Centre de documentation
scientifique

A comparative approach of support vector machine kernel functions for GIS-based landslide susceptibility mapping / Khalil Valizadeh Kamran in Applied geomatics, vol 13 n° 4 (December 2021)  

Public [article]  inApplied geomatics > vol 13 n° 4 (December 2021) . - pp 837 - 851 Titre : A comparative approach of support vector machine kernel functions for GIS-based landslide susceptibility mapping Type de document : Article/Communication Auteurs : Khalil Valizadeh Kamran, Auteur ; Bakhtiar Feizizadeh, Auteur ; Behnam Khorrami, Auteur ; et al., Auteur Année de publication : 2021 Article en page(s) : pp 837 - 851 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Analyse spatiale [Termes IGN] analyse de sensibilité [Termes IGN] apprentissage automatique [Termes IGN] cartographie des risques [Termes IGN] classification par séparateurs à vaste marge [Termes IGN] effondrement de terrain [Termes IGN] fonction de base radiale [Termes IGN] Iran [Termes IGN] méthode fondée sur le noyau [Termes IGN] occupation du sol [Termes IGN] pente [Termes IGN] risque naturel [Termes IGN] système d'information géographique [Termes IGN] utilisation du sol Résumé : (auteur) Landslides are among the most destructive natural hazards with severe socio-economic ramifications all around the world. Understanding the critical combination of geoenvironmental factors involved in the occurrence of landslides can mitigate the adverse impacts ascribed to them. Among the several scenarios for studying and investigating this phenomenon, landslide susceptibility mapping (LSM) is the most prominent method. Applying the machine learning (ML) algorithms integrated with the geographic information systems (GIS) has become a trending means for accurate and rapid landslide mapping practices in the scientific community. Support vector machine (SVM) has been the most commonly applied ML algorithm for LSM in recent years. The current study aims to implement different SVM kernel functions including polynomial kernel function (PKF) (degree 1 to 5), radial basis function (RBF), sigmoid, and linear kernels, for a GIS-based LSM over the Tabriz Basin (TB). To this end, a total number of 9 conditioning parameters being involved in the occurrence of the landslide events were determined and utilized. The LSM maps of the TB were generated based on the different SVM kernels and were statistically validated according to the landslide inventory. The findings revealed that the polynomial-degree-2 (PKF-2) model (AUC = 0.9688) outperforms the rest of the utilized kernels. According to the SLM map generated through PKF-2, the northernmost parts of the TB are extremely susceptible to slope failures than the rest; therefore, the developmental policies over these parts have to be taken into account with privileged priority to hinder any humanitarian as well as environmental catastrophes. Numéro de notice : A2021-858 Affiliation des auteurs : non IGN Thématique : GEOMATIQUE Nature : Article DOI : 10.1007/s12518-021-00393-0 Date de publication en ligne : 28/08/2021 En ligne : https://doi.org/10.1007/s12518-021-00393-0 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=99066 [article]