A carregar...
Publicação
Modelling the rainfall threshold for shallow landslides considering the landslide predisposing factors in Portugal
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 4.53 MB | Adobe PDF |
Orientador(es)
Resumo(s)
Rainfall-triggered landslides pose a significant threat
to both infrastructure and human lives, making it crucial to comprehend the factors that contribute to their occurrence. Specifically, understanding the relationship between these factors and
the amount of rain that is necessary for triggering such events
is essential for effective prediction and mitigation strategies.
To address this issue, our study proposes a statistical modelling
approach using machine learning, specifically the Random Forest
algorithm, to investigate the connection between landslide predisposing factors and the daily rainfall intensity threshold necessary
for the initiation of shallow landslides in Portugal. By leveraging
a comprehensive dataset comprising historical landslide events,
associated critical rainfall, and ten distinct landslide predisposing factors, we developed several models and used cross-validation
technique to evaluate their performance. Our findings demonstrate
that the Random Forest model effectively captures a relationship
among landslide predisposing factors, critical daily rainfall intensity, and landslide occurrences. The models exhibit a satisfactory
accuracy in assessing the spatial variation of critical daily rainfall
intensity based on the predisposing factors, with a mean absolute
percentage error (MAPE) of around 17%. Furthermore, the models
provide valuable insights into the relative importance of various
predisposing factors in landslide triggering, highlighting the significance of each factor. It was found that it takes higher rainfall
intensity to trigger shallow landslides in the north region of Portugal when considering critical rainfall events of 3 and 13 days. Slope
aspect, slope angle, and clay content in the soil are among the main
predisposing factors used for defining the spatial variation of the
daily rainfall intensity threshold.
Descrição
Palavras-chave
Critical rainfall intensity Predisposing factors Shallow landslide Random Forest
Contexto Educativo
Citação
Villaça, C., Santos, P., & Zêzere, J. (2024). Modelling the rainfall threshold for shallow landslides considering the landslide predisposing factors in Portugal. Landslides, 21(9), 2119–2133. https://doi.org/10.1007/s10346-024-02284-y
Editora
Springer