A carregar...
Publicação
Technical note: assessing predictive capacity and conditional independence of landslide predisposing factors for shallow landslide susceptibility models
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.57 MB | Adobe PDF |
Orientador(es)
Resumo(s)
The aim of this study is to identify the landslide
predisposing factors’ combination using a bivariate statistical model that best predicts landslide susceptibility. The best
model is one that has simultaneously good performance in
terms of suitability and predictive power and has been developed using variables that are conditionally independent. The
study area is the Santa Marta de Penaguiao council (70 km ˜
2
)
located in the Northern Portugal.
In order to identify the best combination of landslide predisposing factors, all possible combinations using up to seven
predisposing factors were performed, which resulted in 120
predictions that were assessed with a landside inventory containing 767 shallow translational slides. The best landslide
susceptibility model was selected according to the model degree of fitness and on the basis of a conditional independence
criterion. The best model was developed with only three
landslide predisposing factors (slope angle, inverse wetness
index, and land use) and was compared with a model developed using all seven landslide predisposing factors.
Results showed that it is possible to produce a reliable
landslide susceptibility model using fewer landslide predisposing factors, which contributes towards higher conditional
independence.
1 Introduction
Recent developments in GIS software and increasing computing power allow a substantially high number of independent variables to be used in empirical, data-driven landslide
susceptibility models. Recent studies in landslide susceptibility models usually involve over a dozen variables considered as predisposing factors of slope instability (e.g. Lee et
al., 2002 (13 variables); Lee and Choi, 2004 (15 variables);
van der Eeckhaut et al., 2010 (9 variables); Sterlacchini et
al., 2011 (9 variables)). Nevertheless, the evaluation of the
weight of each landslide predisposing factor within the predictive model through a thorough sensitivity analysis is frequently missing. In addition, the application of statistic bivariate methods to assess landslide susceptibility assumes
conditional independence (CI) of the landslide predisposing
factors (Bonham-Carter et al., 1989; Agterberg et al., 1993;
Van Westen, 1993; Agterberg and Cheng, 2002; Thiart et al.,
2003; Thiery et al., 2007). Blahut et al. (2010) pointed out
that spatial probabilities are overestimated when conditional
independence is not verified.
In this study, the aim is to determine the best combination
of landslide predisposing variables using a bivariate statistical model, based on the assessment of goodness of fit and
predictive power, using variables that have a high degree of
conditional independence. In addition, we assess the number of unique conditions within each landslide susceptibility
model associated to each combination of landslide predisposing variables. This number should be minimized when
landslide susceptibility maps are made for land use planning
and management in order to avoid the over partitioning of the
study area.
Descrição
Palavras-chave
Landslide predisposing factors Shallow landslide susceptibility models
Contexto Educativo
Citação
Pereira, S., Zezere, J. L., & Bateira, C. (2012). Technical note: assessing predictive capacity and conditional independence of landslide predisposing factors for shallow landslide susceptibility models. Natural Hazards and Earth System Sciences, 12(4), 979–988. https://doi.org/10.5194/nhess-12-979-2012.
Editora
Copernicus Publications