Broad Consistency Between Observed And Simulated Trends In Sea Surface Temperature Patterns

Using seven single-model ensembles and the two multimodel ensembles CMIP5 and CMIP6, we show that observed and simulated trends in sea surface temperature (SST) patterns are globally consistent when accounting for internal variability. Some individual ensemble members simulate trends in large-scale SST patterns that closely resemble the observed ones. Observed regional trends that lie at the outer edge of the models' internal variability range allow two nonexclusive interpretations: (a) Observed trends are unusual realizations of the Earth's possible behavior and/or (b) the models are systematically biased but large internal variability leads to some good matches with the observations. The existing range of multidecadal SST trends is influenced more strongly by large internal variability than by differences in the model formulation or the observational data sets.Plain Language Summary Climate model simulations agree well with the observed evolution of the global mean sea surface temperature, but their ability to realistically represent changes in the patterns of sea surface temperatures has been questioned. We show with an unprecedented number of simulations from different models and different initial conditions that the observed and simulated changes in SST patterns are consistent in most regions of the ocean. For each model, a few individual simulations recreate the observed patterns. In some regions, the observed changes may be an extreme realization of the Earth's possible behavior. Alternatively, structural model errors may be hidden by the large range of possible realizations.