Predict the Riverbank Erosion Susceptibility for the Ham Luong River Using the Logistic Regression Model

Riverbank erosion, once considered an inevitable process, has emerged as a severe and unpredictable issue, exacerbated by climate change and human activities. The Vietnamese Mekong Delta (VMD) has faced significant erosion, leading to considerable infrastructural damage and economic losses. This study aims to predict riverbank erosion susceptibility along the Ham Luong River using a logistic regression (LR) model for 100 riverbank locations, classified as eroded or stable, with twelve conditioning variables. The LR model achieved an overall accuracy (ACC) of 0.83 and an area under the receiver operating characteristic curve (AUC) of 0.86 through 15-fold cross-validation. Sensitivity analysis identified the bank slope, soil moisture, and bank height as key factors influencing erosion susceptibility. Probability analysis revealed that an increased bank slope may cause greater riverbank instability, whereas higher soil moisture levels may reduce erosion susceptibility. These findings highlight the importance of stabilizing bank slopes and maintaining soil moisture to mitigate the riverbank erosion susceptibility effectively, emphasizing the need for managing water levels in rivers and canals during the dry season as part of disaster risk management in the VMD.