Susceptibility models Sample Clauses

Susceptibility models. To evaluate the influence of land use change on landslide spatial occurrence we prepared three single models and one combined susceptibility zonation through different multivariate statistical analysis following the procedure proposed by Xxxxx et al. (2010) and further implemented during the project (Fig. 3.2). The prototype SW for regional landslide susceptibility modelling and zonations has been delivered and will be discussed in detail in the D6.2 “Prototype SW for regional landslide susceptibility modelling”. The model exploits different methods of multivariate statistical classification: (i) linear discriminant analysis model (LDA); (ii) quadratic linear discriminant analysis model (QDA); (iii) logistic regression model (LRM); and (iv) self-optimizing neural network model (NNM). Moreover, the model considers the combination (CFM) of the single modeling susceptibility results (LDA, QDA, LRM, NNM) or some of them exploiting a logistic regression approach. The model can be applied at different scales and can use polygon or regular pixel-based mapping units. The analysis requires two input variables for each cartographic unit: a binary grouping variable (i.e. dependent) showing the absence or presence (respectively 0 and 1) of a landslide, and a set of explanatory variables (i.e., independent). The explanatory variables describe the environmental characteristics for each cartographic unit.