Evolution Characteristics and Trend Prediction of Farmland Soil Organic Carbon in Shanxi, China

View this project on my Github: Link Here

Soil organic carbon is an important component of the global carbon sink and plays a crucial role in maximizing soil carbon sequestration capacity and reducing soil respiration, thereby positively impacting carbon emission reduction. In line with sustainable development goals, Shanxi Province can achieve atmospheric carbon neutrality by limiting industrial carbon emissions and enhancing carbon sequestration technologies, such as promoting soil carbon sequestration. Understanding the spatiotemporal variations and future prospects of soil organic matter in Shanxi Province is significant for agricultural production and carbon sequestration in farmland. This study compiled national second general survey of soil resources data (1982), literature data on land fertility evaluation and utilization in various counties and cities of Shanxi Province (2012), and field measurements (2022). The data were combined with climate, topography and vegetation data from Google Earth Engine and the National Tibetan Plateau Data Center. Statistical analysis, geostatistics, Pearson correlation analysis, and machine learning models were employed to analyze the spatiotemporal patterns and predictive prospects of soil organic matter in farmland in Shanxi Province. The research findings are as follows: (1)Descriptive statistics revealed a gradual increase in soil organic matter content in farmland of Shanxi Province from 1982 to 2022. Geostatistical analysis showed moderate spatial correlation of organic matter in different regions of Shanxi Province in 1982, while strong spatial correlation was observed in 2012 and 2022. Kriging interpolation maps displayed an increasing trend of organic matter content from west to east in 1982, and from northwest to southeast in 2012 and 2022. (2)Pearson correlation analysis indicated a weak linear negative correlation between organic matter content and pH value, while a strong linear positive correlation was observed with total nitrogen and alkaline hydrolyzable nitrogen. No significant linear relationships were found between organic matter content and topography, climate, and vegetation factors. There was a strong correlation between organic matter content and anthropogenic fertilizer input. (3)By establishing linear regression, support vector machine, random forest, XGBoost, and LightGBM models, the relationships between the aforementioned factors and organic matter content were studied The results showed that LightGBM and XGBoost models performed the best in the prediction scenario. These two models were used to predict the organic matter content in farmland soil in Shanxi Province under certain scenarios, yielding predicted values of 20.0 g/kg and 19.6 g/kg, respectively. Through the three-step process of descriptive statistics, identification of key influencing factors, and model-based future prediction, this study provides in-depth insights into the spatiotemporal variations and underlying mechanisms of soil organic matter in farmland in Shanxi Province, along with feasible predictions of future organic matter content. This research is expected to provide valuable references for farmland management and carbon sequestration measures in Shanxi Province.