Miraglia, Vittorio (2024) USE OF FUZZY-BASED AI TECHNIQUES FOR THE ANALYSIS OF CLIMATE IMPACTS AND MULTI-RISK ASSESSMENT OF URBAN SETTLEMENTS. [Tesi di dottorato]

Documento PDF Miraglia_Vittorio_37.pdf Download (6MB)

Abstract

Climate change is showing increasingly tangible effects, mainly through extreme weather events such as heatwaves and intense precipitation, which are occurring with increasing frequency. These events directly impact urban systems, affecting buildings, public spaces, and communities. Many studies have explored the repercussions of extreme weather events on urban areas, analyzing vulnerability and associated risk. The urgency of addressing such risks has led to the development of decision-making tools to assess the level of risk and identify the most vulnerable areas. Recent research has focused on the relationship between the built environment and climate phenomena, aimed at defining adaptation strategies to mitigate risks. Methodologies and tools have been developed to deepen the knowledge of climate vulnerability, with statistical approaches such as Principal Component Analysis (PCA) used to map vulnerability in urban contexts. However, the limitations of portability and replicability of existing models represent a significant challenge. To further address the limitations of traditional methods, Artificial Intelligence techniques and fuzzy approaches have been experimented, offering a new paradigm for managing uncertainty. Several experiments have been conducted evaluating the adoption of approaches and methodologies based on Artificial Intelligence (AI) in order to enrich the understanding of the urban system and manage the uncertainty associated with the results. These methods aim to overcome the limitations of traditional approaches, which often require interpretation by experts in the field before the data can be used by decision makers. In this work, some innovative AI techniques applied to the assessment of extreme climate events, in particular heatwaves, and to the assessment of multi-risk are presented. Emotion Detection techniques through the analysis of textual data allow us to acquire qualitative information on the perceptions and reactions of the population to extreme climate events. This approach can reveal how different communities perceive risk, allowing a deeper understanding of the social and psychological factors that influence urban resilience. For example, the analysis of social media posts can identify areas of greater concern and vulnerability, thus guiding targeted communication and intervention strategies. Image Segmentation algorithms applied to satellite image analysis are useful for identifying and classifying land features. This technique is essential for monitoring changes in urban landscapes following extreme events, such as floods or heatwaves. Segmentation allows distinguishing between residential, commercial, and green areas, facilitating a precise assessment of impacts on the different elements of the urban fabric. In addition, image analysis can provide updated data for risk modeling and intervention planning. With Hotspot Analysis it is possible to map the distribution of extreme events and overlay it with socioeconomic, environmental, and infrastructural variables. This analysis can highlight the critical areas most exposed to risk, allowing authorities to concentrate resources more efficiently and develop more effective adaptation strategies based on the specific needs of communities and the most vulnerable subjects. Fuzzy Multi-Criteria Decision Analysis (FMCDA) approaches integrate different decision criteria in a context of uncertainty. Through fuzzy logic, it is possible to represent and manage the uncertainty of information and assessments, which is particularly relevant when considering the impacts of extreme climate phenomena. This approach allows combining qualitative and quantitative assessments, facilitating more informed decisions. For example, in the selection of adaptation interventions, fuzzy MCDA allows to weigh different factors, such as costs, environmental and social benefits, based on expressed preferences, even in situations of uncertainty. These AI approaches aim to improve the understanding of urban systems and to support decision makers in increasing the resilience of the urban system against climate impacts. This research, in line with European initiatives, aims to improve data quality and the use of digital technologies in the assessment of climate risks. The integration of these AI techniques in the assessment of the impacts of extreme climate phenomena allows us to address the complexity and variability of urban systems in a more systematic and informed way. Thanks to the ability to analyze heterogeneous data and to manage uncertainty, these methodologies promote a more systematic and informed understanding of the impacts of extreme climate phenomena.

Downloads

Actions (login required)

Modifica documento