Abstract
Understanding the spatiotemporal relationship between urban spatial structure and carbon emissions is essential for achieving sustainable urban development. However, the underlying mechanisms driving their complex interactions remain insufficiently explored. This study employs machine learning and multiscale geographically weighted regression (MGWR) to investigate the spatial and temporal dynamics of urban spatial structure and their impact on carbon emissions in the Yangtze River Economic Belt (YREB). The results reveal significant spatial heterogeneity, with carbon emissions highly concentrated in Shanghai, Jiangsu, and Zhejiang province, which are situated in the lower of Yangtze River Economic Belt, while other regions exhibit a general upward trend, characterized by urban expansion towards peripheral areas. Driving forces analysis highlights the varying effects of urban form attributes, including breadth, complexity and compactness, on carbon emissions. These findings offer theoretical insights into optimizing urban spatial structures and provide scientific support for policymakers to implement targeted carbon reduction strategies and promote sustainable urban transformation.