the measurement of each exterior point to the dolmen was carried out using the GNSS device connected to the RAP, the Andalusian Positioning Network, with its GNSS network of 22 multi-constellation stations that the Andalusian Government offers to obtain highprecision positioning throughout the Andalusian territory through free differential correction services. Each point was measured ten times and averaged, with possible out-of-tolerance points being reviewed in the process. The coordinates of all measurements were obtained in the official reference system of the Iberian Peninsula and the Balearic Islands (UTM with datum ETRS89). In our study area, we are located in zone 29, so its EPSG encoding would be 25829. The accuracies obtained with this measurement mode are within the centimetre range. Then, those same points were measured employing the TCR705 total station. This measurement system guarantees millimetric relative accuracies, especially when using a precision mini prism at a low height (10 cm) complemented with a clamp tripod for better stabilization during the measurement of each point. This working mode generates coordinates in a relative system, with high internal precision but completely unlocalized in cartography. The same points were then measured using the TPS705. For obtaining coordinates of control points inside the dolmen, the Fig. 3. Simple plan sketch and section of the dolmen, indicating its parts and that which is a contemporary addition. R. Martínez-´ Alvarez et al. Digital Applications in Archaeology and Cultural Heritage 37 (2025) e00417 4 only viable option is to use the TPS705. The interior of the dolmen, especially the corridor area, presents difficulties due to lack of illumination, its narrowness, and extreme length compared to its width. To address the lack of illumination, it was necessary to use a portable spotlight to illuminate the mini prism before each measurement. For surveying the points with the TPS, it was first positioned outside the chamber, in front of the entrance door, from where at least three of the points located outside could be visualized (with the aim of using the third as a control). From this same position, a point (identified as 3 in Fig. 6) located at the entrance door was measured, and four more points were taken inside the chamber. Subsequently, the TPS was positioned inside the chamber in an area facing the corridor, so that its entire length and even the chamber at the end could be visualized. To enter the coordinate system implemented in the first station, a free stationing was carried out, a procedure referred to as such in Leica brand total stations. This procedure involves performing a reverse intersection combined with distance measurements to prisms located at points with known coordinates, which allows obtaining the coordinates of the station in three dimensions. To carry out this free stationing, measurements were taken to the points previously marked inside the chamber, obtaining a positioning accuracy of less than 2 mm. From this position, all the nails and targets located inside the corridor and chamber were surveyed. To verify the accuracy of the obtained coordinates, a final stationing was conducted in the chamber using the same free stationing technique. Points from the chamber and the last section of the corridor were used for this purpose. The measurement of point -3located at the threshold of the entrance door was taken from the first station located outside the chamber, resulting in an accuracy of less than 1 mm in each of its axes. Fig. 4. Detail of the access to Dolmen de La Pastora, Valencina de la Concepci´ on. Fig. 5. Methodology: Georeferentiation and SfM. R. Martínez-´ Alvarez et al. Digital Applications in Archaeology and Cultural Heritage 37 (2025) e00417 5 All coordinates obtained with TPS are in a relative system, but it is necessary to work with georeferenced coordinates, as explained earlier. To perform this transformation, a series of operations must be carried out: - Remove the projection of the exterior points measured with GNSS. To achieve this, a command implemented in TAO (Topografía Asistida por Ordenador, Computer-Assisted Topography), integrated into CAD software (Computer Assisted Design), was utilized. The algorithm in question calculates the average linear anamorphosis coefficient of all points and applies it inversely to eliminate the deformation inherent in UTM projection. It is important to recall that points surveyed with GNSS have an intrinsic accuracy within the centimetre range (1–2 cm), while those surveyed with TPS have an accuracy within the millimetre range. - Following it, the measured points in relative coordinates are input into the drawing environment, and using another TAO command, common points in both surveys are identified. Through a least squares adjustment, these points are then translated, rotated, and adjusted in elevation relative to the GNSS-measured points, from which the projection had previously been removed. This is done to avoid mistakes in adjustments caused by the inherent deformation of the UTM projection. The GNSS-measured points, with an internal precision within the centimetre range as previously indicated, are discarded, as this level of accuracy is insufficient for the precise photogrammetric work being conducted. - Finally, using another TAO algorithm that restores the UTM projection for all transformed points, a correctly georeferenced point cloud is obtained with an internal precision within the millimetre range. In this way all GCPs and points used for georeferencing are in the EPSG25829 coordinate system, and millimetre accuracy in the photogrammetric processes is guaranteed. 2.2.2. Ground Control Points (GCP) for SfM The use of Ground Control Points (GCPs) in the Structure from Motion (SfM) methodology allows us to establish a connection between the model and its real-world location, providing precise metrics to the object or area of interest (Oniga et al., 2020). These points should be distributed across the area of interest to help achieve accurate results (Villanueva and Blanco, 2019). GCPs were marked with targets or nails, depending on the area to be treated. Large nails were used in the exterior of the dolmen and were captured with the GNSS equipment, as the required precision to locate the structure was more than sufficient. In the interior, circular targets and small nails were used, which did not disrupt the area. However, both the external and internal points were measured using the total station (TPS) to ensure that we obtained control points with a precision similar to that provided by the technique to be used: photogrammetry, which entails millimetric accuracy. The square targets used for the exterior were 10 cm on each side and divided into four smaller, equal squares, alternating between white and black colors. For the interior area, circular PLA (Polylactic Acid) targets, measuring 10 cm in diameter, were used, with a visible side marked by a cross formed by two perpendicular diagonals, painted in white and black or yellow and black. Steel nails with semi-spherical heads, 4 mm in size, were also employed. The targets were placed in spacious areas near the study zones, but not within them. In narrow areas and on the pavement inside the dolmen, small steel nails were used. These were placed in cracks or areas with compacted soil (Fig. 7) to avoid disrupting the Cultural Heritage Site (Bien de Inter´ es Cultural or BIC). The points were placed alternately to the right and left of the corridor so as to improve the longitudinal geometry in the photogrammetric process, as the rest of the complex did not allow markings to be made without altering the surroundings. In the chamber, marks on the existing walls, floor, and ceiling due to vandalism after the discovery of the environment were used. 2.2.3. Images taking The photographs were taken with a CANON Powershot G3X camera, with a focal length of 8.8 mm (the equivalent focal length is 24 mm since the sensor is 1 inch), and an image size of 20 Megapixels (5472 x 3648 pixels in 4:3 format). The camera was set to ISO 200, with a shutter speed ranging from 1/15 to 1 s, an aperture of f/8, and an exposure compensation of −0.3 EV. The aperture priority mode was used, along with a multi-point metering mode. Images were saved in both RAW format, for white balance adjustments in certain conditions, and JPG. The choice of this camera model is based on several factors, including its compact size, which facilitates handling in the field; its proven effectiveness in previous data collection work, ensuring reliable results; its ease of use, allowing for an efficient workflow; and the fact that it is a personally owned work instrument, ensuring familiarity with its operation and immediate availability. All shots were taken by fixing the camera to a tripod and using a wireless clicker to obtain images with the greatest possible stability (Fig. 8). While the conditions for data collection outside the dolmens were excellent—coinciding with days of diffuse light—the data collection inside was much more complex. This required turning off the installed lighting in the World Interest Property and projecting our own lighting Fig. 6. External and internal GCP. R. Martínez-´ Alvarez et al. Digital Applications in Archaeology and Cultural Heritage 37 (2025) e00417 6 system composed of a total of four luminaires, although depending on the area being photographed, it became necessary to use a smaller quantity. This lighting system helped to overcome two common problems when using artificial light indoors: 1) chromatic alteration in images caused by variations in the light’s temperature was corrected using the white balance feature in CANON’s Digital Photo Professional 4 software, and 2) the generation of areas with hard shadows, excessive light, or backlighting. During the data collection preparation, the use of a ring light mounted directly onto the lens was considered for the hallway area to avoid any type of shadows. However, the limited distance between walls made this unfeasible, as the homogeneity of the lighting obtained in the images was insufficient. The data collection was basically divided into three distinct areas: the initial part of the entrance to the tholos (the casemate that was without the original roof and is much wider and higher than the rest of the tholos), the corridor area and the final chamber. Due to the linear distribution of most of the walls to be photographed, the images were taken mainly in horizontal rows (Fig. 8), following a zigzag pattern to maintain the necessary overlap between adjacent images, both in those taken consecutively and between images in adjacent rows. In this way, the continuity of the sequence was guaranteed. What is more, the procurement was carried out in sections according Fig. 7. Targets and nails utilized for the signalization. Fig. 8. 1–2: Arrangement of luminaires with photographic equipment on tripod and wireless push button. A-B-C: Detail of images capture sequence. R. Martínez-´ Alvarez et al. Digital Applications in Archaeology and Cultural Heritage 37 (2025) e00417 7 to: the lighting provided by the four luminaires, the space available in the corridor and in the chamber, and the detail of constructive elements. The luminaires were placed at intervals of about 4–4.5 m to ensure good lighting in strips of approximately 3 m continuously along the entire corridor. This allowed us to achieve the highest level of detail on the walls, floor, and ceiling, while ensuring that the wiring infrastructure did not interfere with data collection and, of course, without affecting the overall World Interest Property ensemble (see Fig. 8). In the casemate, imaging focused on the two incomplete walls in height and on the access to the covered part of the corridor. In this first area, as it is a large area with some non-linear elements (entrances, protrusions and reliefs of the access to the covered corridor), the images were taken by combining the linear sweeps with some more specific captures to record everything in detail, using the four available luminaires to avoid dark areas in these images. In the corridor areas almost always two luminaires were used as useable space was scarce, with maximum heights ranging from 1.42 m to 1.77 m and corridor widths ranging from 0.75 m to 0.90 m at the top and 1.05 m at the bottom almost all the way along the corridor. In the final chamber, more luminaires could be used, despite the limited space inside. For this last image capture, a continuous data sequence was chosen for the entire circular surface, from the floor to the ceiling, formed by a large piece. Using our own lighting required a temporary electrical installation that needed to be avoided in the image capture (or removed during the photogrammetric process). 2.2.4. Photogrametry (SfM) This technique uses images to generate the geometry of the photographed elements. Nowadays, it is carried out entirely digitally, using specific software. The process it performs is based on identifying common elements (groups of pixels) for each of the images, and then identifying these elements across images taken from different positions, creating a network of related elements across multiple images that generates a global correlation. This enables the recreation of the geometry of the observed part. This process is repeated until the entire object or the observed part of it (Salaün et al., 2017) is captured. A total of 6891 photos were taken over several days, although approximately 360 had to be discarded as they did not agree the quality criteria set (sharpness, brightness, and focus), supposing a total of 5.22 %. Data collection ensured an overlap of over 80 %, resulting in a minimum of 9 photographs covering every surface inside the tholos. It should be noted that standard SfM techniques typically require trifocal overlaps, to calibrate the cameras and reconstruct a scene (Salaün et al., 2017). Once all the images are captured, they are loaded and processed in the Agisoft Metashape software. This program executes a series of automated algorithms, which are adjusted through guided steps throughout the process. In summary, the process is based on identifying characteristic elements (key points and tie points) in each of the images. Subsequently, the program compares the tie points of each image with those of the rest of the images to determine which ones are adjacent to each other. Once determined, the program recreates the geometry of the visible part by observing the differences in position between several of these common elements among the neighbouring images and generates a low-density point cloud (Yang et al., 2016) (Fig. 9), which also serves to determine the relative positions from where the capture was made (the location of each camera). In this way, the process manages to recreate this situation between all pairs of images that it has determined as neighbours, thus generating what is known as the sparse point cloud. These set of algorithms are known as Bundler adjustment. Then, the GCPs are then loaded, and their validation is carried out manually on the images in which these points appear. In this way, the model is dimensioned and georeferenced (see Table 1). In Table 2, you can see the record of the control points with their distance errors and pixel values. Once this process is completed, an optimization of the camera positions is carried out, finalizing the synchronization of the point cloud with the reference framework established by the GCPs (Fig. 10). From this point, depth maps and dense point clouds are generated, enabling the creation of 3D meshes. As the final part of the process, texture is applied to the meshes, resulting in a high-resolution photorealistic model. The detailed process can be observed in Fig. 11 (Marín-Buz´ on et al., 2021b). This generates a model that is extremely useful for other researchers, providing them with detailed and accurate 3D information. However, the high level of detail results in file sizes that can be challenging to Fig. 9. Scattered point cloud in the corridor and camera arrangement. R. Martínez-´ Alvarez et al. Digital Applications in Archaeology and Cultural Heritage 37 (2025) e00417 8 handle without specialized softwares and high-capacity computers (e.g., Intel Core i7-6700 processor at 3.4 GHz, Radeon Pro WX-4100 graphics card, SSD hard drive, and 64 GB of RAM). On the other hand, if the goal is to make the model accessible to a wider audience through dissemination, access to the model cannot be as demanding in terms of required software and hardware, and it is preferable that it can be viewed on mobile devices (smartphones, tablets, etc.). To achieve this, Blender, an open-source, cross-platform software, was used to optimize the model without compromising on high resolution and visual quality. This optimization significantly reduces the file size, making it easier to view efficiently and within a reasonable timeframe on almost any device (Rohe and Jones, 2022). 2.2.5. Retopology Retopology is a technique used to create a simpler polygon mesh from an existing 3D mesh, essentially redrawing a new polygon structure into a more controlled and efficient mesh without sacrificing visual quality. When models become too heavy, as in this case, they require excessive computational calculations that can affect rendering and virtualization performance. The process involves optimizing the 3D model generated by creating a new low-resolution 3D model while maintaining quality and realism through the implementation of normal maps and optimized textures. To carry out this retopology and mesh optimization process, two essential software tools were used: Instant Meshes (Wenzel et al., 2015) for simplifying the mesh while respecting its geometric structure to the maximum extent, and Blender for retopologizing the mesh. We decided not to simplify the mesh through automated decimation in Blender or Metashape because, although it is an effective technique for simplifying the mesh, the result did not adequately maintain the Table 1 Correspondence from absolute coordinates to relative coordinates. Absolute Coordinates Relative Coordinates X Y Z X Y Z A759748.189 4144785.881 145.125 982.672 5050.172 9.208 C759785.321 4144744.848 144.617 995.691 4996.410 8.700 D759779,459 4144752,577 144.791 994.234 5005.996 8.874 G759790.274 4144749.362 144.959 1002.197 4998.008 9.039 H759787.125 4144754.838 145.299 1002.046 5004.321 9.379 3759792.368 4144753.472 144.61 1005.998 5000.618 8.693 Table 2 GCP coordinates (ETRS89, UTM-29N) and errors. GCP X Y Z Error_(m) E_(Pixel) 21 759808.421 4144762.766 144.104 0.000975 2.446 22 759809.216 4144763.04 144.125 0.002042 0.987 23 759810.43 4144763.099 144.074 0.000332 3.849 24 759811.058 4144763.805 144.044 0.000182 1.723 25 759812.301 4144764.797 143.989 0.0003 1.361 26 759813.364 4144764.652 143.956 0.000141 1.112 27 759814.615 4144765.945 143.879 0.000696 0.913 28 759815.85 4144766.052 143.834 0.000644 0.741 29 759816.643 4144766.415 143.813 0.001049 0.641 30 759817.357 4144767.444 143.782 0.000652 2.298 31 759818.407 4144767.728 143.764 0.000139 2.101 32 759819.555 4144768.04 143.721 0.000348 1.155 33 759820.712 4144769.234 143.675 0.000513 0.847 34 759822.567 4144769.791 143.702 0.001704 2.482 35 759823.21 4144770.785 143.685 0.001602 0.82 36 759824.994 4144771.217 143.642 0.001552 0.544 37 759826.518 4144772.559 143.575 0.001123 1.188 38 759828.278 4144772.762 143.542 0.001036 2.534 39 759828.96 4144773.607 143.584 0.000873 2.04 40 759830.083 4144773.803 143.549 0.001056 1.314 41 759831.113 4144775.034 143.592 0.001982 1.782 42 759832.962 4144775.255 143.54 0.000977 1.742 43 759834.459 4144776.716 143.562 0.001293 0.168 44 759835.231 4144776.561 143.472 0.001033 0.392 101 759808.497 4144761.513 144.165 0.000253 2.788 112 759808.053 4144763.864 145.631 0.001129 1.328 113 759809.38 4144761.257 145.637 0.000853 1.707 Fig. 10. Location of the GCP over the scattered point clouds. R. Martínez-´ Alvarez et al. Digital Applications in Archaeology and Cultural Heritage 37 (2025) e00417 9