T150 Landslide Inventories, Hazard Assessments, and Risk Reduction

Monday, 23 September 2019
1:30 PM - 5:30 PM
Room 224A, North Building

A COMPARATIVE STUDY ON THE PREDICTIVE ABILITY OF THE INFORMATION VALUE METHOD (IVM), FREQUENCY RATIO (FR), LOGISTIC REGRESSION (LR) AND MAXIMUM ENTROPY MODEL (MAXENT) MODELS FOR DEVELOPING DEBRIS-SLIDE SUSCEPTIBILITY MAPS

Paper number: T150-1
Program: Technical Sessions
Time: 1:30 PM – 1:45 PM

Authors

Abstract

Debris-slide has been the most dominant form of slope failure in the Great Smoky Mountains National Park (GRSM). Predicting the spatial probability of future debris-slide events requires a generation of a debris-slide susceptibility map, the process of which can be classified into data-driven and knowledge-driven methods. The aim of the study is to develop four data-driven debris-slide susceptibility models using Information Value Method (IVM), Frequency Ratio (FR), Logistic Regression (LR) and Maximum Entropy Model (Maxent) for the Anakeesta formation in GRSM. The models were developed using six causative factors or geo-factors, i.e., elevation, curvature, soil texture, land use, annual rainfall and geological discontinuity data (kinematic index). Debris-slide locations were digitized from aerial photographs and satellite imagery, which were subsequently divided into 75:25 ratio for model training and testing, respectively. Both IVM and FR models calculate the density of debris-slide within the individual classes of geo-factors using different formulas. Debris–slide Susceptibility Index (DSI) was calculated using ArcGIS 10.6.0, for both the models. Coefficients of LR model were calculated using SPSS software and were imported to ArcGIS platform to develop the debris-slide susceptibility model. Maxent model was developed using presence only debris-slide data in the Maxent software with the help of logistic algorithm. The values of area under Receiver Operating Characteristic (ROC) curve were 0.855, 0.863, 0.856 and 0.853 for IVM, FR, LR and Maxent respectively, which suggests that the models have high prediction accuracy. While each individual model has its unique advantages and disadvantages, FR showed marginally higher ROC curve value indicating slightly better performance than others.

Geological Society of America Abstracts with Programs. Vol. 51, No. 5

doi: 10.1130/abs/2019AM-336050

© Copyright 2019 The Geological Society of America (GSA), all rights reserved. Permission is hereby granted to the author(s) of this abstract to reproduce and distribute it freely, for noncommercial purposes. Permission is hereby granted to any individual scientist to download a single copy of this electronic file and reproduce up to 20 paper copies for noncommercial purposes advancing science and education, including classroom use, providing all reproductions include the complete content shown here, including the author information. All other forms of reproduction and/or transmittal are prohibited without written permission from GSA Copyright Permissions.