IPCC最新评估报告表明，当前全球气候系统处于持续变暖阶段。如何解决二氧化碳等温室气体所导致的气候变暖问题是全球及国家治理的重要组成部分。目前，中国的碳排放量处于显著上升阶段，根据国际能源署(IEA)在2021年世界能源报告中公布的能源数据来看，中国的能源消费增量占世界增长量的四分之三，碳排放量占全球的三分之一，位居世界第一。

中国政府积极采取行动应对全球气候变暖问题，并在国家自主贡献中承诺，2030年左右碳排放总量将达到峰值，并争取在2060年实现碳中和的愿景。面对前所未有的减排压力，探索适合我国现阶段发展需要的节能减排路径意义重大。

首先，本文分析了现阶段中国能源消费现状，通过描述性统计方法描绘中国能源消费特征。其次，通过碳排放系数法核算了2003—2020年中国及东部、中部和西部地区能源消费碳排放总量，并构建了相应的STIRPAT模型，深入探究中国碳排放的经济社会驱动因素。该模型将影响因素分解为人口、技术、经济、能源、产业和社会因素6个方面，并选取这6个方面的11个变量作为初始自变量，然后通过变量筛选方法（Lasso，Elastic-Net）选出重要影响因素，并基于选出的重要影响因素构建最终的“扩展-STIRPAT”模型。研究结果显示，造成碳排放增加的主要原因有人口规模扩大、城市化水平提高和人均GDP增长。相反，调整产业结构、降低能源强度和优化能源结构将有助于减少碳排放。

进一步，本文构建了基于GRU神经网络的中国及各地区能源消费碳排放预测模型，结合情景分析法设置影响因素在高碳、基准和低碳情境下的变化情况，预测了三种情景下2021—2040年全国及东部、中部和西部地区碳排放量变化趋势。结果表明，尽管经济持续增长，各地区能够在至少一种情景下实现碳达峰，考虑到累积碳排放量因素，最早达峰的情景可能并非是最优选择。本文从地区层面出发，寻找全国最优排放路径。结果显示，中国将在2024~2032年实现碳达峰，峰值范围在12.1~12.5Gt CO2，累积碳排放量范围为237.0~248.5Gt CO2。

本文的研究意义在于揭示了中国及各地区的潜在达峰路径，为实现碳达峰目标以及合理制定中长期应对气候变化策略提供理论依据。

The latest assessment report of IPCC shows that the concentration of greenhouse gases such as carbon dioxide in the atmosphere is currently in a stage of continuous and significant rise, and the global climate system will continue to warm. How to solve the problem of increasing carbon emissions and global warming caused by fossil energy consumption is an important part of global and national governance. With the rapid development of the economy, China's carbon emissions are increasing. According to the data released by IEA in the 2021 World Energy Report, China's energy consumption growth accounts for three-quarters of the world's growth, coal consumption accounts for 57.7% of total energy consumption, and carbon emissions account for one-third of global emissions, ranking first in the world. China's carbon emission has come under global spotlight.

Considering the seriousness and urgency of global warming, the Chinese government implements a national strategy of actively responding to climate change. In September 2020, China pledged in its INDCs that China's carbon emission intensity in 2030 would be reduced by 60%-65% compared with 2005, and that total carbon emissions would peak around 2030, with the vision of achieving carbon neutrality in 2060. In the face of unprecedented pressure to reduce emissions, it is of great significance to explore energy-saving and emission-reduction paths that suit China's current development needs.

Firstly, this paper analyzes the current situation of China's energy consumption at this stage, and depicts the characteristics of China's energy consumption through descriptive statistical methods. Secondly, the total carbon emissions of energy consumption in China and the eastern, central and western regions from 2003 to 2020 were calculated by the carbon emission coefficient method, and the corresponding STIRPAT model was constructed to deeply explore the economic and social drivers of China's carbon emissions. The model decomposes the factors affecting China's carbon emissions into six aspects: population factors, economic factors, technical factors, industrial factors, energy factors and social factors, and selects 11 variables in these six aspects as the initial independent variables, and then screens the initial variables through Lasso and other methods, and constructs the final "extended-STIRPAT" influencing factor model based on the variable screening results. The results show that population growth, urbanization and per capita GDP growth will all lead to an increase in carbon emissions. On the contrary, reducing energy intensity and adjusting the industrial structure and energy structure will help reduce carbon emissions.

In this paper, carbon emission prediction models based on GRU neural network for China and its provinces are constructed. Combined with the scenario analysis method, the changes of each influencing factor under high-carbon, benchmark and low-carbon scenarios are reasonably set, and the carbon emissions of the whole country and eastern, central, and western regions under different scenarios from 2021 to 2040 are predicted. We observe that different regions are highly likely to reach peak emissions under at least one scenario. Given the cumulative amount of carbon emissions, the earliest peak scenario may not be optimal. This paper looks for the optimal emission path in China from the regional level. The total estimation results show that China will achieve peak emissions in 2024~2032, with a peak range of 12.1~12.5Gt CO2 and a cumulative carbon emission range of 237.0~248.5Gt CO2.

This paper reveals the potential peaking pathways in China and other regions from a bottom-up perspective, providing an important reference for China to achieve the goal of carbon peaking and reasonably formulate medium- and long-term climate change strategies.