RecentWalker circulation strengthening and Pacific cooling amplified by Atlantic warming

An unprecedented strengthening of Pacific trade winds since the late 1990s (ref. 1) has caused widespread climate perturbations, including rapid sea-level rise in the western tropical Pacific2–5, strengthening of Indo-Pacific ocean currents6,7, and an increased uptake of heat in the equatorial Pacific thermocline1. The corresponding intensification of the atmospheric Walker circulation is also associated with sea surface cooling in the eastern Pacific,which has been identified as one of the contributors to the current pause in global surface warming1,8,9. In spite of recent progress in determining the climatic impacts of the Pacific trade wind acceleration, the cause of this pronounced trend in atmospheric circulation remains unknown. Here we analyse a series of climate model experiments along with observational data to show that the recent warming trend in Atlantic sea surface temperature and the corresponding trans-basin displacements of the main atmospheric pressure centreswere key drivers of the observed Walker circulation intensification, eastern Pacific cooling, North American rainfall trends and western Pacific sea-level rise. Our study suggests that global surface warming has been partly o set by the Pacific climate response to enhanced Atlantic warming since the early 1990s. Given the importance of the recent wind-induced trends in Pacific sea level and surface temperature, it is vital to determine the underlying causes. Recent studies have focused mostly on the low-frequency Pacific climate modes, such as the Pacific Decadal Oscillation10 (PDO) or the Interdecadal Pacific Oscillation11 (IPO), to explain Pacific wind shifts and the current pause in greenhouse warming8,9,12. However, the fact that this unprecedented 1992–2011 equatorial Pacific zonal wind trend1 is not consistent with a Pacificonly sea surface temperature (SST) driving mechanism5 suggests a role for dynamics outside the tropical Pacific in this atmospheric reorganization, such as from the Indian Ocean5,13, the Atlantic14 or both. This scenario is further supported by the fact that the trade wind intensification since the early 1990s is related to a global scale see-saw in atmospheric surface pressure, which is characterized by a positive sea-level pressure (SLP) trend in the Pacific and a negative trend in the Indo-Atlantic region (Fig. 1a). Our study uses a series of climate model experiments in combination with observational analyses to identify potential remote drivers of Pacific equatorial wind changes since the early 1990s and their corresponding impacts on global climate. We first conduct a suite of five-member ensemble sensitivity experiments using the Community Atmospheric Model, version 4 (CAM4) atmospheric general circulation model (AGCM; Methods and Supplementary Table 1) to further elucidate the underlying physical mechanism of the recent inter-basin SLP see-saw (Fig. 1a) and its corresponding effects on the Pacific trade wind systems. We prescribe the trend in observed global SST anomalies (SSTA) over the period 1992–2011 (Fig. 1a), which is characterized by an overall Atlantic warming, an eastern Pacific cooling trend and western subtropical Pacific and Indian Ocean warming. This pattern is in fact quite different from the typical global warming hiatus pattern simulated by coupled general circulation models9, which exhibit similar cooling trends across all tropical oceans. In response to the applied global SST trend forcing, the AGCMexperiment reproduces the observed global SLP see-saw and the related intensification of tropical Pacific trade winds qualitatively well (Fig. 1c,d and Supplementary Figs 1g,h and 3a). It is important to note that in spite of capturing the overall sealevel pressure andwind trend patterns (Fig. 1c,d and Supplementary Fig. 1g,h), the CAM4 ensemblemean underestimates themagnitude of the central Pacific wind stress intensification by a factor of three (Supplementary Table 2). The magnitude of the underestimated wind stress response of these AGCM ensemble simulations is consistent with the weaker response for this period simulated by an ensemble of 25 models of SST anomaly-forced AGCM experiments conducted as part of the Atmospheric Model Inter-Comparison Project, version 5, (AMIP5; Supplementary Information and Supplementary Figs 1c,e and 2a). It should be noted here that the simulated AMIP5 trends in zonal 850 hPa winds (Supplementary Fig. 1d,f and Table 3) are much more consistent with the reanalysis data as compared to the surface stresses, thus indicating AGCM deficiencies in the downward mixing of momentum through the boundary layer. Next we carry out a series of CAM4 simulations, each with five ensemble members, with prescribed sea surface temperature forcing in some ocean areas and a slab mixed-layer ocean in others to isolate the effect of different ocean basins on the Pacific trade wind intensification (Methods and Supplementary Table 1). The origin of the recent Pacific climate trends becomes apparent in the ensemble AGCM experiment that is forced only by Atlantic SSTA trends while using a mixed-layer ocean in the Pacific and prescribed climatological SST in the Indian Ocean (Supplementary Information and Supplementary Table 1). In this experiment the global atmosphere and Pacific SSTs can adjust to the remote observed Atlantic SSTA trend forcing. The ensemble mean results of this experiment demonstrate that the recent Atlantic warming generates a trans-basin (Pacific/Atlantic) SLP see-saw that is very similar to the observations (Fig. 1e and Supplementary Fig. 3b). This generates a significant strengthening of the wind stress in the central Pacific (Fig. 1f), which can account on average for