A new perspective for regional landslide susceptibility assessment





flow-like landslides, landslide inventory, susceptibility mapping, Generalized Additive Model


Landslides pose a severe geohazard in many countries. The availability of inventories depicting the spatial and temporal distribution of landslides is crucial for assessing landslide susceptibility and risk in territorial planning or investigating landscape evolution. In the case of the Italian territory, several landslide hazard and risk maps were produced ranging from regional to national scale. This was made possible leveraging public domain data of the Italian Landslide Inventory (IFFI project; Trigila et alii, 2010), or other geodatabases spanning from local to regional scale. However, the practical utility of this inventory is often limited in many applications due to its spatial inhomogeneity or the use of different mapping methods and classification criteria. Despite the impressive advancements in techniques for assessing natural hazard susceptibility at a national scale over the past years, including statistical models, AI based models (i.e. Neural Networks) and others, the results are still limited by the quality of the data used. Specifically, the effectiveness of these models is closely tied to the quality of the landslide inventory utilized. Currently, recent regional landslide inventories could potentially enhance precision and accuracy compared to the national dataset, primarily owing to their finer resolution compared to the IFFI dataset. In this work, we present a new approach to assess landslide susceptibility at local scale, relying on regional landslide inventories. Using a data-driven technique, we propose to train a single model on a landslide inventory consisting of a composition of regional inventories selected to be representative of the national scenario. The weighted model is now capable of predicting landslide susceptibility in any study area across Italy. The entire analysis has been done using the SRT tool for Google Earth Engine and the SZ-plugin for QGIS. All the data used and processed are freely available and downloadable. The proposed approach has been tested in the framework of the PNRR RETURN project. The evaluation was conducted in two specific areas: the first one encompasses a section of the railway connecting Napoli to Bari (southern Italy), while the second focuses on areas impacted by the Marche region 2022 landslide event (central Italy).




How to Cite

Titti, G., Antelmi, M., Fusco, F., Longoni, L., & Borgatti, L. (2024). A new perspective for regional landslide susceptibility assessment. Italian Journal of Engineering Geology and Environment, 275–283. https://doi.org/10.4408/IJEGE.2024-01.S-30

Most read articles by the same author(s)

1 2 > >>