中国碳排放区域分布特征、预测及差异化政策建议

Regional Distribution Characteristics, Prediction and Differentiated Policy Recommendations Policies for China’s Carbon Emissions

摘要: 全球气候变暖背景下，中国“双碳”目标(2030年前碳达峰、2060年前碳中和)的实现需精准把握碳排放区域特征与未来趋势。针对中国碳排放“高维、非线性、时空异质”的核心特征，以及现有研究方法单一、区域分类粗放、未充分解构时序特征的缺口，本文构建HP-LSTM模型，基于CEADs数据库2000~2020年30个省份47个行业的碳排放数据，通过HP滤波分解排放序列的趋势与周期成分，结合LSTM模型实现精准预测，并与ARIMA模型进行对比验证。研究发现，30个省份可划分为排放顶峰型、趋势增长缓慢型和持续增长型三类，部分省份难以在2030年前达峰；电力、蒸汽和热水生产供应业是碳排放主要贡献来源。基于预测结果，本文提出差异化治理策略与动态监测机制的政策启示，为中国“双碳”目标落地提供科学支撑。

Abstract: Against the backdrop of global warming, the achievement of China’s “dual carbon” goals (peaking carbon emissions before 2030 and achieving carbon neutrality before 2060) requires an accurate grasp of the regional characteristics and future trends of carbon emissions. Addressing the core characteristics of China’s carbon emissions—high-dimensionality, non-linearity, and spatial-temporal heterogeneity—as well as the gaps in existing research (e.g., single research methods, crude regional classification, and insufficient deconstruction of time-series characteristics), this study constructs an HP-LSTM model. Based on carbon emission data from 47 industries across 30 provinces in China (2000~2020) from the CEADs database, the model decomposes the trend and cyclical components of emission sequences through HP filtering, and combines the LSTM model to achieve accurate prediction, with comparative verification against the ARIMA model. The results show that the 30 provinces can be classified into three types: emission peak-reached type, slow trend growth type, and continuous growth type, with some provinces struggling to peak before 2030. The production and supply of electricity, steam, and hot water is the main contributor to carbon emissions. Based on the prediction results, this study proposes policy implications including differentiated governance strategies and dynamic monitoring mechanisms, providing scientific support for the implementation of China’s “dual carbon” goals.