Spatiotemporal pattern and driving factors of atmospheric CO2 concentrations based on satellite remote sensing from 2001 to 2022 in central Yunnan plateau

• A breakthrough in remote sensing allows long-term and pixel-scale monitoring of near-surface CO 2 concentrations in CYP. • Multi-model coupling analysis deepens the GTWR-GD-RF framework considering complex driving mechanisms. • Predictive modelling integrates urban spatial configuration, natural, and socioeconomic factors to foresee key indicators. • Policy transformation innovation of the carbon efficiency via constructing an index to measure the ecological cost. CO 2 is one of the primary greenhouse gases. Dynamic monitoring CO 2 concentrations and quantitative determination of the contributions of its driving factors are crucial for achieving carbon peak and carbon neutrality. Although previous studies have identified changes in atmospheric CO 2 concentrations, they are limited by spatiotemporal resolution, preventing a detailed observation of regional variations and trends. Moreover, the analytical methods focus on using linear regression models exploring a single type of factor and limiting the ability to gain a deeper understanding. Therefore, this study constructs a comprehensive methodological framework by taking the Central Yunnan Plateau (CYP) in southwest China as an example. Based on multi-source remote sensing data, CO 2 concentrations was estimated at pixel scale, and the spatiotemporal dynamics were monitored by using Sen’s slope estimation combined with Mann-Kendall test (SEN-MK) and unitary linear regression. Using Geographically and Temporally Weighted Regression (GTWR) and Geographical detector (GD), a prediction model was established to explore the complex relationship between 14 driving factors of 3 different types (including urban spatial configuration, natural factors and socioeconomic factors) and the annual CO 2 concentrations. Random Forest (RF) algorithm was used to gain insight into the nonlinear effects of these different indicators on near-formation atmospheric CO 2 concentrations. And the carbon benefit index was further constructed to measure the ecological cost caused by economic development. The findings showed that: (1) The CO 2 concentrations of CYP was a turning point in 2006, showing a significant increase, and there is a downward trend in recent years. Carbon benefit showed an obvious upward trend. CO 2 concentrations generally increased from southwest to northeast. (2) The contributions of driving factors demonstrate significantly spatiotemporal heterogeneity. The reduction of elevation and SHAPE_MN, and the increase of PAFRAC could lower the CO 2 concentrations in central and southwestern part of CYP. The increase of air temperature and elevation could contribute the high CO 2 concentrations of the northeastern part of CYP. (3) Elevation and air temperature are the most important centre of interaction. Topography can compel the form of construction land and meteorological conditions, and furtherly affect social and economic activities, especially in the eastern part of the Yuxi. Overall, the study unveils comprehensively the complicated interactions between driving factors and CO 2 concentrations. The findings offer a refined and differentiated reference for policymakers in assessing the current state of the urban construction in CYP, optimizing urban layout, and developing low-carbon cities. Additionally, they serve as a case for other hilly and mountainous cities to conduct remote sensing monitoring of CO 2.