Landslide Susceptibility Modeling Using Machine Learning Algorithm: The Case Of Sile And Elgo Catchments, Gamo Highland, Southern Ethiopia

Abstract

Landslides Are One Of The Most Prevalent Global Geologic Hazards, Causing Significant Damage To People And Property. They Are Prevalent In The Highlands Of Ethiopia. The Primary Purpose Of This Study Was To Develop A Comprehensive Landslide Susceptibility Modeling Utilizing Machine Learning Algorithms In The Sile And Elgo Catchment Areas. This Study Selected Ten Factors Influencing Landslides For Landslide Susceptibility Mapping. These Factors Included Elevation, Slope, Aspect, Curvature, Distance To Streams, Distance To Lineaments, Soil Type, Lithology, Land Use And Land Cover, And Rainfall. Data For These Parameters Were Primarily Derived From Google Earth Imagery, Digital Elevation Models (Dem), And Sentinel-2 Imagery. Through Qgis And Google Earth Imagery Interpretation, 1068 Landslides Were Identified Within The Study Area. The Analysis Revealed That Soil Type, Followed By Slope Are The Most Significant Factors Influencing Landslide Occurrence, As Indicated By The Frequency Ratio Value. The Support Vector Machine (Svm) And K-Nearest Neighbors (Knn) Models Were Employed To Assess Landslide Susceptibility. Model Accuracy Was Evaluated Using The Receiver Operating Characteristic (Roc) Curve, And The Area Under The Curve (Auc) Values For A Success Rate Of 0.799 For Svm And 0.81 For Knn. The Results Indicate That The Support Vector Machine (Svm) Model Is The Most Effective For Assessing Landslide Susceptibility In The Study Area, And Highlights The Efficacy Of Machine Learning. The Study Demonstrates The Efficacy Of Machine Learning In Landslide Susceptibility Mapping While Suggesting Potential Advancements Through Deep Learning Techniques.