Resolving the carbon sink from global carbonate weathering and its environmental controls using a global synthesis of rock tablet data and machine learning

Abstract

[EN]Chemical weathering of carbonate minerals (CWC) plays an important role in global carbon cycle, as it bridges the atmospheric, lithospheric and hydrospheric carbon pools. However, data limitations have hindered an accurate estimation of the carbon sink flux induced by global CWC (CSFCWC), as well as its response to environmental change. Conventional hydrochemical methods, which infer CSFCWC indirectly from riverine hydrochemistry, provide only catchment-integrated signals, yet cannot resolve the specific contribution of CWC across different climate, pedological, and ecological settings within a catchment. Here, we synthesize 2444 globally-distributed in-situ CWC rates measured by rock tablet tests, and investigate the magnitude, spatiotemporal variation and controlling factors of CSFCWC using a machine learning model. We find that soil physicochemical properties (e.g. pH and moisture) play a more important role in determining global CWC spatial variation than climate, hydrology and vegetation factors. The machine-learning model developed in this study explains 68% of the variance in globally observed CSFCWC values. Global application of our model indicates that CWC generates a carbon sink of 0.27 Pg C yr−1 worldwide, which is comparable to previous catchment-integrated estimates derived using different approaches, accounting for approximately 8% of the total terrestrial carbon sink. Over the past two decades, global greening has significantly accelerated global total carbon sink induced by CWC, with this acceleration particularly pronounced in Asia. Overall, this study provides a benchmark estimate of global CSFCWC and advances the mechanistic understanding of carbonate weathering. Our findings contribute to improving existing weathering models and reducing uncertainties in future projections of the terrestrial carbon sink.