Earthquake and rainfall-induced landslide hazard assessment of Kutupalong Rohingya camp using meteorological and geological information

Abstract

The study aims to assess the landslide hazard scenario of the Kutupalong Rohingya camp, considered one of the largest humanitarian shelters for conflict-fled people. The camp area lying in a prominent tectonic setting and meteorological environment, is vulnerable to landslide hazards, specifically in the monsoon season. The worst-case scenario of earthquake and rainfall occurrence at the same time or one after another is considered in this study. Landslide favorable parameters (topographic, soil physical property, and contributing factors) are designed in the study to identify the hazard-prone areas through the factor of safety computation. Topographic parameters are prepared by combining field investigation and remote sensing-based information. Soil physical properties are designed in two different models to account for their uncertainties; normal probabilistic distribution and interpolation-based spatial distribution. The contributing factors (earthquake and rainfall) have been assessed using the probabilistic approach. The Peak Ground Acceleration (PGA) exceedance and rainfall intensities of 50, 75, 100, 200, and 475 years return period are applied in the factor of safety calculation. The factor of safety calculation has been run using two different methods; Monte-Carlo simulation using Python scripts and direct estimation method. Different duration (rainfall intensity) scenarios are also modeled to estimate the vulnerable areas in these models. The findings of the study are finally validated against field investigation-based landslide inventory with more than 84% accuracy at a 90% confidence interval.

Assessment of geological hazard susceptibility based on remote sensing images

“…Tsangaratos et al used a landslide example to compare the different advantages of logistic regression classifiers and classifiers using the Naive Bayes algorithm, giving researchers new ideas for research [37]. Hong et al compared the frequency ratio (FR), certainty factor (CF), and entropy index (IOE) to map the susceptibility of landslide areas due to rainfall and determine their overall performance [42]. Pradhan et al used a back-propagation artificial neural network model to analyze landslide susceptibility and recommended protection accordingly [38].…”

“…The spatial effect will impact the evaluation structure. Hong et al constructed three landslide susceptibility models using FR, CF, and IOE with 15 factors related to landslides and validated and compared these models using known landslide locations and receiver operating characteristic curves [42].…”