**6. Discussion**

The experiments presented above involved point clouds produced with CRP and MMS solutions. The experiment was designed to investigate whether MLS can reach a degree of accuracy to complement CRP for mapping monitored areas; it was motivated for understanding the potential of an innovative acquisition method using MMS in challenging environments.

The MMS proved to be very useful due to the rapidity of acquisition and the ease of use. In the presence of an existing reliable photogrammetric survey, the mobile mapping can be easily constrained, reducing the post-processing actions required; the developed methodology made the combination with existing surveys very easy for expert operators. Kaarta Stencil 2 has a strong potentiality to generate a higher number of points when composing the 3D point clouds. The advantage of using the MMS at the ground level allows users to enrich the point cloud in areas di fficult to acquire with the CRP from the other side, for example, the roads surface and objects hidden or covered by the crown of trees. It is clear that these geomatics technologies may be complementary to one another in creating complete high-quality fully 3D representations. Nevertheless, the noisy data invalidate their usefulness to create high accuracy 3D models of the area, complementary to the photogrammetric one undertaken in the previous campaigns.

This work demonstrates its usefulness for data acquisition and mapping purposes, presenting similarities with the case studies described in [3,13], referring to the hand-held use of MMS. Indeed, we have been able to manage the occlusions, which are unavoidable from the photogrammetric model. Therefore, the information can be considered complementary to CRP for mapping the entire area, i.e., for a comprehensive survey. However, we have corroborated that the MMS is not robust enough to be used for monitoring complex sites such as this Corte de Pallás site. One promising approach is to integrate an accurate di fferential GNSS to the MMS solution.

It is fair, however, to highlight some drawbacks that emerged from this research. First of all, the point cloud cannot be exploited from scratch; in fact, further work of alignment was undertaken (as described in Section 4). The more the traveled survey distance extends, the more the drift error increases. The use of some constraints along the path should reduce the drift error. Despite MMS being designed with the main purpose of collecting 3D data without the need to register the point cloud further, this work proved that, the achievement of a result exploitable for mapping purposes relies upon the co-registration with an existing, and more accurate, point cloud. As the survey needs to be integrated with other data, accurate planning of the survey is required; in other words, it is not enough to just walk (see loop closure issue in Section 3.4) in order to improve the accuracy and facilitate the integration. Monitoring data need a submillimeter accuracy; as we reached an accuracy that is not comparable to the photogrammetric one (5.6 ± 2.3 cm), hand-held MMS as it is now implemented (without an accurate implementation of GNNS) cannot be considered suitable for such monitoring purposes. Nevertheless, it is very useful to complete the mapping of challenging areas in a fast, agile, and a ffordable way.

Another aspect that is worth discussing is the lack of RGB information. Indeed, the Kaarta MMS provides output 3D point clouds without the color information. This is an aspect that cannot be neglected, as environmental monitoring applications require in depth knowledge of a site. To overcome this limitation, the MMS cloud has been colorized, assigning the RGB values of corresponding neighbors from the CRP cloud. The colorized MMS cloud is depicted in Figure 16.

A final note is required with regard to the number of points that can be exploited with respect to the original raw data. As visible in Figure 13, to achieve a satisfactory accuracy comparable with the CRP one, the filtering steps brought to reduce the threshold up to values lower than 10 cm. This step increased the accuracy of the remaining points, at the expense of the number that was significantly reduced. In other words, due to the noise that is introduced by the tool, there is a waste of 3D information.

**Figure 16.** Colorized MMS (Mobile Mapping System) point cloud: (**a**) overview of the entire survey, (**b**) close-up view of a portion of the surveyed area.
