Methodological considerations for the study of rockfall risk along 3.5 km of coast using lidar data
Considérations méthodologiques pour l’étude de l’aléa éboulement d’une falaise côtière à partir de données lidar
Résumé
The erosion of cliffs bordering the Mediterranean sea in the Provence-Alpes-Côte-d'Azur region (PACA) was the subject of a recent thesis (Giuliano, 2015). Results show a low annual average erosion rate (mm to cm / year) depending on the location. In the town of Carry-Le-Rouet, coastal cliffs were identified as particularly at risk (Marçot, 2006), not so much because of a high rockfall hazard, but rather because of dense urbanization on top of the cliff. As a result, the topographic monitoring of 3.5 km of coastline with a mobile lidar installed on a boat was started as early as February 2011, repeated in November / December 2011 and then in July 2012. These lidar surveys have some properties: a partially covered cliff with vegetation, a sinuous coastline with capes and bays and a poor performing technology during the first survey. These elements contribute to making the treatment of point clouds and their interpretation more complex than in other cases discussed in the literature (e.g., Rosser et al., 2007 or Dewez et al., 2013). Giuliano (2015) established the basis of the treatment, but certain technical biases did not allow us to exploit his catalog of erosion scars in a probabilistic analysis of rockfall occurences. A new methodology has therefore been developed to achieve this objective. This methodology takes into account:
•inaccuracies due to the acquisition of lidar data from a moving boat during the first campaign and co-registration errors;
•the vegetation partly covering the cliff, mapped on an orthophoto acquired during the second campaign of November 2011;
•the coast geometry with capes and bays, which requires the unrolling/rolling of the point clouds in order to mark the vegetation points and to confirm that the differences observed between clouds are indeed related to rock collapses;
•a different point density for each lidar acquisition.
The different point clouds are unrolled according to a curve describing the geometry of the coastline, and composed of rectilinear or arcuate portions (Guliano, 2016). The main urban works and the main vegetation areas are extracted from the point clouds using information from orthophotographs of the coast that were acquired at the same time as lidar data.
The diachronic comparison of point clouds is no longer carried out by rasterizing the point clouds on 2.5D grids. Instead, we directly compute a cloud to cloud distance using the M3C2 algorithm (Multiscale Model to Model Cloud Comparison; Lague et al., 2013). This M3C2 algorithm has many advantages over grid differences. Differences M3C2 are calculated by exploiting all observation points, rather than summarizing them with a single cell value in a grid. Moreover, each difference is qualified statistically and takes into account, by construction of the algorithm, the intrinsic roughness of the compared surface portions. We document how we selected the parameters chosen for the M3C2 calculation. These parameters are adjusted in relation to the observed density of points.
Once applied, suspicious differences remain, statistically significant, but positioned in residual areas of vegetation that did not exist on the reference orthophoto or related to urban developments. To exclude them from the erosion database, we use statistical indicators of geometry similar to those of the CANUPO method of Brodu and Lague (2012) to distinguish false positives (erosion not related to a rock start).
References :
Brodu N. and Lague D. (2012) 3D terrestrial lidar data classification of complex natural scenes using a multi-scale dimensionality criterion: Applications in geomorphology. ISPRS J. Photogramm. and Rem. Sens., Vol. 68, pp 121–134
Dewez, T.J.B., Rohmer, J., Regard, V., and Cnudde, C. (2013) Probabilistic coastal cliff collapse hazard from repeated terrestrial laser surveys: case study from Mesnil Val (Normandy, northern France), J. Coast. Res., Sp. Iss. 65, DOI:10.2112/SI65-119.
Giuliano J. (2015) Erosion des falaises de la région Provence-Alpes-Côte d’Azur : évolution et origine de la morphologie côtière en Méditerranée : télédétection, géochronologie, géomorphologie. Sciences de la Terre, PhD, Université Nice Sophia Antipolis
Lague D., Brodu N., and Leroux J. (2013) Accurate 3D comparison of complex topography with terrestrial laser scanner : application to the Rangitikei canyon (N-Z), ISPRS Journal of Photogrammetry and Remote Sensing, Vol. 82, pp 10–26, https://doi.org/10.1016/j.isprsjprs.2013.04.009
Marçot N. (2006) – Prise en compte de la problématique des risques liés aux falaises côtières
dans l’aménagement du territoire en Provence-Alpes-Côte d’Azur – Année 2 : Définition des
enjeux sur le linéaire de falaises côtières, caractérisation et hiérarchisation des risques. Rapport
BRGM RP-54316-FR. 72 p. 27 ill. 1 ann. 12 cartes hors texte
Rosser, N.J., Lim, M., Petley, D.N., Dunning, S.A. & Allison, R. (2007) Patterns of precursory rockfall prior to slope failure. Journal of Geophysical Research (Earth Surface). 2007;112:F04014.
Acknowledgements:
Data were acquired thanks to the French national project VALSE. Financial support was also given by the following French intitutions: DREAL PACA and Conseil Régional PACA.