全球变化知识资源中心

Our results suggest that the hiatus in global mean temperature rise has been strongly influenced by the negative phase of the PDO. The planet is still warming⁸, but the changes in atmospheric circulation and surface winds, in particular, have changed ocean currents and more heat is being sequestered at greater depths. Effects are greatest in northern winter and there are profound regional manifestations. In particular, a cooler Europe is not reflected in the regime shift patterns in Fig. 3a, and instead was related to a trend within the past epoch and cooler years after 2005 in winter^{1, 4} (see Supplementary Fig. 1). In northern winter, teleconnections by means of quasi-stationary atmospheric Rossby waves forced from the tropical Pacific influenced the Arctic and predisposed the North Atlantic Oscillation to be in its negative phase (Figs 3c, 5a and 6a). In turn this favours cold outbreaks in Europe, as occurred in 2009–2010, 2010–2011 and 2012–2013 (Supplementary Fig. 5). These results help explain why the main absence of very recent warming over land has been over Eurasia in winter^{1, 4}.

When atmospheric circulation patterns persist then there is generally anomalous atmospheric forcing, which mainly arises from anomalous SSTs that are most persistent in the central tropical Pacific. Indeed, our results demonstrate that tropical Pacific forcing of the atmosphere such as that associated with a negative phase of the PDO produces many of the pronounced atmospheric circulation anomalies observed globally during the hiatus. Tropical Pacific rainfall variations also occur on shorter timescales, including influences from the Madden–Julian Oscillation that are well established to influence the North Atlantic Oscillation^{29, 30} and other transients, and from El Niño events on year-to-year timescales modulated by the PDO. Consequently, the exact teleconnections at any time vary substantially.

For instance in 2013–2014 northern winter, it has been speculated¹⁰ that increased greenhouse gases may have influenced the SSTs and helped trigger the substantial Rossby wave patterns of a strong ridge over the west coast of North America, a strong cold trough over the eastern United States and wet conditions over the United Kingdom. We note that the tropical Pacific rainfall during this winter featured very heavy rains near 160° E on the Equator associated with ‘westerly wind bursts’ and the developing El Niño event even though the dry region still existed in the central tropical Pacific.

Recent results³¹ suggest that these influences profoundly affected the Arctic, although that study was for annual means. We are able to replicate the essence of the teleconnections using a simple forcing in a full atmospheric model, whereas a coupled model⁷ did not realistically extend the patterns to the Arctic or North Atlantic. Much has been made of late about the atmosphere possibly becoming more ‘wavy’, and claims have been made that increased waviness is associated with loss of Arctic sea ice and Arctic amplification^{32, 33}. Such claims have been disputed^{34, 35, 36}. Our results, and others^{28, 31, 37, 38, 39} suggest an alternative hypothesis where the warmer Arctic is a consequence, not a cause, of the wavy pattern through increased exchanges of cold air from the Arctic with warmer air from lower latitudes, in association with the teleconnections across the region that originated in the tropical Pacific. More fundamentally, the reason the Arctic is a less likely source of forced atmospheric waves is because heating influences from that region are much smaller in amplitude and less persistent. Precipitation and associated latent heat anomalies are an order of magnitude less than in the tropics.

In the Southern Hemisphere, the tropical Pacific SST has been found to affect the Antarctic from 1979 to 2009⁴⁰, although that study did not separate the effects into the two PDO regimes, as we have done here. The change in wave train after 1999 (Figs 3 and 5) is mainly in winter, and again is remarkably well simulated (Fig. 6) with an idealized forcing.

There is a strong predilection for anomalous atmospheric circulation conditions mainly in the winter of each hemisphere and in the subtropics of the summer hemisphere to arise from tropical Pacific SST anomalies. These have been recognized as a dominant mode of natural interannual variability associated with the El Niño/Southern Oscillation, but here we focused on the interdecadal variability that has become strongly evident recently through its manifestation as a pause in the rise of global mean temperatures. Accompanying the recent negative phase of the PDO has been striking changes in tropical and subtropical winds and ocean currents, with profound effects on ocean heat content and sea level. Some of these aspects seem to be unique to the past decade and raise questions about whether natural internal variability itself is being altered by climate change.