The organic carbon (OC) stored in global topsoil (0-30 cm) will be the most active participant in the carbon cycle under future climate change. Due to differences in focus regions or research methods, the spatio-temporal changes of future global topsoil OC stocks and how they will be affected by climate change are not systematically understood, which needs to be further explored. In this study, we developed data-driven models to predict the spatio-temporal dynamics of global topsoil OC stocks by combining 32,579 soil profiles with environmental variables and comprehensively explored the impact of future climate change on topsoil OC. It was found that the topsoil OC stocks were 1249.29 Pg in the baseline period (1971-1990). By 2100, under the normal and high representative concentration paths, it is predicted that the global topsoil OC stocks will decrease by 113.67 ± 25.93 Pg and 193.71 ± 39.76 Pg, respectively. In the future, the largest global topsoil carbon loss will occur in boreal forest areas, which are expected to lose 17.03-27.90 % (66.01-108.13Pg) of their carbon stocks. The influence of climate on topsoil OC stocks is mainly manifested in temperature, which has a negative influence on the global topsoil OC stock, and the contribution rate of temperature to the effect on the global topsoil OC stock is about 26.96 %. Overall, our results provide a high spatio-temporal resolution assessment of global topsoil OC stocks and their relationship to environmental factors, and highlight the spatial heterogeneity, which has been generally ignored in many experimental frameworks and prediction models. These results will help governments to make appropriate management decisions to mitigate climate change.
Keywords: Climate change; Machine learning; Topsoil organic carbon.
Copyright © 2023 Elsevier B.V. All rights reserved.