Livable Cities - London AMPS | City, University of London Page 110 Figure 1. Study Area in and around CCEC. Estimating CO2 Emissions in Chengdu-Chongqing Economic Circle In this study, PBA is used to account for carbon emissions from energy consumption in CCEC cities. Comprehensive energy consumption includes different types of coal, coke, gas, fuel, oil, heat and electricity, etc. In China, energy is measured in ‘Standard Coal,’ whose calorific value is 29,307 kilojoules per kilogram. As such, based on the methodology provided by IPCC, the formula for calculating carbon dioxide emissions from comprehensive energy consumption in units of standard coal is: Cenergy = E × NCV ×CEF ×COF (1) Where Cenergy represents the carbon emission of standard coal, E denotes the consumption amount of standard coal, NCV denotes the average net calorific value of standard coal, CEF is the carbon emission factor of calorific value unit of standard coal, and COF is the carbon oxidation factor of standard coal, with a default value of 1. The values of each parameter are derived from the General Rules for Calculating Comprehensive Energy Consumption (GB/T 2589-2008) and the 2006 IPCC Guidelines for National Greenhouse Gas Inventories. The Geographically Weighted Regression Model and Spatial Autocorrelation GWR is a method used for local linear regression that considers spatially varying relationships. What should be noticed is that, it focuses on observations close to a specific point and gives more weight to nearby data than to data farther away. The model is represented by the following equation: yi = β0 (ui,vi) + ∑j βj (ui,vi) xij + εi (2) where i = 1, 2, … n, (ui,vi) represents the spatial coordinate of the ith observation point; β0 (ui,vi) is the local intercept of the ith observation point; βj (ui,vi) is the parameter of the jth independent variable xij; εi is the random error term. Besides, GWR model also embeds spatial autocorrelation, which refers to a certain degree of spatial interaction between geographic phenomena or an attribute value among spatial units. Moran's I is divided into global autocorrelation and local autocorrelation. Global autocorrelation identifies and measures the spatial pattern of the whole area: Moran’s I = ∑ni=1 ∑nj=1 Wij (Yi-Y) (Yj-Y) / S2 ∑ni=1 ∑nj=1 Wij (3) where S2 =∑ni=1 (Yi-Y) / n, Y = ∑ni=1 Yi / n, Yi denotes the observation value of the ith region (e.g., urban carbon emission in this study), n is the total number of regions, and Wij is the binary proximity spatial weight matrix, which is used to define the proximity of the spatial objects to each other. The value of the global Moran’s I ranges from -1 to 1. If its value is greater than 0, the larger the value is, the stronger the positive autocorrelation of the spatial distribution is. When its value is less than 0, it indicates differences. if its value
