Satellite-observed strong subtropical ocean warming as an early signature of global warming

The escalating concentration of greenhouse gases (GHGs) in the atmosphere over the past century, particularly in recent decades, has led to unprecedented warming and widespread climate change. Understanding the extent of anticipated warming due to rising GHGs and assessing the current status of global warming are crucial for planning effective adaptation measures. Since the ocean is the primary sink for the anomalous heat trapped by increasing GHGs, evaluating ocean warming is key to addressing these questions. While the International Comprehensive Ocean-Atmosphere Data Set (ICOADS) provides a long-term record of ocean temperature data spanning over three centuries, uncertainties arise from inconsistencies in sampling. The advent of satellite observations has significantly improved the quality and coverage of ocean temperature records, providing a valuable 40-year dataset. This satellite data reveals intriguing features: accelerated warming in subtropical regions, particularly along oceanic western boundary currents, and cooling over the subpolar North Atlantic. The former is attributed to intensified and poleward-shifting western boundary currents, while the latter is linked to a weakening Atlantic Meridional Overturning Circulation. Hemispheric differences in warming are also observed, with stronger warming in the Northern Hemisphere and slight cooling in the Southern Ocean, likely due to sea ice expansion, deepwater upwelling, and Antarctic ozone depletion. Despite these known patterns, the study focuses on the dominant large-scale warming observed in subtropical latitudes. This study compares satellite-observed ocean warming patterns with reconstructed mid-Pliocene sea surface temperatures (SSTs) and model simulations of ocean warming at different stages. The authors hypothesize that the observed pattern is closely tied to upper ocean circulation and emerges only in the early stages of anthropogenic warming, when warming signals are concentrated in the upper ocean. They anticipate that on centennial to millennial timescales, subpolar ocean warming will eventually surpass subtropical warming.

Previous research has identified accelerated warming over the subtropical extensions of oceanic western boundary currents and a cold patch over the subpolar North Atlantic. Studies have linked the subtropical warming to the intensification and poleward shift of these currents, while the subpolar cooling is connected to a weakening Atlantic Meridional Overturning Circulation. Hemispheric differences in warming have also been noted, with stronger warming in the Northern Hemisphere compared to the Southern Hemisphere. Several hypotheses have been put forth to explain the observed cooling in the Southern Ocean, including sea ice expansion, upwelling of pristine deepwater, and Antarctic ozone depletion. The dominant large-scale warming observed in subtropical latitudes, however, has been interpreted as a manifestation of natural climate variability, specifically relating it to the negative phase of the Pacific Decadal Oscillation (PDO). Existing studies show inconsistencies in how much of the observed warming is attributable to anthropogenic climate change versus natural variability, emphasizing the need for further investigation and refined analysis techniques.

This study utilizes four types of data: satellite-observed SST data (NOAA OI SST V2), multi-model simulations from CMIP6 and PlioMIP2, mid-Pliocene SST proxies (PRISM database), and sensitivity simulations using FESOM1.4 and AWI-ESM. The satellite SST data covers 1982-2022. CMIP6 data includes the abrupt4xCO2 and piControl experiments from 47 models to analyze the early-stage ocean warming response to rapidly rising GHGs. The historical ssp245 and 1pctCO2 experiments are also examined in supplementary materials. Data is interpolated to a 1°x1° grid. PlioMIP2 data provides mid-Pliocene SST anomalies from seven models (plus COSMOS) using 400 ppmv CO2 for Pliocene simulations and 280/284 ppmv for pre-industrial comparisons. Mid-Pliocene SST proxies from the PRISM database are used for comparison. To study long-term warming patterns, a 3000-year abrupt4xCO2 experiment using AWI-ESM is conducted, and results are compared to pre-industrial control runs. To isolate the impact of ocean dynamics on warming, the FESOM1.4 model is utilized, conducting a control simulation and a Uniform4K experiment with a uniform 4 K surface air temperature increase. This experiment simplifies the surface heat flux using a bulk formulation linearly proportional to SST and air temperature differences. An AWI-ESM simulation employing an age tracer tracks ocean water trajectories, aiding in understanding circulation characteristics. Finally, the study includes a method to remove the influence of PDO and AMO fluctuations from observational SST trends through linear regression and reconstruction. Barotropic streamfunction is calculated to illustrate the spatial structure of ocean gyres.

Satellite data reveals widespread strong subtropical ocean warming, concentrated in western ocean basins, a pattern initially attributed to internal climate variability. However, this warming persists even during positive PDO phases, raising doubts about the dominance of PDO. Removing PDO and AMO influences still shows this enhanced subtropical warming. This pattern correlates strongly (0.53 initially, 0.67 after a decade) with the early stage of the CMIP6 abrupt4xCO2 experiment. The resemblance, however, is short-lived, diverging from observations after 1-2 decades. Both abrupt4xCO2 and 1pctCO2 experiments show similar patterns. A FESOM1.4 Uniform4K experiment, with uniform warming forcing, reveals that surface convergence of subtropical gyres enhances warming in those regions. The observed warming pattern, however, shows maxima on the polar side of subtropical gyres due to a poleward shift in circulation, absent in the Uniform4K experiment. An AWI-ESM simulation using an age tracer indicates that subtropical waters are significantly younger (0-5 years) than subpolar waters, implying surface origin and downwelling. This implies that the surface radiative heating is concentrated in the subtropics, leading to enhanced warming. The observed warming contrasts with paleoclimate reconstructions of the mid-Pliocene which show stronger warming in subpolar regions. Long-term (3000-year) AWI-ESM abrupt4xCO2 simulation shows that, in the long-term, this pattern reverses, and the greatest warming occurs in the subpolar regions. The difference in warming patterns is due to the distinct response times: radiative warming affects subtropical upper oceans quickly but subpolar warming depends on deep ocean warming, taking millennia. This explains the transient enhanced subtropical warming at the early stage and the eventual amplified subpolar warming near equilibrium.

The study demonstrates that the observed strong subtropical ocean warming is not solely a result of internal climate variability but is constrained by ocean dynamics of surface convergence and divergence. The early-stage dominance of subtropical warming is a transient phenomenon due to heat absorption in the upper ocean. The long-term warming pattern, however, will shift towards subpolar regions as the deep ocean warms. This finding contradicts the initial interpretations relating the pattern to PDO-like variability. The discrepancies observed between model simulations and observations in the southern Indian Ocean (attributed to missing iceberg activity) and the tropical eastern Pacific (likely due to cold biases and aerosol forcing misrepresentations) require further investigation. However, these discrepancies do not affect the main conclusions regarding the subtropical and subpolar warming patterns. The observed warming pattern influences regional climate by shifting meridional temperature gradients and driving changes in circulation and storm activity, but these changes may be temporary. The currently observed warming represents an early stage; once high-latitude warming develops, irreversible impacts on ice sheets and sea levels are likely.

This study reveals the dominant subtropical ocean warming observed in satellite data is a transient response constrained by ocean dynamics in the early stages of anthropogenic warming. This contradicts previous interpretations of the warming as purely a manifestation of internal climate variability. The study highlights the importance of considering both ocean circulation and deep ocean warming timescales when interpreting observed warming trends. Future research should focus on improving model representation of processes like iceberg activity and accurately simulating the complexities of ocean heat uptake, regional circulation and its impact on climate sensitivity.

The study notes limitations related to model biases. CMIP6 models struggle to reproduce the observed cooling in the tropical eastern Pacific, potentially due to cold biases in model mean states, aerosol forcing misrepresentations, or inaccurate representations of the Antarctic ozone hole and meltwater impacts. The lack of iceberg activity in CMIP6 models leads to discrepancies between models and observations in the southern Indian Ocean. While these discrepancies do not invalidate the central findings on subtropical and subpolar warming, they emphasize the need for further model refinements.