Predictive Analytical Framework for Identifying Vapor Intrusion Risks across Urban Redevelopment Zones

Abstract

Urban redevelopment initiatives increasingly involve the repurposing of former industrial and commercial lands, many of which contain legacy contaminants capable of migrating into overlying structures through vapor intrusion pathways. Accurately identifying and predicting vapor intrusion risks is essential for protecting public health, guiding land-use decisions, and ensuring regulatory compliance during redevelopment processes. This study proposes a predictive analytical framework that integrates environmental forensics, subsurface vapor transport modeling, and machine-learning-enhanced risk classification to evaluate vapor intrusion potentials across diverse urban redevelopment zones. The framework incorporates high-resolution spatial datasets including soil gas measurements, groundwater contaminant plumes, building characteristics, and land-use variables to generate comprehensive exposure profiles. A hybrid modeling approach is implemented, combining physics-based vapor transport models with data-driven predictive algorithms to capture complex interactions among contaminant concentration gradients, soil permeability, building ventilation patterns, and climatic influences. Feature-ranking techniques and sensitivity analyses were employed to determine the most influential parameters affecting vapor intrusion likelihood. The framework further integrates temporal monitoring data to capture seasonal fluctuations in subsurface vapor behavior and potential intrusion episodes. Model accuracy was assessed through cross-validation against regulatory screening data and field-verified vapor intrusion cases. Results indicate that redevelopment zones with shallow groundwater tables, fractured subsurface geology, aging building foundations, and historical chlorinated solvent use exhibit significantly elevated intrusion risks. The predictive framework also identifies micro-scale spatial variability in intrusion hotspots, highlighting built-environment factors such as sub-slab depressurization, sewer line connectivity, and basement construction materials as critical determinants of exposure. The integrated model demonstrates strong predictive performance, providing a robust and scalable tool for prioritizing mitigation strategies, optimizing site investigation resources, and informing zoning decisions in urban redevelopment contexts. This study contributes a scientifically grounded and operationally practical framework for assessing vapor intrusion risks using predictive analytics. By combining hydrological, geospatial, and machine-learning techniques, the approach advances proactive environmental risk management and supports sustainable, health-protective redevelopment planning in urban communities.