Bausilio, Giuseppe (2024) Urban Geology and Geohazards: an integrated analysis for a resilient urban planning. [Tesi di dottorato]

Anteprima

Testo Bausilio_Giuseppe_36.pdf Download (22MB) | Anteprima

Abstract

This PhD project's objectives were the characterization and modelling of different geohazards, their effect on the urban fabric, and the study of the impact that a single geohazard can exert on other geohazards through a multirisk approach. The entire process was divided into three phases. During the first one, the preprocessing phase, the Geohazards (sinkhole, landslide and flood) inventories, Predisposing factors and Elements at Risk map were prepared. In particular, sinkhole and landslide inventories for the city of Naples underwent an updating phase starting from pre-existing data. The sinkhole inventory update was performed starting from the Guarino and Nisio (2012) published inventory and using national and local news, city council reports, and field work. The landslide inventory used in this project was extracted from the Landslide Inventory of the Campania region (hereafter LaICa), published by Fusco et al. in 2023. LaICa was obtained by gathering and uniforming the inventories of the former Campania Units of Management, the IFFI project, and literature data. The research group of the Department of Earth, Environmental and Resources Science of the University of Naples Federico II contributed and provided support during the update phase. As for the flood inventory, a sampling operation was carried out, extracting random points from the flood hazard polygons of the former Unit of Management "Campania Centrale". This process was necessary due to the lack of an available flood inventory for the city of Naples. The last step in the preprocessing phase was the production of the Damage map used for the Risk Assessment. The first element of Damage, Elements at Risk, was evaluated using the following layers as base data: • Population density (spatially distributed according to the Census Sections and divided into five classes from Very Low to Very High); • Rail transport (considered as a strategic infrastructure); • Road system (divided into non-strategic and strategic networks, the latter mainly consisting of motorways and main arteries); • Natural Reserves (Parco Regionale dei Campi Flegrei and Parco Regionale Metropolitano delle Colline di Napoli); • Buildings divided into strategic structures (hospitals, barracks, schools, railway stations, administrative and government buildings, etc.) and non-strategic ones. Due to the difficulty in defining the Vulnerability for such a heterogeneous and complex territory as the City of Naples, a precautionary measure has been adopted by attributing the maximum Vulnerability value (1) to the entire area. The objective of the second phase of the approach was the production of the Landslide, Sinkhole, and Flood Risk maps, combining Damage and Relative Hazard using Risk Matrix. The Relative Hazard assessment was carried out using an Ensemble Modeling approach, deploying two Machine Learning methods, the Maximum Entropy (MaxEnt) and the Random Forest (RF) algorithms. Two approaches were used to increase the performance obtained from the geohazard susceptibility assessment: • the K-Fold Crossvalidation approach was used on the presence data, allowing the iteration of the modelization process and obtaining a mean value of the performance score; • The Variance Inflation Factor (VIF) was deployed to detect any collinearity problems between the predisposing factors. The Susceptibility map obtained using the Ensemble Modeling approach reached excellent ROC/AUC performance scores (Flood & Hydraulic Erosion, Transport and Deposition susceptibility 0.94; Landslide susceptibility 0.94; Sinkhole susceptibility 0.89). These performance were evaluated using, as test data, 30% of the original inventories that have been kept outside of the susceptibility assessment process, which was carried out on the remaining 70% of the original data. Due to the incomplete temporal information in two of the three inventories used, it was not possible to perform a return period analysis. For this reason, the Susceptibility (Relative Hazard) was used to evaluate the Risk. Damage and Relative Hazard were combined, using a Risk matrix, to obtain the Risk maps. During the third and last step, the mutual influence between geohazards was evaluated through a multi-hazard approach using the Rock Engineering Systems (Hudson, 1992) method. The Interaction Matrix was compiled using the Relative Hazard maps obtained in the previous phase as parameters of the Multirisk system. Through the Matrix, two scores are attributed by the user to every single parameter (the geohazard): (i) a cause score, which is the impact of the parameter over the other parameters and (ii) the effect score, the impact of the other geohazards against the parameter in analysis. The score ranges from 0 (no influence) to 3 (maximum influence). In the following step, all the sums of cause and effect scores for every parameter are evaluated (C+E). The percentage value of C+E for a geohazard against the total sum of the C+E of all the parameters is the interactivity [%]. Three interactivity maps were produced, one for every geohazard, by attributing the interactivity percentage value to the areas where the geohazard was present. The final Multirisk map was obtained by combining the Interactivity Map and the Damage Map using a risk matrix. From the analysis of the final products, the weight of the flood geohazard on the other parameters is evident. Out of the three parameters, the flood-related one is the only dominant one (Causes score > Effects score), while the remaining two are subordinate parameters (Causes score < Effects score). The historic city centre falls within the Medium Multirisk class area despite the lower local interactivity when compared to other places. This is due to the influence of the Damage map, as a high density of strategic structures and infrastructures is located within the city centre. The Natural Reserves' influence is also evident along the western and northern slopes, where the High and Very High Multirisk classes are abundant. As for the eastern sector of the city, the strategic infrastructures (main roads, highways and railways) are heavily influential, as the High and Very High classes are located along these elements. The results obtained are adequate to expectations and, in the future, the integration of further geohazards which have a recognized influence on other parameters (fires-landslides, for example) would allow a more detailed definition of the Multirisk system in an urban context.

Downloads

Actions (login required)

Modifica documento