Limiares de desencadeamento de deslizamentos em áreas de risco do município de Petrópolis, RJ

Abstract

This report summarizes the main steps and results of the study focused on the spatiotemporal analysis of landslide susceptibility in Petrópolis (RJ), based on event records between 1985 and 2022. Historical occurrence data and environmental and anthropogenic conditioning variables were integrated to produce susceptibility maps in years with a higher concentration of landslides, allowing the assessment of spatial variability and the localized effects of control factors. The methodological approach combined Logistic Regression (LR) and Multiscale Geographically Weighted Regression (MGWR), enabling both the global estimation of associations and the identification of local patterns of influence. Among the most relevant variables were: lithology, land use and land cover (LULC), total precipitation (PRCPTOT), extreme events (R100), distance to rivers, soil type, aspect, elevation, and forest cover. Lithology and LULC showed the strongest positive associations with the occurrence of landslides, while forest cover and elevation indicated a protective effect. Predictive performance was evaluated using robust metrics, with AUC ranging from 0.77 to 0.99 and accuracy between 0.77 and 0.97, demonstrating consistency across different time periods. The maps indicated a trend of increasing areas classified as "very high susceptibility" throughout the analyzed period, particularly in urbanized sectors, reinforcing the relevance of the interaction between land use dynamics and physical constraints. As a product, the report consolidates a technical-scientific framework applicable to supporting public policies, urban planning, and risk reduction strategies, by highlighting key factors and their spatial heterogeneity. Additionally, the data collected and the methodologies implemented constitute a reusable database capable of supporting other studies — including year-on-year comparisons, analyses with future climate scenarios, evaluation of mitigation measures, and replication in municipalities with similar geomorphological contexts.