Uncertainty in the response of terrestrial carbon sink to environmental drivers undermines carbon-climate feedback predictions.
D N Huntzinger,A M Michalak,C Schwalm,P Ciais,A W King,Y Fang,K Schaefer,Y Wei,R B Cook,J B Fisher,D Hayes,M Huang,A Ito,A K Jain,H Lei,C Lu,F Maignan,J Mao,N Parazoo,S Peng,B Poulter,D Ricciuto,X Shi,H Tian,W Wang,N Zeng,F Zhao
DOI: https://doi.org/10.1038/s41598-017-03818-2
IF: 4.6
2017-01-01
Scientific Reports
Abstract:Terrestrial ecosystems play a vital role in regulating the accumulation of carbon (C) in the atmosphere. Understanding the factors controlling land C uptake is critical for reducing uncertainties in projections of future climate. The relative importance of changing climate, rising atmospheric CO2, and other factors, however, remains unclear despite decades of research. Here, we use an ensemble of land models to show that models disagree on the primary driver of cumulative C uptake for 85% of vegetated land area. Disagreement is largest in model sensitivity to rising atmospheric CO2 which shows almost twice the variability in cumulative land uptake since 1901 (1 s.d. of 212.8 PgC vs. 138.5 PgC, respectively). We find that variability in CO2 and temperature sensitivity is attributable, in part, to their compensatory effects on C uptake, whereby comparable estimates of C uptake can arise by invoking different sensitivities to key environmental conditions. Conversely, divergent estimates of C uptake can occur despite being based on the same environmental sensitivities. Together, these findings imply an important limitation to the predictability of C cycling and climate under unprecedented environmental conditions. We suggest that the carbon modeling community prioritize a probabilistic multi-model approach to generate more robust C cycle projections.