A multi-scale mapping approach of local climate zones: A case study in Hong Kong

The spatial scale of Local Climate Zone (LCZ) mapping significantly affects classification accuracy and the understanding of the relationship between land surface characteristics and urban climates. Traditional “one-size unit” mapping often fails to capture actual climatic conditions and inconsistencies between LCZs, limiting its ability to precisely identify urban heat island effects. Using Hong Kong as a case study, this study investigated optimal LCZ mapping scales (LCZ-scales) based on homogenous Land Surface Temperature (LST) representation within individual LCZs through semi-variogram modelling. The impact of immediate surroundings on LCZ-scales and LST was examined using Spearman correlation analysis and Generalized Linear Models. A multi-scale LCZ map was developed and its accuracy in classifying land use and representing climate conditions was evaluated using Analysis of Variance (ANOVA). Results show that, first, optimal mapping scales vary across LCZ types: 260–390 m for built types and 320–425 m for land cover types, forming a multi-scale mapping approach. Second, LCZ-scales for LCZ 1, 2, 5, A-B, C, D, E, F and G can be refined depending on surrounding LCZ configurations. Third, the “surrounding effect” on LST highlighted detailed UHI-mitigation strategies—generally, maintaining proximity to vegetation and water bodies within 900 m and 1200 m and beyond 900 m from LCZ 1 and 10 can effectively mitigate urban heat. Fourth, the multi-scale LCZ map better recognizes homogeneous land surface patterns and differentiates thermal characteristics than the “one-size unit” LCZ map. The findings of this study can inform climate-responsive urban planning, especially in urban–rural transition zones.