1. Introduction [1, 2]
The atmosphere-ocean system has exhibited some rather pronounced trends during the past few decades. [3.4] Winters have been getting milder, particularly over northern Europe, Russia, and central and western Canada. Global warming skeptics take delight in pointing out that the strongest warming trends have been observed over the coldest regions of the hemisphere. One notable exception to the trend toward global warming has been Labrador and west Greenland, where winters have been growing more severe. [5] Barometric pressure has been dropping over the Arctic and the westerlies at subpolar latitudes (~55°N) have intensified. [6] Winter storms over the North Atlantic have been getting more intense and wave crests have been getting higher. Winter precipitation over Scotland and Norway has increased. [7] Snow accumulations at higher elevations in Norway have been so heavy that a number of the glaciers have begun to advance, even in the face of the trend toward milder temperatures. [8.9] Meanwhile, winter rainfall over southern Europe has been decreasing, with major droughts during the early 1990's and again last year. Winter temperatures in the lower stratosphere 20 kilometers above the earth’s surface have been cooling over the Arctic polar cap region, with temperature drops of 5 C or more, and there has been appreciable thinning of the stratospheric ozone layer poleward of 40°N.
In this lecture I will argue that all the trends noted in the previous paragraph are all linked to the same phenomenon and that equally pronounced changes over and around the Antarctic are linked to its Southern Hemisphere analogue.
2. Some personal reminscences [10]
The challenge of understanding climatic change is promoting a remarkable synthesis of the earth sciences. A generation ago, when I entered the field, geology, meteorology and oceanography were largely separate disciplines, and each of them touched on only limited aspects of the climate problem. The study of climates of the past was largely the province of geologists who, in collaboration with paleo- botanists and zoologists were assembling an increasingly detailed picture of the record of past climate variations based on proxy evidence. Some of them carefully scrutinized their proxy records of past climate with hopes of finding evidence of orbital and other periodocities might lend some insight into the underlying causes of climate variability, but beyond that, most of the scientists working in this field weren't in a position to do much more than gather evidence and speculate. They lacked the tools to perform a global synthesis of the proxy records and to investigate the dynamical mechanisms that link climate variations in widely separated parts of the world.
The meteorologists were developing the tools needed for investigating climate variability. They were already having some success in using numerical models based on fundamental physical principles to predict time evolution of the global wind field, but at that time the skill of weather forecasts was limited to a few days. By performing extended runs with the same numerical models, researchers were able to realistically simulate many of the features of the present climate, a notable example being the transport of water vapor in the global hydrologic cycle, as documented by the late Professor Jose Peixoto of the University of Lisboa. .... Yet despite the promising developments in climate modeling, I think it is fair to say that back at that time most meteorologists at that time tended to view climate as unchanging, and climatology as a stagnant field. It was only a few of our peers who recognized the great potential of numerical models for the study of climate variability.
A generation ago much of the research on the oceans was focused on regional phenomena. The more global thinkers in the field [11] like Henry Stommel were still struggling with fundamental issues, like the factors that govern the strength of the western boundary currents, and the rate of recirculation of water between the surface layer of the ocean and the deep abysses below. Ocean general circulation models were just in their infancy. Yet already, developments within the field of oceanography were helping to set the stage for a renaissance in climate research. Marine sediment cores were providing a wealth of detailed information on past climate variability. Leading chemical oceanographers of that time like [12] Wallace Broecker were beginning to consider the carbon cycle and other biogeochemical cycles from a global, interdisciplinary perspective. And pioneering efforts in coupled atmosphere-ocean modeling were already on the drawing boards.
3. The El Nino Southern Oscillation phenomenon [13]
By the late 1970's momentum began to build for an expanded program of climate research. Coincidentally, just about that time, the landmark papers of [14] Jacob Bjerknes on the El Niño-Southern Oscillation (ENSO) phenomenon began attracting widespread attention in both the atmospheric sciences and physical oceanography communities. ENSO provided a concrete example of how the atmosphere and oceans, as a coupled system, are capable of supporting a mode of space / time variability altogether different than the modes of variability of the individual components of the system.
[15] ENSO involves an atmosphere-ocean feedback loop. During the onset of a warm 'El Niño' event, weakening of the equatoral Pacific tradewinds leads to a deepening of the thermocline on the eastern side of the basin and decreased upwelling, both of which enable equatorial sea surface temperatures to warm toward the ambient temperature of the tropical oceans. Raising the temperature of the surface waters imparts additional buoyancy to the overlying air, enabling it to support deep cumulus convection. The increased convective rainfall along the equator favors a further weakening of the trades, which allows sea surface temperature to warm further--- a positive feedback. [16] The redistribution of the tropical rainfall also affects the upper air circulation at higher latitudes, producing a complex array of temperature and rainfall anomalies at the earth's surface. [17] For example during winters of 'El Niño years' when the surface waters of the equatorial Pacific are abnormally warm, temperatures over Alaska and western Canada tend to be warm as well [18] and regions of the Americas at subtropical latitudes tend to be wetter than normal. El Niño years are also usually marked by drought over Northeast Brazil, Central America, and parts of Africa, and a suppression of hurricanes over the Carribbean Sea.
The equatorial Pacific sea surface temperature anomalies that occur in association with El Niño events typically persist for about a year: much longer than the characteristic 'memory time' of either the winds or the equatorial ocean currents alone. The longer time scale is thus attributable to the coupling between the atmosphere and the ocean. By virtue of this extended memory, the coupled system is far more predictable than the atmosphere alone. This extended predictability is being exploited to make forecasts of El Nino-related climate anomalies out to several seasons in advance, using state-of-the-art coupled atmosphere-ocean models. The principal beneficiaries of these forecasts are residents of tropical countries, many of whom suffer drought during El Niño years, and residents of the countries of North and South America. The effects of ENSO on European climate are more subtle, but the forecasts are nonetheless of some value.
The ENSO paradigm has been instrumental in demonstrating the existence of a certain degree of order in the year-to-year variations in temperature and rainfall in the tropics and over some parts of the extratropics. It has also revealed important linkages between the physical variables and the biosphere. [19] For example, the deepening of the thermocline in the eastern Pacific during the El Niño years results in a decreased availability of nutrients to the euphotic zone, to the detriment of many species of marine life. [20] But the increased rainfall along the coastal drainage basins of Ecuador during these intervals provides increased nutrients for shrimp farms the coastal estuaries and habitat for mosquitoes. For the same reasons, sea birds languish while land birds thrive on the Galapagos Islands during El Niño years. These ENSO-related variations in the biosphere are of considerable interest in their own right, and they provide valuable clues as to how the physical variables in the climate system (temperature, ocean currents, winds, etc.) influence marine and terrestrial ecosystems, not only on the year-to-year time scale, but potentially on much longer time scales as well.
Some time around 1976-77 the ENSO cycle underwent a discernible shift. [21] Within the 30-year interval prior to that time, cold phases tended to be slightly more intense and more prevalent than the warm phases, whereas in the years since then, the opposite has been the case. [22] This 1976-77 'regime shift' is more strongly apparent in North Pacific climatic time series and in various indicators of the state of the marine ecosystem. In fact, the change that occurred around 1976-77 was originally identified as a biospher'c 'regime shift' and only later linked to the physical climate variables. There are indications of an analogous shift, but in the opposite sense around 1946, and perhaps another just within the past 2-3 years, coinciding with the extended cold phase of the ENSO cycle that has persisted from 1998 until the last month or two. This interdecadal-scale ENSO-related variability is referred to in some papers as the Pacific Decadal Oscillation. Whether it is a distinctive phenomenon in its own right, or merely a peculiar feature of the ENSO cycle remains to be seen, but it's generated considerable interest in the fisheries community [23] because it appears to be strongly correlated with the decade to decades ups and downs in salmon recruitment along the coasts of Alaska, British Columbia and Washington.
Research on ENSO is continuing. Efforts are underway to improve the skill of ENSO prediction, to extend the record of the ENSO cycle farther back into the past using proxy records, to quantify the impacts of ENSO on physical climate variables, on marine and terrestrial ecosystems, and on human activities, to understand the nature of ENSO-like variability on decadal time scale, and to predict whether and how global warming might impact ENSO. But even as research on ENSO continues, another phenomenon that may ultimately prove to be equally important from a global perspective, and perhaps even more important from a European perspective, has been attracting increasing attention-- [24] a phenomenon variously known (in different segments of the research community) as the North Atlantic Oscillation, the Arctic Oscillation, and the Northern Hemisphere annular mode.
4. Early history of the North Atlantic Oscillation [25]
The observed tendency for winter temperatures over Greenland and Scandinavia to vary out of phase (i.e., for one to be above normal while the other is below normal) can be traced back 200 years or more. The realization that this distinctive pattern of temperature anomalies is linked to an equally distinctive north-south see-saw of sea-level pressure anomalies between Iceland and a broad east-west belt centered on Portugal, dates back roughly 100 years. The French meteorologist Tesserenc de Bort, the German meteorologists Julius Hann and Felix Exner, and the British applied mathematician / meteorologist Sir Gilbert Walker all contributed to its discovery.
[26] Here is Exner's rendition of the pressure and temperature patterns, published in a paper in 1913. [27] The contours represent correlations with a time series of winter sea-level pressure over the Arctic. During winter months when the pressure over the Arctic is above normal, the pressure over Portugal and much of temperate latitudes tends to be below normal, as indicated by the belt of negative correlations centered over Portugal. In a similar manner, when the pressure over the Arctic is below normal, the pressure over Portugal tends to be above normal. [28] Since the winds tend to follow the pressure contours with low pressure to the left, it follows that below normal pressures over the Arctic favor abnormally strong westerlies over the North Atlantic from Nova Scotia toward Scotland and Scandinavia and vice versa. Hence, the pressure pattern defined by Exner's correlations can equally well be characterized as a modulation in the strength of the westerlies across the North Atlantic shipping lanes and, more generally, at subpolar latitudes. The corresponding surface air temperature correlation pattern indicates that high pressure over the Arctic is conducive to above normal temperatures in Greenland and above normal temperatures over northern Europe and vice versa. Exner reasoned that the temperature and pressure patterns are related by way of the winds: e.g., abnormally strong westerlies that prevail when Arctic pressure is below normal carry an abundance of warm air off the Atlantic into northern Europe causing temperatures to be above normal there. The influence of the winds upon the temperatures is clearly apparent the figure: when the westerlies are strong the western limit of subfreezing temperatures, indicated by the edge of the bluish shading, lies near the border between Germany and Poland, whereas when the westerlies are weak subfreezing conditions extend all the way into France.
[29] It was Sir Gilbert Walker who coined the term 'North Atlantic Oscillation' in a paper published in 1924. In that same paper he also coined the term, 'Southern Oscillation', which is the SO in ENSO, and the less enduring term 'North Pacific Oscillation' describing a suite of statistical relationships in the Pacific sector. By attaching names to suites of statistical relationships, Walker elevated them to the status of phenomena, at least in the minds of those who read his papers and took them seriously. In the case of the Southern Oscillation, his intuition has proven correct: the statistical relationships embodied in that name have proven to be the signature of the response of the global atmosphere to a distinctive mode of atmosphere-ocean interaction in the equatorial Pacific. But it was not until the work of Bjerknes and his successors, nearly half a century after Walker coined that term, that proof was forthcoming. Prior to that time, believing in the Southern Oscillation required something of a leap of faith. In an exchange of correspondence some years later, Walker was challenged to defend his belief. When one of his contemporaries gently chided him about the lack of a plausible mechanism that would explain the global pattern of correlations that he identiified with the Southern Oscillation, he replied: "I think the relationships of world weather are so complex that our only chance of explaining them is to accumulate the facts empirically... there is the strong presumption that when we have the data of the pressure and the temperature at [altitudes of] 10 and 20 km, we shall find a number of new relationships that are of vital importance." He was absolutely right: upper air data proved to be instrumental in understanding the ENSO phenomenon. Later in this lecture, we will need to consider whether Walker's intuition in naming the North Atlantic Oscillation has been vindicated in a similar manner.
In a series of papers published during World War II, [30] Carl Gustaf Rossby (regarded by many as the father of dynamical meteorology), and others at Massachusetts Institute of Technology, promoted the concept of an 'index cycle', which they hoped would prove to be a useful tool in forecasting weather on the 1-2 week time scale. This notion was based, not on the prior work cited in previous paragraphs (which they were apparently unaware of at the time) but upon the fact that (1) observations that the strength of the prevailing westerlies in middle latitudes fluctuates from one week to the next and (2) their simplified analysis of the governing equations suggested that these relatively slow fluctuations should be reflected in the configuration of planetary waves, which determines the longitudes where temperatures are above or below the longitudinally averaged values. For example, one week, when the westerlies are abnormally strong, Europe might lie within a cold trough in the planetary wave pattern, whereas a week later, when the westerlies are weaker, it would lie within a warm ridge. The 'zonal index' was originally defined as a measure of the strength of the westerlies that encircle the North Pole at latitudes ~45°N.
By the late 1940's Rossby had lost interest in the index cycle, but in 1951 two younger members of the same department published papers that seemed to imply that this concept still held promise. [31] In one of the papers Jerome Namias proposed a redefinition of the index cycle: instead of monitoring the strength of the westerlies as in the earlier papers, Namias argued that it would be more informative to characterize the hemispheric circulation in terms of the mean latitude of the westerlies. [32] Based on his experience as a synoptic meteorologist, he reasoned that when the westerlies are poleward of their mean position, cold, Arctic air masses tend to be confined within the polar cap region, and temperatures in midlatitudes tend to be above normal. Conversely, when the westerlies are displaced equatorward of their mean position, incursions of cold air masses into middle latitudes are more severe and more frequent. As a measure of the mean latitude of the westerlies Namias used an index based on the difference between the strength of the westerlies along 55 N and 35 N. [33] He used the term 'high index' as characterizing the state of the hemispheric circulation when the westerlies are north of their mean position, [34 ]and 'low index' as characterizing the state when they are south of their mean position. The other paper was one of the first publications of a new professor by the name of Edward Lorenz: a statistical analysis of the correlations between the pressures and winds at various latitudes. Lorenz's study was reminiscent of Exner's study nearly 40 years earlier, but for one important distinction: the pressure data were averaged around latitude circles before the correlations were calculated. But despite this difference in the methodology, Lorenz's results were remarkably similar to those of Exner, which he wasn't aware of. In particular, Lorenz confirmed that sea level pressure over the polar cap region tends to be negatively correlated with pressure along 45 N (the latitude of Lisbon). And consistent with the more subjective impression of Namias, he reported that this favored pattern in his statistical analysis is well represented by an index consisting of the strength of the winds averaged around 55 N.