Landslide Detection and Susceptibility Modeling on Cameron Highlands (Malaysia): A Comparison between Random Forest, Logistic Regression and Logistic Model Tree Algorithms

Viet-Ha Nhu,Khalil Valizadeh Kamran,Himan Shahabi,Hoang Nguyen,Wei Chen,Sadra Karimzadeh,Baharin Bin Ahmad,Ataollah Shirzadi,Ayub Mohammadi,Nadhir Al-Ansari,Marten Geertsema,Victoria R Kress

doi:10.3390/f11080830

Abstract

We used remote sensing techniques and machine learning to detect and map landslides, and landslide susceptibility in the Cameron Highlands, Malaysia. We located 152 landslides using a combination of interferometry synthetic aperture radar (InSAR), Google Earth (GE), and field surveys. Of the total slide locations, 80% (122 landslides) were utilized for training the selected algorithms, and the remaining 20% (30 landslides) were applied for validation purposes. We employed 17 conditioning factors, including slope angle, aspect, elevation, curvature, profile curvature, stream power index (SPI), topographic wetness index (TWI), lithology, soil type, land cover, normalized difference vegetation index (NDVI), distance to river, distance to fault, distance to road, river density, fault density, and road density, which were produced from satellite imageries, geological map, soil maps, and a digital elevation model (DEM). We used these factors to produce landslide susceptibility maps using logistic regression (LR), logistic model tree (LMT), and random forest (RF) models. To assess prediction accuracy of the models we employed the following statistical measures: negative predictive value (NPV), sensitivity, positive predictive value (PPV), specificity, root-mean-squared error (RMSE), accuracy, and area under the receiver operating characteristic (ROC) curve (AUC). Our results indicated that the AUC was 92%, 90%, and 88% for the LMT, LR, and RF algorithms, respectively. To assess model performance, we also applied non-parametric statistical tests of Friedman and Wilcoxon, where the results revealed that there were no practical differences among the used models in the study area. While landslide mapping in tropical environment such as Cameron Highlands remains difficult, the remote sensing (RS) along with machine learning techniques, such as the LMT model, show promise for landslide susceptibility mapping in the study area.

Highlights

Landslides are geohazards that claim lives and damage property every year [1,2]
Remote sensing and geographic information system technologies have been widely used in the application of different models to generate landslide susceptibility maps, which are very important for policy and decision-makers to monitor and protect areas prone to landslides
The current study released the results of comprehensive research on landslide susceptibility mapping in the Cameron

Summary

Introduction

Landslides are geohazards that claim lives and damage property every year [1,2]. One of the main challenges of land managers is to predict where landslides will occur, and landslide susceptibility maps (LSMs) are created to provide authorities with valuable information [3]. Landslides have been a challenge in the Cameron Highlands of Malaysia for some time and have damaged infrastructure, property, and natural resources [4,5,6,7]. There is limited spatial information about landslides in the Cameron Highlands of Malaysia, making it difficult for authorities to manage landslide-prone areas. Improvements of landslide detection through remote sensing and innovative machine learning algorithms can bolster landslide susceptibility mapping (LSM) efforts and stand to help authorities manage landslide prone areas [8,9]

Methods

Results

Discussion

Conclusion