Constructing cross-scale local carbon emission zones in four global metropolises: A city-specific framework for mitigating spatial scaling effects

Accelerating the transition to low-carbon urbanisation was critical for climate change mitigation. The delineation of local carbon emission zones (LCEZs) offers a promising approach for integrating urban morphology with CO₂ emissions patterns for targeted planning. However, existing LCEZs frameworks are affected by scale dependence and often provided limited, context-insensitive explanations of emission heterogeneity. This study developed a cross-scale LCEZs framework and applied it to London, New York, Paris, and Sydney. First, the collapse method, grounded in finite-size scaling, was used to identify morphology factors exhibiting cross-scale statistical regularity, thereby mitigating the scale effect of the modifiable areal unit problem. Second, an optimal parameter-based geographical detector (OPGD) model was used to identify city-specific combinations of morphology factors associated with heterogeneity in CO₂ emissions to construct LCEZs. Eleven factors passed the collapse screening, including transport, building, and landscape factors. In the constructed LCEZs results of the four cities, at least one transportation-related variable was retained in every combination, while mean building volume (MBV) emerged as a core factor in three of the four metropolises, demonstrating high universality and influence. The resulting LCEZs were statistically validated by Kruskal-Wallis tests, with effect sizes ranging from 0.143 to 0.233, and intra-zone coefficients of variation were reduced by over 80% relative to the global baseline in each metropolis, confirming strong internal homogeneity. This framework provided an basis for comparing morphology–emission associations and informing low-carbon interventions. It offered a repeatable diagnostic procedure to quantify trade-offs and tailor measures to local contexts, bridging computational urban science with planning practice.