Development of an urban carbon emission accounting and dynamic warning framework using multi-source cross-domain big data: A case study of municipalities, China

Cities play a pivotal role in reducing carbon emissions and combating climate change, making them essential for achieving dual-carbon goals. High-frequency and reliable carbon accounting is the foundation for enabling government agencies to reduce emissions efficiently and reach carbon peaks. However, traditional carbon accounting methods are hindered by their poor timeliness, low accuracy, and coarse granularity. To overcome these challenges, this study proposed a high-frequency, high-precision, and traceable city-level carbon accounting framework. This framework enhanced the accuracy and real-time performance of carbon accounting while incorporating a dynamic early warning system for urban carbon emissions, strengthening government emergency response capabilities, and ensuring informed decision-making for low-carbon urban development. The effectiveness of the model was demonstrated through a case study using data from four municipalities in China. Our principal findings were as follows: (1) The proposed framework integrated multi-source, cross-domain big data, significantly improving carbon accounting frequency and precision, enhancing early warning accuracy and timeliness, and ensuring data traceability. (2) The proposed ResTCN surpassed traditional methods, achieving a mean absolute percentage error (MAPE) of less than 4.5 %. (3) The dynamic carbon emission early warning method achieved a MAPE of less than 2 % in carbon emission forecasting. These results demonstrated that the framework not only enhanced data traceability but also improved the accuracy and timeliness of carbon accounting and early warnings, providing a robust foundation for cities to develop targeted carbon reduction policies and accelerate the transition to low-carbon cities.