Finding the driver of local ocean-atmosphere coupling in reanalyses and CMIP5 climate models

Identification of the driver of coupled anomalies in the climate system is of great importance for a better understanding of the system and for its use in predictive efforts with climate models. The present analysis examines the robustness of a physical method proposed three decades ago to identify coupled anomalies as of atmospheric or oceanic origin by analyzing 850 mb vorticity and sea surface temperature anomalies. The method is then used as a metric to assess the coupling in climate simulations and a 30-year hindcast from models of the CMIP5 project. Analysis of the frequency of coupled anomalies exceeding one standard deviation from uncoupled NCEP/NCAR and ERA-Interim and partially coupled CFSR reanalyses shows robustness in the main results: anomalies of oceanic origin arise inside the deep tropics and those of atmospheric origin outside of the tropics. Coupled anomalies occupy similar regions in the global oceans independently of the spatiotemporal resolution. Exclusion of phenomena like ENSO, NAO, or AMO has regional effects on the distribution and origin of coupled anomalies; the absence of ENSO decreases anomalies of oceanic origin and favors those of atmospheric origin. Coupled model simulations in general agree with the distribution of anomalies of atmospheric and oceanic origin from reanalyses. However, the lack of the feedback from the atmosphere to the ocean in the AMIP simulations reduces substantially the number of coupled anomalies of atmospheric origin and artificially increases it in the tropics while the number of those of oceanic origin outside the tropics is also augmented. Analysis of a single available 30-year hindcast surprisingly indicates that coupled anomalies are more similar to AMIP than to coupled simulations. Differences in the frequency of coupled anomalies between the AMIP simulations and the uncoupled reanalyses, and similarities between the uncoupled and partially coupled reanalyses, support the notion that the nature of the coupling between the ocean and the atmosphere is transmitted into the reanalyses via the assimilation of observations.