Deus ex machina long-term cooling of the eastern Pacific cold tongue in ocean reanalysis data

The emergence of an observed distinctive, meridionally confined surface cooling trend in the tropical Pacific cold tongue over recent decades contrasts sharply with the rapid warming simulated by most climate models, representing a key unresolved feature of climate change. Ocean-only simulations, even when forced with observed atmospheric conditions, likewise fail to reproduce the observed long-term absence of surface layer warming over the eastern Pacific cold tongue. Here we examine the long-term heat budget of the surface layer in the equatorial Pacific using atmospheric and oceanic reanalysis data, quantifying contributions from surface heat fluxes, ocean advection, and vertical diffusion inferred through Richardson number-based diffusivity estimates. In the Ocean ReAnalysis System 5 (ORAS5), the observed cold tongue cooling cannot be reproduced by the model’s intrinsic dynamics alone; instead, it depends on a surface heat flux adjustment imposed during data assimilation. A reduction in the warming effect associated with this adjustment in the eastern Pacific cold tongue over time emerges as the dominant contributor to the long-term cooling in the reanalysis data, a deus ex machina obscuring the actual physical drivers of change in the real ocean. We discuss potential origins of this deus ex machina cooling effect, including the influence of analysis increments, uncertainties in surface forcing and possible problems in the representation of oceanic processes, particularly subsurface turbulent heat flux induced by oceanic mixing processes. This work emphasizes that reanalysis-based assessments remain subject to inherent biases from the ocean models they are built upon. Resolving the discrepancy between observed and simulated Pacific trends will require confronting the structural limitations of models.