Contstraining CMIP5 simulations of gross carbon fluxes with atmospheric carbonyl sulfide

Growth in terrestrial gross primary production (GPP) may provide a feedback for climate change, but there remains strong disagreement on the extent to which biogeochemical processes may suppress this GPP growth at the ecosystem to continental scales. The consequent uncertainty in modeling future carbon storage by the terrestrial biosphere creates one of the largest unknowns in global climate projections for the next century. Here we provide a global, measurement-based estimate of GPP growth during the last century using long-term atmospheric carbonyl sulfide (COS) records derived from ice core, firn, and ambient air samples. We interpret these records using a model that relates changes in COS concentration to changes in its sources and sinks, the largest of which is proportional to GPP. The COS record was most consistent with simulations that assume large GPP growth during the last century. Carbon-climate models that assume little to no GPP growth predicted trajectories of COS concentration over the anthropogenic era that differ from those observed.