Global riverine land-to-ocean carbon export constrained by observations and multi-model assessment

Abstract

Rivers are a key component of the global carbon cycle. They receive vast quantities of terrestrial carbon, of which a large fraction is ultimately exported to the coastal ocean. Our review of previously published assessments reveals that substantial uncertainties remain with regard to the spatial distribution and speciation of the carbon export. Accurate quantification of the relative contributions of dissolved, particulate, organic and inorganic carbon to the total amounts is, however, of crucial importance for the coupling between the terrestrial and marine carbon cycles. Breaking down existing spatially explicit assessments over large river basins, we find a disagreement in flux estimates that exceeds two orders of magnitude for more than half of the basins. Using machine-learning techniques in combination with a multi-model ensemble and an updated database of observations, we overcome the inconsistencies in existing assessments and narrow down uncertainties in riverine carbon exports. Our revised assessment yields a global riverine export of 1.02 ± 0.22 (2σ) PgC yr−1. This carbon flux is partitioned into 0.52 ± 0.17, 0.30 ± 0.14, 0.18 ± 0.04 and 0.03 ± 0.02 PgC yr−1 of dissolved inorganic, dissolved organic, particulate organic and particulate inorganic carbon, respectively. We estimate the carbon contribution through groundwater export to be minor (0.016 PgC yr−1). Our assessment suggests an underestimation of the land-to-ocean carbon flux by 0.24 PgC yr−1 by the Intergovernmental Panel on Climate Change (IPCC) and calls for a revision of the oceanic carbon budget. Global riverine carbon export to the ocean may exceed previous estimates, underscoring the important role of riverine carbon to the carbon budget, according to a multi-model ensemble assessment constrained by a global dataset.