Scale-Dependent Performance of CMIP5 Earth System Models in Simulating Terrestrial Vegetation Carbon

• Published in: Journal of climate 2015-07, Vol.28 (13), p.5217-5232
• Main Authors: Jiang, Lifen, Yan, Yaner, Hararuk, Oleksandra, Mikle, Nathaniel, Xia, Jianyang, Shi, Zheng, Tjiputra, Jerry, Wu, Tongwen, Luo, Yiqi
• Format: Article
• Language: English
• Subjects: 21st century; Biomes; Carbon; Carbon cycle; Carbon dioxide; Carbon sequestration; Climate change; Climate models; Data processing; Earth; Ecosystem models; Ecosystems; Feedback; Goodness of fit; Intercomparison; Modeling; Modelling; Permafrost; Residence time; Residence time distribution; Simulation; Simulations; Spatial distribution; SPECIAL C4MIP COLLECTION; Stocks; Uncertainty; Variability; Variables; Vegetation; Working groups
• ISSN: 0894-8755; 1520-0442
• DOI: 10.1175/jcli-d-14-00270.1

Model intercomparisons and evaluations against observations are essential for better understanding of models’ performance and for identifying the sources of uncertainty in their output. The terrestrial vegetation carbon simulated by 11 Earth system models (ESMs) involved in phase 5 of the Coupled Model Intercomparison Project (CMIP5) was evaluated in this study. The simulated vegetation carbon was compared at three distinct spatial scales (grid, biome, and global) among models and against the observations (an updated database from Olson et al.’s “Major World Ecosystem Complexes Ranked by Carbon in Live Vegetation: A Database”). Moreover, the underlying causes of the differences in themodels’ predictions were explored. Model–data fit at the grid scale was poor but greatly improved at the biome scale. Large intermodel variability was pronounced in the tropical and boreal regions, where total vegetation carbon stocks were high. While 8 out of 11 ESMs reproduced the global vegetation carbon to within 20% uncertainty of the observational estimate (560 ± 112 Pg C), the simulated global totals varied nearly threefold between the models. The goodness of fit of ESMs in simulating vegetation carbon depended strongly on the spatial scales. Sixty-three percent of the variability in contemporary global vegetation carbon stocks across ESMs could be explained by differences in vegetation carbon residence time across ESMs (P< 0.01). The analysis indicated that ESMs’ performance of vegetation carbon predictions can be substantially improved through better representation of plant longevity (i.e., carbon residence time) and its respective spatial distributions.