RINA - Démonstrateur d’utilisation de l’intelligence artificielle pour la gestion opérationnelle des risques naturels gravitaires géologiques | FEREC 2020

The PCA Graph represents how much of each variable is explained by each component. Figure 10 (c) and (d) shows how PC1 and PC2 are formed. We observe that all 5 features seem spread on the 4 corners. No two features are clustered together. Plotting the data on the two principal components with the events being colored it is possible to look for clustering of events and try to identify any clustering. Figure 10 (e) and (f) shows a concentration of points close to the origin. This makes sense as the majority of the original data is also clustered as the majority of the days there is no rain. The main objective of this exercise was to try to identify a visual separation between the days with rock falls and the others. Neither of the graphs provides a clear separation that could be exploited. 3.4. Spatial data Having the topography, the geology and the information about the protective elements installed along the cliff values were assigned, along the road, to represent the geomorphological characteristics of the cliff. This spatial study aims to improve the understanding of what characteristics of the cliff are linked to the blocks reaching the road. The protective elements were simply described as present or not present for each point of the road. Various protective structures are used for different parts of the road and they can also have different dimensions depending on the demand, but those were not considered. The final separation before and after the constructions are shown in the figure 11. Presence of protective elements before constructions of 2008 Presence of protective elements after constructions of 2008 Figure 11: protection zone (0 no protection/ 1 with protection) 19