**3. Data Acquisition**

Since 2017, a monitoring plan to the rock-wall of the lakeside reservoir has been commissioned by the Infrastructures Area of the Diputació de València. This change detection operation is performed yearly. After a first geodetic survey to detect fixed targets and select some check points, the data acquisition was carried out with a reflex camera for a CRP from different points of view to cover the whole area of interest. In 2019, in addition to the two techniques mentioned, a third survey was added: the MMS was tested to make this type of survey more complete, and verified through a comparison operation between point clouds. The combination of the two surveys had the objective of completing the 3D mapping of the site, needed to support the monitoring plan of the rock-wall.

#### *3.1. Cortes de Pallás Test Site*

Cortes de Pallás is a Cretaceous limestone area with historical geotechnical issues. On 6 April 2015, a cliff called La Muela partially collapsed, and some facilities of the electricity power plant and the main access road to the village were seriously damaged. At the end of 2017, once the consolidation works were finished, the Infrastructures Area of the Diputació de València commissioned a deformation monitoring plan to detect possible displacements of huge boulders or potential malfunction of the installed anchoring systems [2]. However, the detection of possible displacements of some centimeters with the required level of significance in a short period, e.g., two or three years, is a quite challenging task that can only be approached by means of high-precision geodetic techniques due to the peculiar topography of the zone. The whole area involves distances from 500 to 2000 m with height differences reaching 500 m. Moreover, 10 geodetic pillars mounted on presumably stable locations at 15 target points (referred to as check points) were installed in the rock wall by professional climbers using abseiling techniques, because they are not directly accessible. Furthermore, the measurements have to be undertaken by necessity from the opposite shoreline which is about 600 m away because the cliff of interest is facing a water reservoir (Figure 1).

**Figure 1.** Perspective view of Cortes de Pallás site area.

Therefore, the Diputació de València opted for periodical geodetic surveys along with image-based techniques like CRP and long-range TLS (since 3D models derived from CRP and TLS techniques proved compatible, only the former will be use for the experiment) as the most feasible join solution for the deformation monitoring plan. The periodical geodetic surveys, which are based on sub-millimetric EDM techniques and performed annually, have a triple objective: (i) to establish and monitor a high-precision 3D reference frame realized by 10 pillars, coded 8000+ (Figure 2a), (ii) to determine the 3D coordinates of 15 additional check points, represented by 360◦ prisms coded 1000+ (Figures 2b and 3) placed in specific points of interest on the monitored rock wall, and (iii) to provide a ground control network for image-based techniques.

As a demonstration of the potentiality of the high-precision geodetic techniques used [23,24], the displacements found for the reference frame pillars between the years 2018 and 2019 are shown in Table 1. Except for pillar number 8005 (Figure 2a), where a significative vertical displacement of −6.12 mm (see Table 1) was detected, the reference frame can be considered stable and well-controlled. The resulting reference frame has an overall accuracy of 1 mm and its scale is metrologically consistent with the unit of length of the International System (SI). Further information about the geodetic surveys is given in Section 3.2.

This high-precision geodetic method, which is very demanding and time-consuming, can be only applied to a limited number of relevant points of interest, which are 15 targets in the case at hand.

Complementary geomatic techniques like long-range TLS or CRP are needed to massively collect information with a density of around 300 points/m2. However, the accuracy of this type of points is reduced up to 1 to 3 cm. The main problem with both geomatic techniques is that photographs and cloud points provided by respectively CRP and TLS need to be registered in the same reference frame. In the case at hand, the reference frame as well as the coordinates of the 15 check points of interest, which are periodically provided by the sub-millimetric geodetic techniques, are considered the ground truth for the integration of the three techniques so that the combined numeric models are fully consistent.

(**a**) (**b**)

**Figure 2.** Example of target devices used on pillars of the reference frame and check points used for monitoring the site: (**a**) 50 cm white spherical point on the pillar (number 8005); (**b**) check point (in the center of the 360◦ prism) with a 15 cm white sphere on top. Please note both images are not to the same scale.

**Figure3.**Exampleofcheckpoint(centerofthe 360◦ prism)onsite:(**a**)number1009;(**b**)number

(**a**) (**b**)

 1010.


**Table 1.** Displacement of the reference frame pillars obtained by using sub-millimetric electronic distance meter (EDM) techniques between years 2018 and 2019. Points with statistical T above the cut-off value 5.739 are considered to have a significative displacement with a 99% probability. Pillar 8006 was not measured in the 2018 campaign.

An additional problem with both CRP and TLS techniques, especially when they are performed statically, like in the case at Cortes de Pallás, is the presence of occlusions. Since the CRP images can easily miss some shadowed areas, the 3D model obtained in each campaign usually has small patches with no information. A possible and efficient solution could be the use of mobile mapping solutions like the Kaarta Stencil 2 (https://www.kaarta.com/products/stencil-2-for-rapid-long-range-mobile-mapping/). Being dynamic, this system can provide a continuous 3D survey of the problematic area in Cortes de Pallás in less than one hour.

However, prior to being accepted as part of the combined 3D numeric models, the consistency of the points cloud obtained with the Kaarta system has to prove that it is consistent with the data provided by the EDM and CRP techniques. Since EDM and CRP are compatible at the 1–3 cm level (see Table 2), the point cloud obtained with the Kaarta system can be compared with the CRP solution in two ways to facilitate the analysis: first, using the dense CRP point cloud; second, employing manually measured natural photogrammetric check points.

**Table 2.** Differences between the close-range photogrammetry (CRP) and the EDM coordinates obtained for the check points.

