High resolution spatial distribution of CO₂ emissions from building operations and driving factors in the Guangdong-Hong Kong-Macao Greater Bay Area

Abstract

Building operation is an important source of greenhouse gas emissions. As China's most densely populated region, the Guangdong-Hong Kong-Macao Greater Bay Area (GBA) contributes 22% of national building sector CO₂ emissions, but systematic assessment of the Building Operational CO₂ Emissions(BOCE) characteristics remain scarce. Clarifying these patterns and their drivers is essential for achieving regional low-carbon development goals. This study integrates top-down and bottom-up approaches to construct a 1 km × 1 km gridded CO₂ emission inventory for BOCE in the GBA, employing multi-scale geographically weighted regression (MGWR) to analyze the spatial heterogeneity of its driving factors. The results show that: (1) From 2015 to 2020, the total CO₂ emissions from the building operation phase in the GBA increased by 27.54 % (from 141 to 180 million tons). (2) Although high-emission areas occupy only a small proportion of urban land, they contribute a disproportionately large share of BOCE. High-emission clusters in urban cores of Guangzhou and Shenzhen, accounting for 64 % of municipal totals, significantly surpassing suburban levels. (3) Key drivers of BOCE ranked by influence in 2015: tertiary industry GDP (GDP3) > population (POP) > Normalized Difference Vegetation Index (NDVI) > per capita disposable income (IN). Compared with 2015, the impact of GDP3 declined (coefficient decreased from 0.857 to 0.213) in 2020, while POP's influence strengthened (coefficient rose from 0.547 to 0.751). There is spatial heterogeneity in the impact of different drivers, the impact of POP and IN exhibited a “west-strong-east-weak” spatial pattern, but the areas most affected by POP shifted eastward in 2020. These findings provide a scientific basis for formulating region-specific decarbonization policies and fostering cross-sectoral collaboration in the GBA's building sector.