Uncertainty Reduction in Flood Susceptibility Mapping Using Random Forest and eXtreme Gradient Boosting Algorithms in Two Tropical Desert Cities, Shibam and Marib, Yemen

Approximately 9.93% to 15.69% of Shibam City and 8.42% to 9.89% of Marib City are highly vulnerable to floods, according to a study that reduced uncertainty in flood susceptibility mapping using advanced machine learning algorithms. The research employed random forest (RF) and extreme gradient boosting (XGB) methods, incorporating a variable drop-off loop function to address model uncertainty and overfitting issues. By analyzing historical flooding data alongside twelve topographic and geo-environmental factors, the study demonstrated that the drop-off loop function significantly improved accuracy and processing efficiency in flood prediction models. These findings contribute valuable strategies for mitigating natural disaster risks, particularly in tropical desert regions like Yemen.

Al-Aizari, A. R., Alzahrani, H., AlThuwaynee, O. F., Al-Masnay, Y. A., Ullah, K., Park, H.-J., Al-Areeq, N. M., Rahman, M., Hazaea, B. Y., & Liu, X. (2024). Uncertainty Reduction in Flood Susceptibility Mapping Using Random Forest and eXtreme Gradient Boosting Algorithms in Two Tropical Desert Cities, Shibam and Marib, Yemen. Remote Sensing, 16(2), 336. https://doi.org/10.3390/rs16020336