PSI - Issue 44

Brunella Cutrone et al. / Procedia Structural Integrity 44 (2023) 713–720 Brunella Cutrone et al. / Structural Integrity Procedia 00 (2022) 000–000

716

4

3. Critical analysis for the determination of the Landslide Attention Class: Level 0 compilation of the census forms The census of bridges, planned at Level 0, consists in cataloging all the bridges present in the area in order to know the number of structures to be managed and their characteristics, both in relation to the geometry and structural elements, and relatively to the road network in which they are inserted and to the location site. This phase represents the starting point for the implementation of the multilevel approach; the information collected in the census makes it possible to identify an order of priority useful for planning the visual inspections in situ. With regard to the geomorphological classification of the area, from the census form (Annex A of the Bridge Guidelines) it is possible to notice that the suggested reference documentation is the PAI plans (Hydrogeological Arrangement Plans). The PAI plans, drawn up pursuant to Italian Law 183, 1989, Legislative Decree 180, 1998 and Legislative Decree 152, 2006, through the application of restrictions and regulations for the use of the territory, certainly constitute a fundamental tool for correct territorial planning, however they map the risk associated with the calamitous event of hydrogeological nature without providing information on its characteristics and, above all, by making an estimate of the expected damage linked to its occurrence (ISPRA, 2021, ISPRA 2018). ISPRA, in order to update the landslide hazard map on the entire national territory, proceeded to create the new national mosaic (v. 4.0 - 2020-2021) of the hazardous areas of the PAI Plans. From the analysis, significant inhomogeneities of representation and classification emerge, mainly due to the different methodologies used. The greatest differences are found between the hazard maps that classified only landslide polygons (e.g. Upper Adriatic catchment areas, Autonomous Province of Bolzano) and those that classified the entire territory (e.g. Valle d'Aosta Region, Autonomous Province of Trento, Arno Basin). In fact, by analyzing the distribution of landslides in the IFFI Inventory, a decidedly higher homogeneity is found. The presence of non-homogeneity in the perimeters had already been highlighted in the 2015 and 2018 ISPRA reports (ISPRA, 2015, ISPRA 2018) on the basis of planning data available at the time. In addition, the areas reported in the PAI are often summarily traced without taking into account any counter-slopes, valleys engraved by water courses and of the same shape which, in theoretical conditions is determined by the type of movement (Manzo et al., 2013). This also has consequences in terms of the calculation of the landslide density and therefore in the construction of future susceptibility models. As a first proposal for improvement, it seems appropriate to integrate the consultation of the hazardous areas of the PAI plans with the consultation of the landslides mapped in the IFFI inventory, given the greater precision and conceptual coherence with respect to the perimeter areas in the PAI plans. In addition to the consultation of the PAI plans cartography and the IFFI maps, it is also considered appropriate to adopt methodologies aimed at estimating the landslide susceptibility of the area, in order to circumscribe areas potentially affected by instability by planning in a more complete way the subsequent on-site inspection phase envisaged in Level 1 of Bridge Guidelines. The methods applied for the analysis of landslide susceptibility are generally divided into qualitative and quantitative approaches (Romeo, Mari, Floris, Pappafico, Gori, 2011). The former are generally based on traditional techniques, such as field survey, geomorphological analysis and aerial photo analysis, also supported by GIS techniques to produce thematic maps. The latter can be divided into two categories: statistical methods and deterministic methods. Statistical methods are based on the principle that areas where landslides occurred in the past will most likely be affected by landslides also in the future and that areas characterized by predisposing conditions similar to those already affected by landslides have a greater propensity to trigger landslides. The advantages of these methods are the evaluation of landslide susceptibility in an objective way; the limit is that the goodness of the results is strongly conditioned by the accuracy of the input data. Deterministic methods are based on the use of physical models; analytical models that allow the calculation of the stability of a slope by analyzing the forces involved and the geological-technical characteristics of the soils (cohesion, internal friction angle, interstitial pressure, etc.). The result is the determination of a safety factor which can also be used in the design phases of infrastructures or containment works. The different methods can be chosen and used based on the scale of the study, the availability of data and the purpose of the analysis. In contexts in which certain information on recognized kinematics is not available or in which complete and reliable databases on the triggering and predisposing factors of failure mechanisms are lacking, the methods that lead to a reliable, albeit subjective, estimate are empirical methods. As a second improvement proposal, it is suggested the adoption of a methodology which, if on the one hand requires the operator's expertise, on the other hand allows a more objective assessment of the failure phenomenon. In particular, for the estimation of landslide susceptibility we propose the application of a bivariate heuristic-statistical method