Omar Yasser, Omar Ads, Izni Zahidi, Chow Ming Fai
ABSTRACT Urban heat islands (UHIs) pose a significant environmental issue in rapidly urbanizing areas, intensifying the need for accurate methods to assess and mitigate their effects. The Local Climate Zones (LCZ) classification system is a standardized land use approach for urban heat island (UHI) research. Many studies reported difficulties in mapping LCZs with high accuracy, and the current state of the literature shows discrepancies in the accuracy of LCZ mapping methods. To address these inconsistencies, this study compares commonly used LCZ mapping methods, namely, WUDAPT, GEE, CNN, GIS, and combined methods. The LCZ layouts of Kuala Lumpur and Singapore were classified using six mapping schemes from the literature to determine the most accurate LCZ mapping method for UHI studies. Remote sensing (RS)‐based schemes employed Landsat 8 scenes, while the GIS‐based scheme used building footprint and land use data. The CNN schemes achieved the highest accuracies, with the 300‐m resolution model scoring an overall accuracy of ~95% for both cities. The GIS‐based and combined methods failed to map LCZs with high accuracy and fell behind the RS‐based methods. This study acts as a valuable benchmark for LCZ classification and offers insights that can guide future developments in this field.