Abstract: Carbon emissions of 30 provinces, metropolis or autonomous regions (except Hong Kong, Macao, Taiwan and Tibet) of China from 1995 to 2010 were calculated using the IPCC method. Based on total carbon emission, carbon emission intensity and carbon emission per capita, the three indices, the 30 provinces, metropolis and autonomous regions were sorted into 3 categories or grades, i.e. high, moderate and low in carbon emission, through cluster analysis of multivariable panel data. In Grade High, there were Hebei, Shanxi, Inner Mongolia, Liaoning, Shandong, Ningxia, in Grade Low,Beijing, Zhejiang, Anhui, Fujian, Jiangxi, Hubei, Hunan, Guangdong, Guangxi, Hainan, Chongqing, Sichuan, Yunnan, Shaanxi, Qinghai, and in Grade Moderate, Tianjin, Jilin, Heilongjiang, Shanghai, Jiangsu, Henan, Guizhou, Gansu, Xinjiang. Besides, cointegration test and regression analysis was performed of carbon emissions per capita and GDP per capita of the three groups of provinces. Results show that a co-integration relationship had long been existing between carbon emissions per capita and GDP per capita and fit the environmental Kuznets inverse U curve, and the knee of the theoretical curve of the three groups, high, moderate and low was 41 046, 50 219 and 47 049 yuan (RMB) in GDP per capita, respectively. It is predicted that the total carbon emission of the country will keep go on increasing for a long time, however, with a growth rate of carbon emission lower than that of the GDP, which means the carbon emission intensity will keep on declining.

Key words: carbon emissions, regional difference, environmental Kuznets curve, China