Global land carbon sink estimates: towards right answer with right reasons

Estimates of the global land carbon sink differ significantly between top-down and bottom-up approaches. Top-down methods, such as those used by atmospheric inversion and the Global Carbon Project, estimate the sink at around -3 PgC y-1, whereas bottom-up approaches trained with FLUXNET generally provide implausibly large sinks, several times those of top-down estimates. Some global land carbon flux maps have achieved plausible global land carbon sink estimates; however, the histograms of annual net ecosystem exchange (NEE) from those maps show very narrow ranges, such as -100 to 100 gC m-2 y-1. This contradicts FLUXNET field observations, which show much wider ranges from -1000 gC m-2 y-1 to 500 gC m-2 y-1. Therefore, global maps that achieve both realistic global C sink estimates and ranges of annual NEE are not available. This is the critical bottleneck in global land carbon studies. We discuss potential reasons, including forest age, disturbance, fire, and the decoupling of photosynthesis and ecosystem respiration.