Under global warming and rapid urbanization, understanding the link between residential spatial layouts and carbon emissions, considering microclimate effects, is crucial. Using the Local Climate Zone (LCZ) framework, this study selected 60 residential neighborhoods in Suzhou (hot summer and cold winter zone). A microclimate–neighborhood morphology coupling model was built by integrating the Urban Weather Generator (UWG) and Urban Metabolism-Induced Energy (UMI) model. The factor method was used to simulate energy use and carbon emissions of building clusters. The study systematically analyzed eight spatial form indicators, their influence on carbon emissions, and developed a predictive model. Main conclusions: (1) Carbon emission intensity ranks: LCZ6 > LCZ3 > LCZ5 > LCZ4 > LCZ2 > LCZ1. (2) Shape factor is positively correlated with carbon emissions across all LCZs, being the most stable and significant predictor. (3) Dominant factors vary by LCZ: sky view factor for compact high-rise (LCZ1); scattering degree for open high-rise (LCZ4); shape factor for compact mid-rise (LCZ2), open mid-rise (LCZ5) and open low-rise (LCZ6); no significant factor for compact low-rise (LCZ3). These findings can support low-carbon residential design, energy-efficient renovation of old neighborhoods, and optimization of residential evaluation standards.