Skip to main content
Passa alla visualizzazione normale.

CHRISTIAN CONOSCENTI

Slope units-based flow susceptibility model: using validation tests to select controlling factors

  • Authors: Rotigliano, E; Cappadonia, C; Conoscenti, C; Costanzo, D; Agnesi, V
  • Publication year: 2012
  • Type: Articolo in rivista (Articolo in rivista)
  • Key words: Landslide susceptibility, Univariate multiparametric model, validation, Mapping units
  • OA Link: http://hdl.handle.net/10447/62271

Abstract

A susceptibility map for an area, which is representative in terms of both geologic setting and slope instability phenomena of large sectors of the Sicilian Apennines, was produced using slope units and a multiparametric univariate model. The study area, extending for approximately 90 km2, was partitioned into 774 slope units, whose expected landslide occurrence was estimated by averaging seven susceptibility values, determined for the selected controlling factors: lithology, mean slope gradient, stream power index at the foot, mean topographic wetness index and profile curvature, slope unit length, and altitude range. Each of the recognized 490 landslides was represented by its centroid point. On the basis of conditional analysis, the susceptibility function here adopted is the density of landslides, computed for each class. Univariate susceptibility models were prepared for each of the controlling factors, and their predictive performance was estimated by prediction rate curves and effectiveness ratio applied to the susceptibility classes. This procedure allowed us to discriminate between effective and non-effective factors, so that only the former was subsequently combined in a multiparametric model, which was used to produce the final susceptibility map. The validation of this map latter enabled us to verify the reliability and predictive performance of the model. Slope unit altitude range and length, lithology and, subordinately, stream power index at the foot of the slope unit demonstrated to be the main controlling factors of landslides, while mean slope gradient, profile curvature, and topographic wetness index gave unsatisfactory results.