Sources of Uncertainty in Modeled Land Carbon Storage within and across Three MIPs: Diagnosis with Three New Techniques

作者：Sha Zhou, Junyi Liang, Xingjie Lu, Qianyu Li, Lifen Jiang, Yao Zhang, Christopher R. Schwalm, Joshua B. Fisher, Jerry Tjiputra, Stephen Sitch, Anders Ahlström, D. N. Huntzinger, Yuefei Huang, Guangqian Wang, Yiqi Luo · 发表于：Journal of Climate · 年份：2018 · DOI：10.1175/jcli-d-17-0357.1 · 被引用次数：41 · 研究领域：Atmospheric and Environmental Gas Dynamics、Climate variability and models、Plant Water Relations and Carbon Dynamics

Terrestrial carbon cycle models have incorporated increasingly more processes as a means to achieve more-realistic representations of ecosystem carbon cycling. Despite this, there are large across-model variations in the simulation and projection of carbon cycling. Several model intercomparison projects (MIPs), for example, the fifth phase of the Coupled Model Intercomparison Project (CMIP5) (historical simulations), Trends in Net Land–Atmosphere Carbon Exchange (TRENDY), and Multiscale Synthesis and Terrestrial Model Intercomparison Project (MsTMIP), have sought to understand intermodel differences. In this study, the authors developed a suite of new techniques to conduct post-MIP analysis to gain insights into uncertainty sources across 25 models in the three MIPs. First, terrestrial carbon storage dynamics were characterized by a three-dimensional (3D) model output space with coordinates of carbon residence time, net primary productivity (NPP), and carbon storage potential. The latter represents the potential of an ecosystem to lose or gain carbon. This space can be used to measure how and why model output differs. Models with a nitrogen cycle generally exhibit lower annual NPP in comparison with other models, and mostly negative carbon storage potential. Second, a transient traceability framework was used to decompose any given carbon cycle model into traceable components and identify the sources of model differences. The carbon residence time (or NPP) was traced to basel...