Mapping local climate zones for cities: A large review

The local climate zone (LCZ) system provides a universal classification mechanism for urban and natural landscapes and plays an increasingly important role in urban climate research. With the rapid development of various LCZ mapping methods, a thorough survey of the LCZ mapping literature is urgently needed to better understand current progress, challenges, and future directions. Accordingly, this study provided a comprehensive review of the LCZ mapping literature during 2012–2021, with a detailed analysis on literature statistics, research topics, LCZ cities, and active research groups. Furthermore, remote sensing (RS)-based LCZ mapping methods were elucidated from feature sets, classification units, training areas, classification algorithms, and accuracy assessment; geographic information system (GIS)-based LCZ mapping methods were elaborated from LCZ parameters, basic spatial units, classification algorithms, and accuracy assessment; and their combination methods were summarized from two typical integration strategies. Finally, several challenges and future directions for LCZ mapping were discussed. The topics include exploiting multi-source RS and GIS data, determining appropriate LCZ mapping unit sizes, acquiring high-quality LCZ ground truth data, improving LCZ classification algorithms, optimizing LCZ parameters and subclasses, exploring the transferability of LCZ models, conducting global interannual LCZ mapping, and expanding the application of LCZs. The research community can quickly obtain abundant information on the LCZ mapping literature, understand the frameworks of different LCZ mapping methods, and inspire new directions for future research. •Progress, challenges, and prospects for LCZ mapping are systematically reviewed.•Frameworks of RS- and GIS-based LCZ mapping methods are elaborated.•Data sources, mapping unit sizes, classification algorithms, and validation strategies are analyzed.•Deep learning models, domain adaption methods, and benchmark datasets improve LCZ mapping.•GEE, deep learning, crowdsourcing, and collaboration facilitate global interannual LCZ mapping.