Using AI to Spatially Map Urban Features and Urban Heat Island Effects in Huntsville City Alabama
Board Location: #182
Discipline: Computer Sciences and Information Management
Session: 2
Chance Williams - Alabama A&M University
Co-Author(s): Ranjani W. Kulawardhana, Alabama A&M University, Huntsville, Alabama
The Urban Heat Island (UHI) effect, where urban areas experience temperatures up to 7°C higher than surrounding rural zones, poses critical challenges to climate resilience, energy consumption, and public health. Urban structures such as buildings, roads, and impervious surfaces contribute significantly to this phenomenon by absorbing and retaining heat. Quantifying the impacts of urban landscapes at varying scales remains a significant challenge due to their complexity and variability. This study seeks to evaluate applicability of Artificial Intelligence (AI) to detect UHIs using satellite Remote Sensing (RS) data and products. Our specific objectives are: 1) to spatially map urban features using AI and Landsat 8 multispectral data & 2) to detect UHIs using satellite derived Land Surface Temperature (LST) products and AI. AI offers a scalable, data-driven approach to analyze UHI contributors with unprecedented precision. Using high-resolution satellite imagery over urban areas, machine learning algorithms can map impervious surface areas, vegetation cover, and heat-retentive materials at a spatial resolution as fine as a few centimeters. Huntsville city, located within Madison and Limestone counties of Alabama is one of the fastest growing cities in the southeastern United States and # 16 fastest in the country. Thus, for this study, Madison and limestone counties were selected as the study area. The study consists of two overarching objectives, to spatially map urban features in Huntsville using AI and Landsat 8 multispectral data & detect UHIs using satellite derived Land Surface Temperature (LST) products and AI. Our preliminary findings reveal the applicability of AI for mapping urban areas and identifying building borders. However, challenges in using AI algorithms at large scale mapping primarily due to density of urban features and image resolution leading to polygon overlapping were detected. Thus, further processing of AI-derived outputs and reporting of accuracy estimates remains a critical first step prior to using said outputs in management and decision making relating to any urban planning and mitigation strategies.
Funder Acknowledgement(s): DOE RENEW Grant DE-SC0024614
Faculty Advisor: Ranjani Kulawardhana, ranjani.kulaward@aamu.edu
Role: data processing, analysis and writing