Unveiling the Dynamic Patterns and Driving Forces of Soil Organic Carbon in Chinese Croplands From 1980 to 2020
Junchen Ai, Zipeng Zhang, Yang Cheng-lin, Jinhua Cao, Zhiran Zhou, Xiangyu Ge, Xiangyue Chen, Jingzhe Wang
Abstract
ABSTRACT Soil organic carbon (SOC) in cropland is a critical component of the global carbon cycle, representing the most dynamic segment of the carbon pool, and is vital to addressing both “dual‐carbon” goals and food security challenges. However, the current research on SOC in China's croplands has limitations in timeliness, continuity, and accuracy. This study constructed a machine learning model to assess the spatial–temporal distribution and changes of cropland SOC across China. It maps the annual distribution of cropland SOC in China over the past four decades (1980–2020), leveraging data from 2399 cropland sampling points collected from the second soil census of China and the integration of multi‐platforms combined with 22 environmental excoriates. The model's accuracy ( r = 0.82) could meet the needs of the analysis and perform reliably in predicting cropland SOC across China, with high uncertainty only in some areas, such as the northeast. The study reveals that while there have been fluctuations in SOC stocks in China's croplands over the years, the overall trend has been upward, increasing at a rate of 0.012 Pg C y −1 , and generally functions as carbon sinks. Furthermore, the Shapley additive explanations indicate that temperature strongly correlates with SOC in croplands, followed by precipitation and topography. The outcomes of this research provide essential data support for formulating policies on cropland protection, land degradation, and carbon peak strategies in China.