Machine Learning Model-Based Estimation of XCO2 with High Spatiotemporal Resolution in China

Abstract

As the most abundant greenhouse gas in the atmosphere, CO2 has a significant impact on climate change. Therefore, the determination of the temporal and spatial distribution of CO2 is of great significance in climate research. However, existing CO2 monitoring methods have great limitations, and it is difficult to obtain large-scale monitoring data with high spatial resolution, thus limiting the effective monitoring of carbon sources and sinks. To obtain complete Chinese daily-scale CO2 information, we used OCO-2 XCO2 data, Carbon Tracker XCO2 data, and multivariate geographic data to build a model training data set, which was then combined with various machine learning models including Random Forest, Extreme Random Forest, XGBoost, LightGBM, and CatBoost. The results indicated that the Random Forest model presented the best performance, with a cross-validation R2 of 0.878 and RMSE of 1.123 ppm. According to the final estimation results, in terms of spatial distribution, the highest multi-year average RF XCO2 value was in East China (406.94 ± 0.65 ppm), while the lowest was in Northwest China (405.56 ± 1.43 ppm). In terms of time, from 2016 to 2018, the annual XCO2 in China continued to increase, but the growth rate showed a downward trend. In terms of seasonal effects, the multi-year average XCO2 was highest in spring (407.76 ± 1.72 ppm) and lowest in summer (403.15 ± 3.36ppm). Compared with the Carbon-Tracker data, the XCO2 data set constructed in this study showed more detailed spatial changes, thus, can be effectively used to identify potentially important carbon sources and sinks.

An Interpolation and Prediction Algorithm for XCO2 Based on Multi-Source Time Series Data

“…He et al validated XCO 2 results generated by random forest against ground-based data, with an RMSE of 1.123 ppm. These results are consistent with our analysis, further supporting the reliability and validity of our findings [45]. The error of the validation results is depicted in Figure 4.…”

“…Li [43] 1.71 ppm --Zhang [44] 1.18 ppm 0.99 ppm −0.6 He [45] 1.123 ppm --Ours 1.01 ppm 0.75 0.4 The prediction of CO 2 concentration requires a parameterized model, with each parameter or variable having different scales in the dataset. To prevent parameters with large value ranges from exerting excessive influence, feature normalization is performed to scale all features equally.…”

“…In addition, this study estimates XCO 2 data using environmental variables, but did not incorporate anthropogenic factors into the modeling process. Existing research has not adequately addressed this point [43][44][45], and in the future, incorporating social science factors into the model may improve our estimation accuracy. 2.…”

High-Spatiotemporal-Resolution Estimation of Ground-Level Ozone in China Based on Machine Learning

“…Regarding the spatial distribution, we identify high-value areas of O 3 concentrations in eastern coastal regions such as Shandong, Jiangsu, and Zhejiang provinces, where the values concentrated around 210 µg/m 3 . High-density regions characterized by substantial industrial, transportation, and residential emissions are responsible for elevated levels of O 3 precursors, including NOx and VOCs [44,45]. Moreover, the culmination of summer characterized by elevated temperatures and increased thunderstorms intensified the increase in ground-level O 3 levels.…”

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.…”