Dataset of City-level CO2 emissions in China during 1992-2023

Based on current CO2 emission accounting methods and results, this study employed nighttime lights (NTL) remote sensing data, coupled with the XGBoost model and a multi-task learning model, to develop a data alignment technique for carbon accounting by integrating correction, supplementation, and prediction functionalities. As a result, a dataset of CO2 emissions covering 329 cities (including 4 municipalities and 325 prefecture-level cities) from 1992 to 2023 was compiled.基于当前二氧化碳（CO2）排放核算方法与核算结果，本研究采用夜间灯光（NTL）遥感数据，结合XGBoost模型与多任务学习模型，开发了集成校正、补全与预测功能的碳核算数据对齐技术。最终构建了1992年至2023年覆盖329座城市（含4个直辖市与325个地级市）的CO2排放数据集。