Unraveling the diurnal-nocturnal thermal dynamics across urban functional zones: thresholds and interactions

Rapid urbanization has intensified urban thermal stress, while increasingly frequent extreme heat events have made urban heat island (UHI) mitigation a critical priority for sustainable urban development. The UHI effect persists throughout the diurnal cycle and is shaped by multiple urban thermal environmental factors. However, how these factors influence land surface temperature (LST) across different urban functional zones (UFZs) and observation periods remains insufficiently quantified. Therefore, this study employed extreme gradient boosting (XGBoost) model and Shapley additive explanations (SHAP) model to investigate the relationships between urban thermal environmental factors and diurnal LST across UFZs, using Hangzhou in China as a case study. The results indicated that: (1) During daytime, the dominant drivers of LST differed across both functional zones and observation times, with three-dimensional urban morphology generally playing a major role. At night, the LST variations across UFZs were largely governed by three-dimensional urban morphology and socioeconomic factors. (2) Within different UFZs, the dominant factors showed pronounced temporal differences and nonlinear threshold effects. The warming effect of mean building density (MBD) increased sharply when it exceeded approximately 0.15 in industrial zones. Stable cooling effects emerged when mean tree height (MTH) exceeded approximately 10–12 m in residential and public service zones and approximately 15 m in commercial zones. Nighttime radiative cooling generally occurred when sky view factors (SVF) exceeded approximately 0.50–0.60 in residential and commercial zones and 0.75–0.80 in industrial zones. (3) LST in different UFZs was also affected by interactions among urban thermal factors. In residential zones, MTH values above 10 m contributed to cooling when floor area ratio (FAR) was below approximately 1.0 at 08:56 and 1.5 at 16:26, whereas this cooling effect weakened or shifted to warming under higher-FAR conditions. These findings provide quantitative evidence and decision support for time-specific and function-oriented optimization to improve the urban thermal environment.