Sources of variation in simulated ecosystem carbon storage capacity from the 5th Climate Model Intercomparison Project (CMIP5)

Ecosystem carbon (C) storage strongly regulates climate-C cycle feedback and is largely determined by both C residence time and C input from net primary productivity (NPP). However, spatial patterns of ecosystem C storage and its variation have not been well quantified in earth system models (ESMs), which is essential to predict future climate change and guide model development. We intended to evaluate spatial patterns of ecosystem C storage capacity simulated by ESMs as part of the 5th Climate Model Intercomparison Project (CMIP5) and explore the sources of multi-model variation from mean residence time (MRT) and/or C inputs. Five ESMs were evaluated, including C inputs (NPP and [gross primary productivity] GPP), outputs (autotrophic/heterotrophic respiration) and pools (vegetation, litter and soil C). ESMs reasonably simulated the NPP and NPP/GPP ratio compared with Moderate Resolution Imaging Spectroradiometer (MODIS) estimates except NorESM. However, all of the models significantly underestimated ecosystem MRT, resulting in underestimation of ecosystem C storage capacity. CCSM predicted the lowest ecosystem C storage capacity (~10 kg C m^-2) with the lowest MRT values (14 yr), while MIROC-ESM estimated the highest ecosystem C storage capacity (~36 kg C m^-2) with the longest MRT (44 yr). Ecosystem C storage capacity varied considerably among models, with larger variation at high latitudes and in Australia, mainly resulting from the differences in the MRTs across models. Our results indicate that additional research is needed to improve postphotosynthesis C-cycle modelling, especially at high latitudes, so that ecosystem C residence time and storage capacity can be appropriately simulated.