Surface urban heat island disparities and the role of urban form across the urban-rural gradient

Urban Heat Island (UHI) poses serious challenges to urban sustainability, energy efficiency, and public health. Despite growing concerns, the interactions across urban-rural zones and the factors driving Surface UHI Intensity (SUHII) remain insufficiently understood. This study investigates SUHII dynamics along urban-rural gradients in 258 Chinese cities using Geographical Convergent Cross Mapping (GCCM) and interpretable machine learning. To capture spatial variations, four zones were defined based on urban land density: urban core, inner urban, suburban, and urban fringe. Results show a clear thermal gradient, with SUHII decreasing from 2.589 °C in the urban core to 1.265 °C at the fringe. Northern cities, especially those in the middle temperate zone, exhibit the highest SUHII in the urban core, reaching up to 3.374 °C. GCCM analysis reveals asymmetric predictive influence: the suburban zone has a cooling influence on the core, while the core's heating effect diminishes with distance. Machine learning analysis highlights that the influence of urban form factors varies along the gradient. Two-dimensional (2D) factors consistently play a more prominent role than three-dimensional (3D) ones. In the inner urban zone, the impervious surface fraction is the most influential factor, with a significantly positive effect. At the fringe, patch density becomes dominant, showing a negative correlation with SUHII. Additionally, building footprint coverage at the fringe has a significantly positive effect on SUHII once it exceeds 6 %. In the urban core, 3D factors are more critical. Average building height begins to reduce SUHII once it surpasses 17 m. Similarly, building volume density exhibits a positive effect up to 17 m³/m², beyond which the influence reverses. These findings underscore the need for zone-specific strategies to mitigate SUHII.