Coastal Flood risk assessment using ensemble multi-criteria decision-making with machine learning approaches

Area of Science:

• Environmental Science
• Climate Change Adaptation
• Geographic Information Systems

Background:

• Coastal areas face escalating flood risks due to climate change-induced sea-level rise.
• Increased flood frequency and intensity are linked to anthropogenic activities and unplanned development.
• Effective flood susceptibility mapping is crucial for coastal flood risk management.

Purpose of the Study:

• To identify and map coastal flood susceptibility in Bandar Abbas, Iran.
• To evaluate the performance of ensemble machine learning models for coastal flood prediction.
• To assess flood risk under different climate change scenarios (RCP 2.6 & RCP 8.5).

Main Methods:

• Application of the Analytical Hierarchy Process (AHP) for multi-criteria decision-making.
• Ensemble modeling combining AHP with Support Vector Machine (AHP-SVM) and Decision Tree (AHP-DT).
• Statistical validation using Friedman and Wilcoxon signed rank tests, accuracy, sensitivity, and specificity metrics.

Main Results:

• The AHP-DT model achieved a high Area Under the Curve (AUC) of 0.95.
• Flood susceptibility mapping revealed low to very low risk in the northern and western Raidak Basin River areas due to topography.
• High flood susceptibility (CFSM) was identified in the eastern, middle section, with coastal populations facing low to medium exposure.

Conclusions:

• Ensemble machine learning models, particularly AHP-DT, are effective for coastal flood susceptibility mapping.
• Topographic characteristics significantly influence flood risk in coastal regions.
• The findings provide critical data for implementing targeted risk reduction strategies in vulnerable coastal zones.