The Increasing Impact of Seasonality Biases on Model‐Based Estimates of the Ocean Carbon Sink

Abstract Ocean carbon cycle syntheses combine observation‐based surface ocean p CO 2 products and global ocean biogeochemical models (GOBMs) to estimate air‐sea CO 2 fluxes. These efforts show that GOBMs underestimate the ocean's anthropogenic CO 2 sink compared to observation‐based products, though the causes remain uncertain. Observations and models indicate that the seasonal amplitude of surface p CO 2 is increasing due to anthropogenic carbon (C ant ) accumulation, potentially altering net annual CO 2 fluxes. Here, we test the hypothesis that systematic biases in model representations of surface p CO 2 seasonality contribute to long‐term biases in annual ocean carbon uptake trends. Using simulations from seven GOBMs, we evaluate changes in surface p CO 2 seasonal amplitude over 1980–2018, separating contributions from C ant accumulation and climate‐state changes. Within each model, seasonal p CO 2 amplitude growth is primarily caused by C ant ‐driven enhancement in p CO 2 sensitivity to seasonal thermal and biophysical processes. Across models, differences in C ant ‐driven p CO 2 amplification are closely linked to differences in initial seasonal variability of surface temperature and dissolved inorganic carbon. Models with larger seasonal p CO 2 amplitudes in 1980–1984 exhibit proportionally larger C ant ‐driven growth, widening the across‐model disagreement in seasonal p CO 2 amplitudes over time. Consequently, growing model disagreement in wintertime sea‐air p CO 2 disequilibrium causes divergence in cumulative ocean carbon sink estimates. Seasonal wind variability may further modulate CO 2 flux differences, but its role could not be directly assessed. These findings demonstrate that systematic seasonal p CO 2 biases contribute to diverging annual ocean carbon uptake trends, highlighting the importance of improving seasonal carbon cycle process representations to reduce uncertainty in future climate change projections.