Urbanization has dramatically reshaped the urban landscape and intensified the urban heat island (UHI) effect in recent decades. This study investigates the spatiotemporal dynamics of UHI across 31 major Chinese cities from 2000 to 2020, using the local climate zone (LCZ) framework, Moderate Resolution Imaging Spectrometer-derived land surface temperature (LST), and machine learning models. Existing research has found significant differences in the level of UHI variation among different cities, but few studies attributing it from the perspective of urban morphology changes. We quantify urbanization processes through nine LCZ-derived indicators and assess their influence on daytime and nighttime UHI in summer. Results show that UHI intensity increased in 80.6% of cities during the day and in 77.4% at night, with over half experiencing statistically significant intensification (p < 0.05). Daytime UHI was stronger in southern cities (mean: 3.12 βC), while nighttime UHI was more prominent in northern regions (mean: 1.86 βC). Urban expansion (UE) and vegetation reduction were identified as the primary contributors to UHI intensification, with random forest and Shapley Additive exPlans analyses indicating that UE accounted for up to 28.3% of UHI variance. Functional transformation of urban areas showed mixed thermal effects, with stronger nighttime impact. High-rise (LCZ4) and mid-rise (LCZ5) building expansions were particularly linked to elevated LST. Moreover, three land use indices exhibited significant correlations with UHI trends, especially in cities with significant UHI growth. This study provides a comprehensive attribution analysis of UHI dynamics at the national scale, offering empirical evidence to guide sustainable urban planning and climate-adaptive city design.