Comparison of Low-Frequency Internal Climate Variability in CMIP5 Models and Observations

• Published in: Journal of Climate
• Main Authors: Cheung, Anson H., Mann, Michael E., Steinman, Byron A., Frankcombe, Leela M., England, Matthew H., Miller, Sonya K.
• Other Authors: Univ Arizona, Dept Geosci, Department of Geosciences, The University of Arizona, Tucson, Arizona, Department of Meteorology and Atmospheric Science, and Earth and Environmental Systems Institute, The Pennsylvania State University, University Park, Pennsylvania, Department of Earth and Environmental Sciences, and Large Lakes Observatory, University of Minnesota Duluth, Duluth, Minnesota, ARC Centre of Excellence for Climate System Science, and Climate Change Research Centre, University of New South Wales, Sydney, New South Wales, Australia
• Format: Article in Journal/Newspaper
• Language: English
• Published: AMER METEOROLOGICAL SOC 2017
• Subjects: Pacific; North Atlantic
• Online Access: http://hdl.handle.net/10150/624456; https://doi.org/10.1175/JCLI-D-16-0712.1

Low-frequency internal climate variability (ICV) plays an important role in modulating global surface temperature, regional climate, and climate extremes. However, it has not been completely characterized in the instrumental record and in the Coupled Model Intercomparison Project phase 5 (CMIP5) model ensemble. In this study, the surface temperature ICV of the North Pacific (NP), North Atlantic (NA), and Northern Hemisphere (NH) in the instrumental record and historical CMIP5 all-forcing simulations is isolated using a semiempirical method wherein the CMIP5 ensemble mean is applied as the external forcing signal and removed from each time series. Comparison of ICV signals derived from this semiempirical method as well as from analysis of ICV in CMIP5 preindustrial control runs reveals disagreement in the spatial pattern and amplitude between models and instrumental data on multidecadal time scales (>20 yr). Analysis of the amplitude of total variability and the ICV in the models and instrumental data indicates that the models underestimate ICV amplitude on low-frequency time scales (>20 yr in the NA; >40 yr in the NP), while agreement is found in the NH variability. A multiple linear regression analysis of ICV in the instrumental record shows that variability in the NP drives decadal-to-interdecadal variability in the NH, whereas the NA drives multidecadal variability in the NH. Analysis of the CMIP5 historical simulations does not reveal such a relationship, indicating model limitations in simulating ICV. These findings demonstrate the need to better characterize low-frequency ICV, which may help improve attribution and decadal prediction. U.S. National Science Foundation [AGS-1263225]; Australian Research Council 6 month embargo; Published Online: 13 March 2017 This item from the UA Faculty Publications collection is made available by the University of Arizona with support from the University of Arizona Libraries. If you have questions, please contact us at [e-mail].