Laser scanner and UAV digital photogrammetry as support tools for cosmic-ray muon radiography applications: an archaeological case study from Italy

Abstract

The use of light detection and ranging technologies, i.e. terrestrial laser scanner (TLS), airborne laser scanner (ALS) and mobile laser scanner (MLS), together with the unmanned aerial vehicles digital photogrammetry (UAV-DP) and satellite data are proving to be fundamental tools to carry out reliable muographic measurement campaigns. The main purpose of this paper is to propose a workflow to correctly plan and exploit these types of data for muon radiography aims. To this end, a real case study is presented: searching for hidden tombs in the Etruscan necropolis of Palazzone (Umbria, Italy). A high-resolution digital elevation model (DEM) and three-dimensional models of the ground surface/sub-surface of the study area were created by merging data obtained using different survey methods to achieve the most accurate three-dimensional environment. Indeed, the simulated muon flux transmission used to infer relative transmission values, and the estimated density distribution, depends on the reliability of the three-dimensional reconstructed ground surface model. The aim of this study is to provide knowledge on the use of TLS and UAV-DP data and GPS-acquired points within the transmission-based muography process and how these data could improve or worsen the muon imaging results. Moreover, this study confirmed that muography applications require a multidisciplinary approach.

Figure 1

Schematic workflow followed to carry out the presented archaeological application of muography. Five main phases were defined: (1) preliminary assessment; (2) muographic campaign; (3) data processing; (4) data visualization; (5) data interpretation and results. Remote sensing methods and outputs appear throughout the workflow from phases 1 to 5.

Figure 2

Geographic location and photos of the study area: (a) Italy and Umbria region; (b) in white, the perimeter of Perugia and, on the eastern side of the city, the area involved in this study is highlighted by the orange rectangle; (c) aerial view of the Palazzone necropolis, the white polygon delimits the touristic sector of the necropolis and the green one the studied area. The information reported in (b,c) are draped on a satellite base map uploaded on the free and open source QGIS software. At the bottom of the figure, some explicative photos show some of the many underground spaces accessible at the necropolis (the position of the numbered yellow circles in (c). These images show the appearance of the local archaeological heritage represented by shallow-to-depth underground tombs (1, 2, 4) and the warehouse of the touristic site (3).

Figure 3

Three-dimensional processed models of the study area based on RS-acquired data. Four datasets were combined to create the most reliable and accurate three-dimensional environment to be used for muon absorption simulations within the target hill of the necropolis: ministerial ALS of 10 × 10 m² and 1 × 1 m² resolution, TLS of 0.02 × 0.02 m² and UAV-DP of 0.2 × 0.2 m². In (a), the mesh of the whole involved area obtained from the ALS survey, with a resolution of 10 × 10 m², is shown; in (b), the same mesh of (a) was merged with the 1 × 1 m² resolution on the Palazzone necropolis area; (c) zoom on the eastern side of the necropolis (the studied target, see Fig. 2); (d) ministerial data (in blue), TLS data (in red) and UAV-DP data (in green) merged together to obtain the most reliable three-dimensional reconstruction of the studied environment, comprising the most precise components from the processed point clouds. A visual comparison between (c) and (d) already offers insight into the extent of differences in the ground surface reconstruction within the two three-dimensional models.

Figure 4

Analysis of the point cloud section draped on the UAV-DP model: (a) view of the analyzed data section passing through the center of the MIMA detector and “cutting” the target hill. The section consists of all the available datasets, visualized on the RGB UAV-DP point cloud; in (b) the same data of (a) are shown but with the altimetric values visualized as a color scale.

Figure 5

Point cloud data comparison along the section shown in Fig. 4: (a) section of the 10 × 10 m² point cloud from ministerial ALS; (b) 1 × 1 m² point cloud from ministerial ALS; (c) 0.2 × 0.2 m² point cloud from the UAV-DP survey, the segmentation in ground and vegetation points shows many missing data zones under the vegetation; (d) 0.02 × 0.02 m² point cloud from the TLS survey, few ground points are visible under the vegetation and a section of a known tomb chamber is present within the studied hill; (e) section of the point cloud obtained by merging the ALS,TLS and UAV-DP datasets cleaned by vegetation; (f) all the available datasets together with the MIMA detector location, the yellow arrows highlight the effective and lost acceptance angles.

Figure 6

Elevation differences along the z-axis draped on the best three-dimensional model of the studied area: (a) Cloud-to-Mesh (C2M) signed distance calculated between the 1 × 1 m² ministerial ALS mesh and the point cloud obtained by merging UAV-DP and TLS point clouds. Underground tomb #11 is shown, and the white dotted lines represent a two-dimensional view of the MIMA detector acceptance cone. In (b), the same orthogonal view of (a) without transparency, arrows, lettering and, in (c), the using a different view angle.

Figure 7

Three-dimensional analysis of the Colombario area: in (a,b), the RGB image and the UAV-DP obtained point cloud of the scene, relatively. (c–f) Photos of the GPS in situ survey to obtain the ground truth. In (g,h) the C2C absolute distance values computed on CC between the GPS-acquired points and the ministerial 1 × 1 m² ALS and UAV-DP datasets, respectively.

Figure 8

Three-dimensional analysis of the hill without the Colombario area: in (a,b) photos of the dense vegetation that characterizes the ground hill. In (c) the TLS obtained point cloud of the scene. (d–f) Photos of the GPS in situ survey to obtain the ground truth. In (g,h) the C2C absolute distance values computed on CC between the GPS-acquired points and the ministerial 1 × 1 m² ALS and TLS datasets, respectively.

Figure 9

Depth maps differences calculated along the LoS of the MIMA detector that was placed in the underground warehouse of the necropolis: (a) depth map obtained by comparing the 1 × 1 m² with the 10 × 10 m ALS-obtained ministerial surfaces, on the X–Y axis the degree from the center of the tracker, the color scale shows the depth differences in meters, and the dotted white lines highlight the perimeter of the studied hill and that of the higher mountains behind. The differences between these two datasets can reach up to ± 20 m; (b) zoom of (a) showing in blue the perimeter of the known tomb #11 located inside the hill surveyed during the TLS campaign, (c) depth map obtained by comparing the 1 × 1 m² ministerial surface with the ground surface obtained merging all the available data (UAV-DP, TLS and the 1 × 1 m² ministerial; see Fig. 3d). In (d) the zoom of (c) allows to understand the surficial differences among the compared datasets. In (e) a three-dimensional view of the depth map is shown in (c) to facilitate the understanding of this kind of map.

Figure 10

The TLS survey campaign was carried out using a RIEGL VZ-1000: (a) aerial photo of the study area taken from the UAV survey, in which every underground/surficial scan point and target is highlighted and some photos relative to some of these points; photos 1 and 5 show the TLS measuring in two different locations, photos 2 and 4 show the underground measurement point in the warehouse and inside a known ruined tomb, photo 3 shows one of the thirty used reflectors. In (b) the TLS-obtained point cloud (more than 150 million points) with elevation values. The same point cloud is shown in (c) with the RGB values inferred from the UAV-DP data.

Figure 11

Point cloud data from the UAV-DP survey: (a) aerial photo of the study area taken from the UAV survey, ten surface markers were used to correctly georeference the created three-dimensional model; (b) the Employed UAV, a DJI Mavic Air 2 (see Table 1 for technical information); (c) the RGB dense point cloud of the study area, about 55 million of points; (d) visualization of the classified UAV-DP-obtained point cloud, two classes were identified: ground and non-ground points (vegetation and manufacts).

Figure 12

In (a), the MIMA detector measuring the muon flux underground in a side corridor of the warehouse of the Palazzone necropolis. In (b), zoom on the upper clinometer shows a pointing direction of (20 ± 1)° with respect to the horizon. Some technical information about the MIMA detector are reported in Table 1.