H. Norouzi, A. Bah, M. Muntaha, L. Metlitsky, R. Blake
City University of New York,
United States
Keywords: urban heat, machine learning, digital twin, remote sensing, subway microclimate, smart cities, predictive modeling, infrastructure resilience, grid load, geospatial AI
Summary:
Extreme heat is an emerging threat to public transit systems, passenger safety, and infrastructure resilience in major cities. This project introduces an AI-enabled heat monitoring and predictive modeling platform designed for the Metropolitan Transportation Authority (MTA) subway network in New York City. Our system integrates distributed temperature and humidity sensing across multiple stations with satellite-based remote sensing, machine learning, and digital-twin technologies to forecast heat conditions in real time. Using a network of sensors, we continuously monitor environmental variables at both platform and street levels. These in-situ measurements are tarined by machine-learning models predict platform-level heat indices under varying meteorological and passenger-flow scenarios. The resulting predictive framework identifies “hot-spot” stations and temporal risk zones, providing decision support for targeted cooling strategies, ventilation optimization, and passenger advisories. The pilot implementation, conducted across selected NYC subway stations, demonstrates the feasibility of real-time temperature mapping, early-warning alerts, and integration with digital twin visualization tools for operational planning. The platform’s modular design enables scalability to other cities and transit systems facing similar climate challenges. This work advances the intersection of geospatial AI, distributed sensing, and infrastructure resilience, offering a replicable model for mitigating urban heat risk and enhancing commuter safety in a warming world.