South-Central Section - 59th Annual Meeting - 2025

Paper No. 22-6
Presentation Time: 8:30 AM-5:00 PM

PREDICTING URBAN HEAT ISLANDS AND LAND USE DYNAMICS IN NORTHWEST ARKANSAS USING MACHINE LEARNING


VAHID BELARGHOU, Rasool, Environmental Dynamics Program, University of Arkansas, Fayetteville, AR 72703 and ALY, Mohamed, Department of Geosciences, University of Arkansas, Fayetteville, AR 72701

Arkansas, often called the "natural state", is experiencing rapid urban expansion and significant land use/land cover (LULC) changes, which are impacting its environmental sustainability. This study investigates the relationship between land surface temperature (LST) and LULC changes in northwest Arkansas (NWA), with the goal of informing urban planning and promoting sustainable development strategies. The research evaluates the effectiveness of advanced machine learning techniques, particularly convolutional neural networks (CNN), in predicting future LST and LULC changes, which can enhance land use planning and disaster management practices. A key focus of the study is the identification and analysis of urban heat islands (UHIS) in NWA cities, examining how LST and LULC variations contribute to temperature dynamics, especially in urban areas. By leveraging Landsat-5/-8 satellite data, the study tracks seasonal land cover indices and correlations between LST and LULC changes over two decades (2001-2021). The findings highlight how urbanization, vegetation cover, and other land cover types influence local climate conditions, including the formation and intensity of urban heat islands. To further support long-term climate resilience and urban sustainability, the study employs a combination of machine learning algorithms — specifically the cellular automaton (CA)-Markov chain model — to simulate future urbanization scenarios and forecast their impact on LST and LULC. This research not only provides empirical insights into the role of land cover change in shaping local climate patterns but also offers a methodological framework for future studies in urban sustainability, climate resilience, and deep learning applications in environmental science. The findings aim to guide policymakers, urban planners, and environmental stakeholders in developing strategies to mitigate the effects of urban heat islands, optimize land use, and promote climate adaptation in the face of growing urban pressures.