Deriving Full‐Coverage and Fine‐Scale XCO₂ Across China Based on OCO‐2 Satellite Retrievals and CarbonTracker Output

Abstract

Carbon dioxide (CO 2 ) is the most abundant greenhouse gas in the atmosphere and is considered one of the main contributors to climate change (Kump, 2000). The radiative forcing attributable to CO 2 accounts for approximately 56% of the total greenhouse warming effect (Soden et al., 2018). Intensified anthropogenic emissions from sources such as fossil fuel combustion caused a rapid increase in atmospheric CO 2 from 388.76 in 2010 to 412.44 ppm in 2020 (Dlugokencky & Tans, 2021). Many countries or regions, including China, the United States, the European Union, and Japan, have set the goal of realizing carbon neutrality by promoting renewable energy sources, carbon capture and storage, and forest recovery (Corbin & Holl, 2012;Gil & Bernardo, 2020). According to the Publicly Available Specification (PAS) 2060, "carbon neutral" describes a situation of no net greenhouse gas emissions (Finkbeiner & Bach, 2021). As the largest CO 2 emitter, China accounted for 29.3% of total global emissions in 2017 and has announced two goals of achieving the peak of CO 2 emissions by 2030 and carbon neutrality by 2060 (Muntean et al., 2018).Comprehensive information on the spatiotemporal distributions of CO 2 provides a fundamental basis for investigating the carbon source-sink dynamics and the corresponding impact on climate change (Deng et al., 2011). Site-based observations obtained from surface observation networks, such as the Total Carbon Column Observing Network (TCCON), are insufficient for delineating the broad distributions of CO 2 (Rayner & O'Brien, 2001). Process-based models are commonly employed to simulate carbon source-sink dynamics (Gurney et al., 2002).

Seasonal and Diurnal Variations in XCO2 Characteristics in China as Observed by OCO‐2/3 Satellites: Effects of Land Cover and Local Meteorology

“…Consistent patterns of seasonal variation in satellite‐derived XCO 2 observed by OCO‐2 and OCO‐3 over China are evident in Figures 3a and 3b, respectively, illustrating that XCO 2 concentration was higher in winter and spring, followed by autumn, and lowest in summer (Table S1 in Supporting Information ), and that it exhibited apparent spatial heterogeneity with a gradient of increase from the west toward the east in every season, consistent with the findings of previous studies based on GOSAT data (Bie et al., 2020; Lv et al., 2020), a weather–biosphere model (Dong et al., 2021), and machine learning models (C. He et al., 2022a; M. Zhang & Liu, 2023). Specifically, large‐scale high concentrations of XCO 2 over the Northern China Plain (NCP) and the Yangtze River Delta (YRD) during winter were measured by both OCO satellites, whereas OCO‐3 additionally captured high concentrations of XCO 2 in central China that were missing in the OCO‐2 data, and it also filled the gaps in OCO‐2 observations over Northeast China (NEC).…”

“…Consistent patterns of seasonal variation in satellite-derived XCO 2 observed by OCO-2 and OCO-3 over China are evident in Figures 3a and 3b, respectively, illustrating that XCO 2 concentration was higher in winter and spring, followed by autumn, and lowest in summer (Table S1 in Supporting Information S1), and that it exhibited apparent spatial heterogeneity with a gradient of increase from the west toward the east in every season, consistent with the findings of previous studies based on GOSAT data (Bie et al, 2020;Lv et al, 2020), a weatherbiosphere model (Dong et al, 2021), and machine learning models (C. He et al, 2022a;M. Zhang & Liu, 2023).…”

Improved Consistency of Satellite XCO2 Retrievals Based on Machine Learning

“…High‐quality observations requiring strict cloud filtering only account for a small fraction of total observations (Taylor et al., 2022) and have seasonal variability, leading to issues of spatial representation and uncertainty in flux inversion estimates (Houweling et al., 2015). Recent studies have utilized machine learning (ML) models to derive global full‐coverage XCO 2 maps based on data from GOSAT, OCO‐2, or both (He et al., 2022; Siabi et al., 2019; Wang et al., 2023; Zhang & Liu, 2023; Zhang et al., 2022), facilitating analysis of carbon dynamics (He et al., 2022). However, these seamless data sets lack sufficient accuracy and do not provide uncertainty estimates used for observational inputs in carbon flux inversion.…”

XCO2 Super-Resolution Reconstruction Based on Spatial Extreme Random Trees

“…For instance, parameter tuning in support vector machines significantly affects model performance and requires careful optimization in experiments, while artificial neural network models typically require a large amount of sample data for training. In the downscaling of carbon dioxide concentration, machine learning methods have also been widely applied, and scholars have conducted extensive research to enhance predictive performance [46][47][48]. Methods such as random forest, XG-Boost, and gradient-boosting decision trees (GBDTs) have shown excellent applications in the high-resolution reconstruction of carbon dioxide.…”