Combining data-driven models to assess susceptibility of shallow slides failure and run-out

• Autor(a) principal: Melo, Raquel
• Data de Publicação: 2019
• Outros Autores: Zêzere, José, Rocha, Jorge, Oliveira, Sérgio
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10451/41272
• topic: Shallow slides; Susceptibility to failure; Logistic regression; Run-out modeling; Cellular automata

This research is focused on the susceptibility assessment of shallow slides by modeling the failure and run-out areas separately. The shallow slides failure is evaluated using a statistical method (logistic regression) and for the run-out assessment, a simple cellular automata model is proposed. The existence of shallow slides inventories occurred in distinct time periods allowed the separation of data into two independent groups (modeling and validation) and the adoption of the temporal criterion for the independent validation. The logistic regression model showed a very good predictive capacity (area under the receiver operating characteristic curve of 0.90), although it may be overestimated, as well as the susceptibility scores obtained. The run-out modeling, using a simple cellular automata model developed for this study, provided good results, with an overlap between the simulation and the real cases of 77%. Lastly, a final shallow slide susceptibility map was constructed including both failure and run-out areas. This work accomplished a combination of low-cost methodology with limited input data that allowed a good performance of the landslide susceptibility assessment and can be easily applied to other regions.