A new study, supported by CPO’s Climate Variability and Predictability Program (CVP) demonstrates that realistic Atlantic meridional overturning circulation (AMOC) decadal variations can be simulated in a coupled climate model by sea-surface temperature (SST) and sea-surface salinity (SSS) restoring. As a heat transport current in the North Atlantic, AMOC is an important component in Earth’s climate, and having realistic simulations of it is key to decadal climate predictions. Forced ocean – sea ice (FOSI) model simulations are used to initialize decadal climate predictions in a fully coupled model. However, due to systematic differences in the atmospheric state used for FOSI and that of the coupled climate model, achieving a balanced atmosphere-ocean state continues to be a struggle. As a result, prediction skill may be lower than expected.
This study was performed by researchers from Texas A&M University, National Center for Atmospheric Research (NCAR), Ocean University of China, and Qingdao National Laboratory for Marine Science and Technology. Texas A&M’s Qiuying Zhang and Ping Chang, and NCAR’s Stephen G. Yeager and Gokhan Danabasoglu, funded by CVP, were part of the team that compared two sets of 6-member ensemble fully coupled Community Earth System Model version 2 (CESM2) simulations. The first set utilized a strong SST-restoring only as used in seasonal prediction, while the second set utilized both strong SST- and SSS-restoring. The group used FOSI values as a basis to assess the skill of the simulations.
Published in Geophysical Research Letters (GRL), the study shows that only the second set was able to realistically simulate AMOC decadal variation, indicating the important role of SSS in representing realistic AMOC decadal variability in fully coupled climate systems. The set also produced an atmospheric mean state that is closer to that of the free coupled simulation than to the observations used in FOSI. This reduces the atmosphere-ocean state imbalance when decadal climate predictions are initiated, potentially improving prediction skill.
For more information, contact Jose Algarin.