Solving the spatial extrapolation problem in flood susceptibility using hybrid machine learning, remote sensing, and...

Area of Science:

• Environmental Science
• Geospatial Analysis
• Artificial Intelligence

Background:

• Floods are a major natural hazard with increasing impacts on human life and economies.
• Effective water resource management requires improved flood susceptibility modeling.
• The extrapolation of flood models to new regions remains a significant challenge.

Purpose of the Study:

• To develop and evaluate machine learning (ML) models for addressing the extrapolation problem in flood susceptibility modeling.
• To compare the performance of deep neural network (DNN) models integrated with various optimization algorithms.

Main Methods:

• Utilized deep neural networks (DNNs) combined with six optimization algorithms: Earthworm Optimization Algorithm (EOA), Wildebeest Herd Optimization (WHO), Biogeography-Based Optimization (BBO), Satin Bowerbird Optimizer (SBO), Grasshopper Optimization Algorithm (GOA), and Particle Swarm Optimization (PSO).
• Applied models to flood susceptibility mapping in Quang Nam Province and tested extrapolation to Nghe An Province.
• Evaluated model performance using Root Mean Square Error (RMSE), Receiver Operating Characteristic (ROC), Area Under the ROC Curve (AUC), and Accuracy (ACC).

Main Results:

• All developed models demonstrated high performance in flood susceptibility mapping, with AUC values greater than 0.9.
• The DNN-BBO model achieved the highest AUC (0.99), closely followed by DNN-WHO (0.99) and DNN-SBO (0.98).
• The models successfully addressed the extrapolation problem, proving their capability to assess flood susceptibility in new geographical areas.

Conclusions:

• The novel ML models effectively solve the extrapolation problem in flood susceptibility modeling.
• These models offer a valuable reference for urban planners and decision-makers in coastal regions facing flood risks.
• The developed approach can be adapted to evaluate flood susceptibility globally, enhancing disaster preparedness.