Novel Hybrid Evolutionary Algorithms for Spatial Prediction of Floods

Area of Science:

• Environmental Science
• Artificial Intelligence
• Geographic Information Systems (GIS)

Background:

• Flood spatial modeling is crucial for disaster management in vulnerable areas.
• Traditional methods often lack the precision required for accurate flood prediction.
• Ensemble artificial intelligence (AI) models offer advanced capabilities for complex environmental modeling.

Purpose of the Study:

• To develop and evaluate novel GIS-based ensemble AI models for flood spatial modeling.
• To compare the performance of ANFIS-ICA and ANFIS-FA models against existing machine learning techniques.
• To identify the most effective model for flood risk assessment and sustainable management.

Main Methods:

• Adaptive Neuro-Fuzzy Inference System (ANFIS) integrated with Imperialistic Competitive Algorithm (ICA) and Firefly Algorithm (FA).
• Application of ANFIS-ICA and ANFIS-FA models for flood spatial modeling using ten influential factors in the Haraz watershed.
• Model validation using statistical error indices (RMSE, MSE), statistical tests (Friedman, Wilcoxon signed-rank), and Area Under the Curve (AUC).

Main Results:

• Both ANFIS-ICA and ANFIS-FA models demonstrated good fit and prediction accuracy.
• The ANFIS-ICA model achieved the highest prediction accuracy (AUC = 0.947).
• ANFIS-ICA outperformed other models including Bagging-LMT, BLR, LMT, ANFIS-FA, LR, and RF.

Conclusions:

• The ANFIS-ICA model shows significant potential as a reliable tool for flood spatial modeling.
• This GIS-based ensemble AI approach offers a promising method for sustainable management of flood-prone areas.
• The study highlights the effectiveness of advanced AI techniques in improving environmental risk assessment.