Livable Cities - London AMPS | City, University of London Page 109 Spatial Correlation and Differences The spatial dimension stands as an important trait of cities and an underlying factor influencing carbon emissions. Commonly used spatial measurement models include the Ordinary Least Squares regression model (OLS), Geographically Weighted Regression model (GWR), and more generalized Multi-scale Geographically Weighted Regression model (MGWR). To put it more specific, OLS is a linear non-spatial regression model without introducing spatial distance weights, while GWR acts as an improvement. In addition, before using spatial measurement models, spatial autocorrelation needs to be tested through Moran's I index method. Wang, S.J. et al. found the evolution of carbon emission in Chinese cities has obvious spatial effects, with neighboring cities influencing each other. 7 Socioeconomic Factors and Impacts Three theoretical models are mainly used to study socio-economic factors affecting urban carbon dioxide emissions, namely Kaya, Stochastic Impacts by Regression on Population, Affluence, and Technology (STIRPAT) and the input-output model, respectively. 8 Among the many exponential decomposition analysis (IDA) methods derived from Kaya model, the Log Mean Dichotomous Index Method (LMDI) is recognized as the best for it has no residuals and effectiveness in preventing pseudo-regression problems. This method breaks down carbon emission drivers into the following factors: population size, economic development, industrial structure, energy intensity and energy structure. 9 Relevant studies have found that economic development has a contributing effect on carbon emissions, while the energy structure and population urbanization show an inverted U-shaped relationship with carbon emissions. In addition, urbanization of public services helps reduce carbon emissions. 10 Research Significance and Innovation In summary, although there has been a certain degree of research depth and results, research gaps still exist: (1) Due to incomplete energy consumption data in Chinese cities, most research focuses on national or provincial levels rather than city-level emissions; (2) Underdeveloped cities in the western region are often left behind, as current studies tend to focus on major developed cities in the Pearl River Delta and the Pan- Yangtze River Delta; (3) Existing studies mostly conclude spatial autocorrelation of the whole region, lack explanation of local spatial relationship and spatial heterogeneity. To address these gaps, this paper first estimates the urban carbon emissions of 18 cities in the CCEC from 2012 to 2021 using the PBA methodology. Secondly, it employs OLS regression model to identify overall drivers of city-level carbon emissions, and then GWR model to provide a separate parameter set for local drivers. Last but not least, decarbonization strategies for CCEC cities are proposed. MATERIALS AND METHODS Case Choice The Chengdu-Chongqing Economic Circle (CCEC) is located in western China and acts as one of the main channels connecting international and domestic markets. It covers 16 major cities of Chongqing and Sichuan province, with a total area of 185,000 square kilometers and a total population of over 98 million. In 2021, GDP of this area reached 7391 billion, accounting for 6.5% of the national gross domestic product.11 In this paper, 18 cities in and around the CCEC are defined as the study area, as is shown in Figure 1.
