Hybrid Supervised-Unsupervised Modeling for Post-Hurricane Private Well Contamination Risk Score Using Empirical...

Area of Science:

• Environmental Science
• Public Health
• Data Science

Background:

• Hurricane-related flooding mobilizes contaminants, posing risks to private wells.
• Limited well testing and disaster response capacity exacerbate risks for private well users.

Purpose of the Study:

• To develop and validate a data-driven framework for quantifying post-hurricane private well contamination risk.
• To integrate geospatial variables and machine learning for risk assessment.

Main Methods:

• A hybrid machine learning approach combining supervised and unsupervised learning was used.
• Seventy-eight geospatial variables were integrated into hazard, vulnerability, and capacity modules.
• The framework was applied to western North Carolina post-Hurricane Helene and validated with well testing data.

Main Results:

• Higher risk scores were significantly associated with increased total coliform contamination (p=0.006).
• The framework demonstrated value in identifying areas with elevated contamination risk.
• Predictive performance was moderate, indicating a need for further evaluation.

Conclusions:

• The data-driven framework provides a practical tool for identifying high-risk private well areas after extreme weather events.
• The framework supports disaster preparedness, prioritizes well testing, and protects public health.
• The transferable framework can be adapted to other regions and storm events.