by Javier Vinós & Andy May
“These shifts are associated with significant changes in global temperature trend and in ENSO variability. The latest such event is known as the great climate shift of the 1970s.”Anastasios A. Tsonis, Kyle Swanson & Sergey Kravtsov (2007)
For those that prefer it, Christian Freuer has translated this post into German here.
While the study of weather variability has a long tradition, the science of climate change is a very young scientific topic, as attested to by the 1984 discovery of the first multidecadal oscillation, the primary global climate internal variability phenomenon, by Folland et al. The impact of this fundamental feature of the global climate system was discovered ten years later by Schlesinger and Ramankutty (1994), after modern global warming had already been blamed on CO2 changes, illustrating the risk of reaching a consensus with insufficient understanding of the topic at hand. The Pacific (inter) Decadal Oscillation (PDO) was discovered three years later (Mantua et al. 1997; Minobe 1997). The Atlantic Multidecadal Oscillation (AMO) was not named until just two decades ago (Kerr 2000).
Prior to the 1980s, it was generally thought that climate changed so slowly as to be almost imperceptible during the span of a human lifetime. But then it became clear that abrupt climate changes took place during the past glacial period (Dansgaard et al. 1984), Dansgaard–Oeschger events demonstrated that regional, hemispheric, and even global climate could undergo drastic changes in a matter of decades. The problem was that modern climate-change theory was built around gradual changes in the greenhouse effect (GHE) and did not have much room for abrupt, drastic, global changes that could not be properly explained by changes in greenhouse gas (GHG) radiative forcing.
The first explanations for glacial Dansgaard–Oeschger events involved drastic changes in meridional transport (MT) by the Atlantic Meridional Overturning Circulation (AMOC). AMOC is part of the global conveyor theory that tries to explain the flow of heat through the Earth’s oceans. The AMOC explanation, better known as the salt-oscillator hypothesis (Broecker et al. 1990), falls short however, as there is no evidence that the AMOC has undergone the abrupt and drastic changes required to produce the events. The current theory on MT is based on what is known as the Bjerkness compensation hypothesis, where changes in one of MT components (the oceanic or the atmospheric) are compensated for by similar changes of the opposite sign in the other. Current interpretations of the Dansgaard–Oeschger phenomenon are based on rapid sea-ice changes taking place in the Nordic seas that abruptly released a great amount of ocean-stored heat under the sea-ice (Dokken et al. 2013).
Dansgaard–Oeschger events have turned out to be a glacial-world phenomenon with little applicability to current Holocene conditions, but it is clear that abrupt climate changes are a reality that requires an explanation. Studies of Holocene climate change have identified at least 23 abrupt climate events (Fig. 4.1e; Vinós 2022) during the past 11,700 years (about two per millennium on average). They are well reflected in several proxies of different nature and identified as such in the paleoclimatic literature. From their different climatic signatures, it is clear that these events are not a response to a single cause. Yet modern climate-change theory has left us with only two possibilities, changes in radiative forcing produced by changes in atmospheric GHGs, or volcanic activity. These simple processes cannot explain them all. Changes in CO2 can be ruled out as a cause, since from 11,000 years ago to 1914 it remained between 250 and 300 ppm, and decadal to centennial oscillations in CO2 have only varied a few ppm according to ice cores. Volcanic forcing presents a problem, because its Holocene evolution has been opposite to temperature evolution. It was stronger when the planet warmed and weaker when it cooled, reaching a minimum c. 3,000 years ago, so it cannot be a strong driver of centennial temperature change. In fact, the Little Ice Age (LIA), the most recent abrupt climate event prior to modern global warming, cannot be explained by CO2 or volcanic forcing. According to the GISP2 ice core volcanic sulfate record (Fig. 4.1c; Zielinski et al. 1996), volcanic activity was above average between 1166–1345 AD, but was below average during most of the LIA, only becoming elevated again towards the end of it, in the 1766–1833 AD period, as the world began to warm.
In Fig. 4.1, (a) is the black curve, a global temperature reconstruction from 73 proxies (after Marcott et al. 2013; with original proxy dates and differencing average), expressed as the distance to the average in standard deviations (Z-score). The purple curve, (b), is Earth’s axial tilt (obliquity) in degrees. The red curve (c) is the Holocene volcanic sulfate in the GISP2 ice core in parts per billion summed for each century in the BP scale (rightmost point is 0–99 or 1851–1950), with the quadratic trendline shown as the thin red line. Data are from Zielinski et al. 1996. Curve (d) is light blue and shows CO2 levels as measured in the Epica Dome C (Antarctica) ice core. Data are from Monnin et al. 2004. The light grey bars, (e), are abrupt climatic events during the Holocene determined from ice-rafted petrological tracers (Bond et al. 2001), methane changes (Blunier et al. 1995; Kobashi et al. 2007; Chappellaz et al. 2013), Dead Sea level changes (Migowski et al. 2006), δ18O isotope changes in Dongge Cave (Wang et al. 2005), North Levant precipitation changes (Kaniewski et al. 2015), and dolomite abundance changes in the Gulf of Oman eolian deposition record (Cullen et al. 2000). Dark grey boxes at bottom give their approximate dates in ka. The figure is from Vinós 2022
Modern climate theory has a problem explaining past abrupt climate changes and has developed a vague explanation that uses concepts from chaos theory about thresholds that are crossed and tipping points that are reached when a forcing gradually increases over a noisy chaotic background. The problem is that there is no evidence for the existence of such thresholds and tipping points other than the existence of the abrupt climate changes that they try to explain. Theoretical positive feedbacks are also invoked, but the general climate stability that has been life-compatible for the past 450 Myr indicates that it is a system dominated by negative feedbacks. As is usually the case with a troubled paradigm, it takes refuge on the least known aspects of climate, like the importance of the poorly measured themohaline circulation for climate change, finding some support in general circulation models, but not on the observational evidence, that suggests AMOC is a lot more stable than previously thought (Worthington et al. 2021), and does not appear to depend much on deepwater formation (Lozier 2012).
Besides abrupt climate events that took place centuries or millennia ago, current climate also undergoes rapid regime shifts every few decades. The regime shift concept was developed in ecology to explain rapid transitions between alternative stable states, mainly in grazing ecosystems. Lluch-Belda et al. (1989) used the concept to explain the alternation between sardine and anchovy regimes simultaneously in several of the world oceans, possibly in response to climate change. Their data showed that at least two shifts between sardine and anchovy regimes had taken place during the 20th century prior to the 1980s.
4.2 The climate shift of 1976-77
At the 1990 7th Annual Pacific Climate Workshop, Ebbesmeyer et al. (1991) presented a study demonstrating that in 1976 the Pacific climate had undergone a step change in 40 environmental variables, including air and water temperatures, the Southern Oscillation, chlorophyll, geese, salmon, crabs, glaciers, atmospheric dust, coral, carbon dioxide, winds, ice cover, and Bering Strait transport. The changes suggested that one of Earth’s largest ecosystems occasionally undergoes abrupt shifts. Nicholas Graham (1994) analyzed the abrupt changes that took place in the boreal winter circulation over the Northern Hemisphere (NH) and in the coupled ocean/atmosphere system of the tropical Pacific and concluded that these changes resemble a muted, quasi-permanent El Niño, that began when the coupled ocean-atmosphere system did not recover fully from the 1976-77 El Niño and were best described as a change in the background climate state. In addition, mid-latitude winter boreal circulation became more vigorous, with a southward excursion of the westerlies, significant changes in geopotential heights and sea level pressure, accompanied by a southward migration of the Aleutian low-pressure center in winter.
Examination of past climate and fishery data from the North Pacific by Mantua et al. (1997) and by Minobe (1997) led to the identification of a 50–70-year climate oscillation that was named the Pacific [inter] Decadal Oscillation (PDO; Mantua et al. 1997). Regime shifts in the PDO were identified in both articles c. 1925, 1947, and 1977. The pan-Pacific coordinated changes in climate and ecological variables were apparent in many sea-surface temperature (SST) and sea-level pressure (SLP) indices, like the Southern Oscillation Index, defined by Gilbert Walker in the 1920s, and known since the 1960s to track atmospheric El Niño-linked changes in the Walker circulation. Mantua and Hare defined the PDO as the leading principal component of an empirical orthogonal function of monthly SST anomalies over the North Pacific (poleward of 20°N; Mantua & Hare 2002). As changes in SLP lead changes in SST by about two months, Shoshiro Minobe (1999) focused on SLP, using the North Pacific Index (Trenberth & Hurrell 1994) that tracks seasonal SLP changes in an ample region of the North Pacific centered on the Aleutian Low. Using this index, Minobe showed that there were two oscillations causing climate shifts. The major oscillation, already identified, had a period of c. 55 years. It affected SLP variability during both winter and spring atmospheric circulation, and presented shifts at c. 1922/23, 1948/49 and 1975/76 (Minobe 1999). The minor oscillation had a period of c. 18 years, and only affected winter circulation. Three periods of the minor oscillation (i.e., shifts at c.1923/24, 1946/47 and 1976/77) nearly coincide in time and sign with pressure changes of the major oscillation (Fig. 4.2).
In Fig. 4.2, the central graph (a) is the wavelet-transform coefficient of the winter North Pacific index as the area-averaged sea-level pressure anomaly (hPa) in the region 160°E−140°W, 30−65°N. It is a three-dimensional representation of the time domain (1899–1997, X-axis), the frequency domain (periodicity, Y-axis), and the amplitude of the changes (color scale, hPa) of the pressure changes in a region centered in the Aleutian Low. Blue color indicates a deeper Aleutian Low associated with PDO positive phases. The thin black-solid, black-dashed and grey contours indicate the significance at the 95, 90 and 80 % confidence levels, respectively. The plot is from Minobe 1999. The thick black line, left scale, (b) is the phase and amplitude sine wave of the diurnal lunar nodal cycle K1 tidal constituent. It has been overlain to show that both the phase and period of the bidecadal component in the instrumental record is that of the 18.6-year lunar nodal cycle. The sine wave is after McKinnell and Crawford (2007). The dates of Pacific climate regime shifts (c) are shown as vertical lines, as identified by Mantua et al. (1997).
The North Atlantic also presents a multidecadal oscillation, the AMO, and a bidecadal one (Frankcombe et al. 2010). The relationship between the bidecadal and multidecadal oscillations remains unclear. A subharmonic relationship is unlikely despite their coupling. In the North Pacific they have a different seasonal dependency, and in the North Atlantic the bidecadal oscillation is best seen in subsurface temperatures, while the multidecadal one affects mainly surface temperature and Arctic deep-water salinity (Frankcombe et al. 2010). McKinnell and Crawford (2007) propose that the North Pacific bidecadal oscillation is a manifestation of the 18.6-year lunar nodal cycle in winter air and sea temperatures. This lunar cycle strongly affects the magnitude of the lunar diurnal tide constituent (K1) and is synchronized in phase and period to the bidecadal oscillation. In Fig. 4.2 increasing K1 (upward sinusoidal) is associated with decreasing SLP (blue colors) and decreasing K1 with increasing SLP. According to McKinnell and Crawford, the bidecadal component of variability association to the 18.6-yr lunar nodal cycle appears in proxy temperatures of up to 400 years in duration. A tidal cause for the bidecadal oscillation certainly provides an explanation for the subsurface temperature effect in the North Atlantic. Tides provide over half of the energy for the vertical mixing of water in the oceans.
The work of Schlesinger and Ramankutty (1994) made clear that multidecadal variability had effect on global temperature, which also displays a c. 55–70-year oscillation when detrended. Interdecadal oscillations have been described in most oceans, including the Arctic, affecting a variety of climatic phenomena including SST, SLP, sea subsurface temperature, salinity, sea-ice, wind speed, sea-level, and atmospheric circulation. It was necessary to take a global view integrating all this natural variability into a single hypothesis of global multidecadal internal climate change. This is what Marcia Wyatt accomplished when she developed the “stadium-wave” hypothesis in her thesis (Wyatt 2012). She identified a multidecadal climate signal that propagated across the Northern Hemisphere through the indices of a synchronized network (Fig. 4.3). While Wyatt could not identify the nature of the signal, or the cause of its period of c. 64 years, Wyatt and Curry (2014) identified the Eurasian Arctic sea-ice region as the place where the signal was first generated. As we saw in Part III, this is the main gateway for atmospheric winter meridional transport into the Arctic (e.g., see Figs. 3.6 & 3.8b), which is very sensitive to sea-ice.
Fig. 4.3 shows the 20th-century signal propagation through a 15-index-member network. Selected indices are a sub-set of a broader network. Four clusters of indices are highlighted (I through IV, each can be positive or negative). Each cluster is termed a “Temporal Group”. Peak values of Group indices represent stages of climate-regime evolution. The plot is from Wyatt and Curry 2014.
Clearly a hemispherically synchronized multidecadal variability in the ocean-atmosphere coupled system takes place in the NH during winter. Most modern global warming has also taken place since 1976 in the NH during the winter months (Fig. 4.4). It is obvious to anybody endowed with independent thinking that the climate shift that affected the NH winters since 1976 and the global warming that mainly affected the NH winters since 1976 must be causally related. At the very least, natural multidecadal variability must be responsible for an important part of the global warming experienced in the 1976–2000 period. Yet by the time multidecadal warming and climate regime shifts were known to climatologists (in the 1990s–2010s), modern climate theory had already played a trump card assigning the 1976 climate shift to aerosols. As Tsonis et al. write:
“The standard explanation for the post 1970s warming is that the radiative effect of greenhouse gases overcame shortwave reflection effects due to aerosols. However, … the observations with this event, suggests an alternative hypothesis, namely that the climate shifted after the 1970s event to a different state of a warmer climate, which may be superimposed on an anthropogenic warming trend.”Tsonis et al. (2007)
Despite knowing this, modern climate theory has refused to incorporate the effect of climate shifts, which are poorly reproduced and never predicted by models. This sets the theory up for failure as the same trump card cannot be played again when the next shift comes. Can the proponents of modern climate theory ignore a new shift? Or will they recognize the theory’s shortcomings, after committing western economies to a profound decarbonization?
Fig 4.4 shows the Northern Hemisphere and Southern Hemisphere average temperature anomaly for December-February (continuous), March-May (long dash), June-August (short dash), and September-November (dotted) for the 1970–2000 period. The data are from Jones et al. 2016.
4.3 Despite all the people studying climate, the 1997–98 shift went unnoticed
The failure to incorporate climate shifts to the modern climate theory is one of the reasons the climate shift that took place in 1997–98 (97CS) was not noticed and properly described. Another reason is that many of its effects were erroneously assigned to the increasing radiative forcing from climbing GHGs levels and used to raise the level of climate alarm. As the 1976 shift changed the NH climate to a warmer state (Tsonis et al. 2007) by increasing the rate of warming (Fig. 4.4), the 97CS did the opposite and changed the climate state by reducing the rate of warming. Embarrassingly, it was not climate scientists who noticed this change, since it did not fit their biases with regard to increasing GHGs, but it was a geologist and paleontologist skeptical of modern climate theory who first reported it:
“There IS a problem with global warming… it stopped in 1998”(Carter 2006)
After the pause in global warming was identified, hundreds of articles were published on it in scientific journals and a great controversy erupted over its reality, with some authors denying its existence (Lewandowsky et al. 2016) and even altering the official datasets to reduce its significance (Karl et al. 2015), and other authors asserting it was a real phenomenon that needed an explanation (Fyfe et al. 2016).
The pause controversy was a third factor obscuring recognition of the 97CS despite clear evidence of its existence. This factor, together with the absence of climate shifts in modern climate theory and models, and the incorrect attribution of its effects to increasing GHG forcing, kept the obvious conclusion out of the mainstream. Lluch-Belda et al. (1989) identified global sardine and anchovy regime shifts suggesting a climate change cause. These fishery shift points were later identified as Pacific climate shift points that had global teleconnections (Mantua & Hare 2002). Chavez et al. (2003) reported in the journal Science that a new multidecadal regime shift in Pacific fisheries had taken place in the mid-1990s. The warm “sardine regime” had changed to a cool “anchovy regime.” The authors advised (obviously to no avail) that these large-scale, naturally occurring variations should be taken into account when considering human-induced climate change. The 97CS continues to be unrecognized by climate scientists. The next shift will probably occur in the late 2020s to early 2030s. It would be shameful if climate scientists, at that time, are still unprepared for the change, and do not recognize multidecadal variability contribution to modern global warming.
4.4 How the climate shifted globally at the 97CS
Science is so specialized and compartmentalized these days that nobody has pulled together all the evidence that confirms the 1997-98 global climate shift. A shift that is obviously unexplained by changes in GHGs levels and modern climate theory. Solar activity changed from high at solar cycle (SC) 22 to low at SC24 (Fig. 4.5a, black line). This change can be better appreciated in the great decrease in the antipodal amplitude magnetic index (Fig. 4.5a, red line), that measures geomagnetic disturbances caused mainly by the solar wind, to centennial low values. We have already mentioned the famous pause in global warming (Fig. 4.5b), that can be better characterized as a reduction in the rate of global warming after the mid-1990s, and is still ongoing despite its interruption by the strong 2015–16 El Niño, after which no more warming has taken place, as of the middle of 2022.
At the 97CS, a predominantly Niño frequency pattern in ENSO turned into predominantly Niña, as determined by the cumulative multivariate El Niño index (MEI v.2). The summed index showed an increasing trend during the previous climate regime, peaked in 1998, and shows a declining trend after (Fig. 4.5c, black line). Warm water volume at the equator decreased in variability (Fig. 4.5c, red line), and strongly negative anomalies in the warm water volume, that used to happen once a decade, stopped after 2000. A westward shift in atmosphere-ocean variability in the tropical Pacific took place at the 97CS, characterized by a decrease of ENSO variability that coincides with the suppression of subsurface ocean temperature variability and a weakening of atmospheric-ocean coupling in the tropical Pacific. The shift manifested as more central Pacific versus eastern Pacific El Niño events, and a frequency increase in ENSO, linked to a westward shift of the location of the wind-SST interaction region (Li et al. 2019). The changes in the tropical Pacific atmospheric-ocean coupling had a reflection in the stratosphere. Global (60°N-S) stratospheric water vapor decreased abruptly in 2001 (Fig. 4.5d). Simultaneously, the tropical tropopause cooled (Randel & Park 2019), indicating that a step change in the tropical troposphere-stratosphere coupling also took place.
Fig. 4.5 shows series that illustrate the big climatic shift of 1997–98. Many nearly simultaneous changes in climate related phenomena took place globally between 1995 and 2005. Panel (a) shows Oct–Jan sunspots (thin black line) and the 11-yr average Oct–Jan sunspots (thick black line). Solar activity decreased from high (108 sunspots 1980–1995) to low (54 sunspots 2005–2015). The data are from WDC–SILSO. The antipodal amplitude (Aa) geomagnetic index 13-month average is the thin red line, and the 11-year average is the red thick line. The Aa geomagnetic index measures magnetic disturbances mainly caused by the solar wind. The data are from ISGI. In panel (b) the global surface average temperature anomaly in °C is plotted. It displays the 1998–2013 pause in warming. The data are from MetOffice HadCRUT 4.6 annual data. Panel (c) is the cumulative multivariate ENSO index v.2 as a black line. It changed from increasing to decreasing in 1998, indicating a shift in the ENSO pattern. The data are from NOAA. The change is also reflected in a strong reduction in the warm water volume anomaly (WWVa, shown in red) variability at the equator (5°N–5°S, 120°E–80°W above 20 °C), where after 2000 negative values of minus one are no longer reached. The data are plotted in 1014 m3 from TAO Project Office of NOAA/PMEL. The stratospheric water vapor monthly anomaly at 60°N–S, 17.5 km height, from solar occultation data (black line), and microwave sounder data (red line) in ppmv are plotted in panel (d). The large drop in stratospheric water vapor in 2001 occurs at about the same time as a drop in tropical tropospheric temperature (not shown).
Panel (e) plots the cloud cover anomaly, monthly as a thin line, and yearly as a thick line. The anomaly is for 90°S–90°N cloud cover in percent. Data are from EUMETSAT CM SAF dataset, after Dübal & Vahrenholt (2021). Panel (f) is the ensemble mean annual Hadley cell intensity anomaly (in percent from the mean) for the NH from eight reanalyses (black line), and ensemble mean annual-mean Hadley cell edge anomaly (in degrees latitude) for the NH from eight reanalyses (red line). The plot is from Nguyen et al. 2013. Panel (g) is the average annual change in length-of-day (ΔLOD) in millseconds, inverted (up is a shortening in LOD due to acceleration in Earth’s spin). The data are from IERS LOD C04 IAU2000A. Panel (h) is the yearly increase of the 10-year running mean of the ocean heat content (black line), and the annual mean Earth energy imbalance obtained as the difference between the incoming solar radiation and the total outgoing radiation (red line). Both are in W/m2. The plot is after Dewitte et al. 2019. The illustration is from Vinós 2022.
At the 97CS, low cloud cover decreased (Fig. 4.5e; Veretenenko & Ogurtsov 2016; Dübal & Vahrenholt 2021), and Earth’s albedo anomaly reached its lowest point in 1997 and started increasing (Goode & Pallé 2007). The increase was due to increasing high and middle altitude cloud cover. During the 1995–2005 period a tropicalization of the climate took place and the tropics expanded as the Hadley cells increased their extent and intensity (Fig. 4.5f; Nguyen et al. 2013). The atmospheric angular momentum decreased causing the speed of rotation of the Earth to increase, reducing the length of the day by 2 milliseconds (Fig. 4.5g). All these changes altered the energetics of the climate system. The Earth’s energy imbalance, the incoming solar radiation minus the total outgoing longwave radiation (OLR) as measured by the CERES system, started to decrease, (Fig. 4.5h, red line; Dewitte et al. 2019). The global energy change at the 97CS resulted in a change in trend in the ocean heat content (OHC) time derivative (Fig. 4.5h, black line; Dewitte et al. 2019). This change indicates OHC started to increase more slowly, which dismisses claims that the missing heat resulting from the pause in warming was going into the oceans (Chen & Tung 2014).
These are some of the global climate variables that display a rapid change at, or soon after, the climate regime shift identified by Chavez et al. (2003) as a transition from a “warm” sardine to a “cool” anchovy regime, the opposite of the 1976 shift that was identified in the 1990s. Twenty-five years after the recognition of climate shifts, the 97CS still is not acknowledged by modern climate science. That the 76CS has been recognized and the 97CS has not is a strong sign that modern climate theory is an obstacle to climate change understanding and is causing scientists to dismiss facts that are inconsistent with their theory.
4.5 The Arctic shift and polar amplification
When the global climate shifted in 1997–98, the Arctic climate was strongly affected. In Part III, when reviewing how heat is transported during the winter to the Arctic, we mentioned that little Arctic amplification had taken place by 1995 despite two decades of intense warming, echoing the words of Curry and colleagues:
“The relative lack of observed warming and relatively small ice retreat may indicate that GCMs are overemphasizing the sensitivity of climate to high-latitude processes.”Curry et al. (1996)
At the 97CS, Arctic amplification increased greatly and suddenly, but displayed a striking seasonality. Arctic summer temperatures are not increasing (Fig. 4.6a, black line). Any increase in net heat transported in summer to the Arctic is in a great part stored, by warming the ocean and melting ice and snow, until the arrival of the cold season when it is returned to the atmosphere, by the reverse process. Winter surface temperature shows a very pronounced increase since c. 1998 (Fig. 4.6a, red line).
The effect of the 97CS on Arctic sea-ice extent was spectacular. Between 1996 and 2007 September Arctic sea-ice extent decreased by a whopping 45 % (Fig. 4.6b), leading to fears among experts that it had entered a death-spiral (Serreze 2010). But after 11 years of loses Arctic sea-ice adapted to the new regime and 14 years later September Arctic sea-ice extent was higher than in 2007. Since sea-ice loss was used as a poster child of enhanced greenhouse warming and Arctic amplification, and used to raise alarm and money, it cannot now be properly attributed to the 97CS without losing face. The reduction in Arctic sea-ice was accompanied at the 97CS by an increase in Greenland meltwater flux (Fig. 4.6c, black line), and a decrease in Greenland ice-sheet mass balance (Fig. 4.6c, red line), that display the same dynamics of rapid change in the years after the climatic shift followed by a stabilization to the new regime levels.
Fig. 4.6 shows the data characteristics of the big Arctic climatic shift of 1997–98. Panel (a) shows the 80–90°N summer (JJA, black line), and winter (DJF, red line) temperature anomalies in °C. The data are from the Danish Meteorological Institute. Panel (b) shows the September average Arctic sea-ice extent (106 km2). The data are from NSIDC. Panel (c) is the Greenland freshwater flux (black line, km3). The data are from Dukhovskoy et al. 2019. The red line is the Greenland ice-sheet mass balance in Gt, after Mouginot et al. 2019. Panel (d) is the Arctic Ocean Oscillation (AOO) index. It reflects the alternation between sea-ice and ocean anticyclonic circulation (blue bars) and cyclonic circulation (red bars). It is after Proshutinsky et al. 2015. Panel (e) shows the number of NH blocking events per year, after Lupo 2020. Panel (f) is the winter (DJF) latent energy transport across 70°N by planetary scale waves, in PW. The thin line is the annual change, and the thick line is the 5-year moving average. The plot is after Rydsaa et al. 2021. The illustration is from Vinós 2022.
Since the 97CS is unrecognized, scientists cannot explain many of the altered climatic parameters. This is true of the Arctic Ocean Oscillation index (AOO; Fig. 4.6d), defined by Andrey Proshutinsky (2015) as the oscillation between cyclonic (anti-clockwise) and anticyclonic (clockwise) ocean circulation in the Arctic Beaufort gyre, with a period of 10–15 years. The problem is that during the 97CS the oscillation stopped and the system got stuck in the anticyclonic regime, which leads to freshwater accumulation in the Arctic. 1996 was the last cyclonic AOO year, as of late 2022. Proshutinsky has no explanation and the index stopped being updated in 2019, however he became worried that the increasing Beaufort gyre freshwater accumulation is a “ticking time bomb” for climate. The accumulation may lead to a salinity anomaly in the North Atlantic with a magnitude comparable to the Great Salinity Anomaly of the 1970s, that traveled the sub-polar gyre currents from 1968 to 1982 and may have contributed to the early 1970s cooling.
Additional unexplained changes in the Arctic climate at the 97CS include the increase in winter blocking events, particularly in the NH (Fig. 4.6e). We also reviewed, in Part III, how blocking conditions stop the normal westerly zonal circulation at mid-latitudes during winter. They have two outstanding effects. They stabilize weather patterns for days over the same location, leading to extreme weather events in temperature and precipitation; and they also greatly increase MT towards the Arctic since they deflect cyclones poleward. It is clear, but not explained, that MT towards the Arctic increased at the 97CS, and this is the underlying cause of many of the changes observed afterward in the Arctic climate. Evidence for the increase in winter heat and moisture transport into the Arctic comes from the increase in planetary-scale latent heat transport (Fig. 4.6f), while synoptic scale latent heat transport decreased during winters, but increased during summers (Rydsaa et al. 2021). The increase in winter heat and moisture transport into the Arctic leads to higher cloud formation, which shifts the strongest radiative cooling from the surface to cloud tops, which are frequently warmer in winter due to temperature inversions. At the sea-ice border, winter heat intrusions cause a temporary retreat of the ice margin, leading to enhanced heat loss by the ocean until the ice forms again (Woods & Caballero 2016).
Arctic amplification has turned out to be mainly a cold season phenomenon that started between 1995–2000 for reasons unknown to most climate scientists and models. Arctic amplification is dependent on changes in MT, and the rate of Arctic amplification appears to be opposite to the rate of global warming.
4.6 Climate regimes as a meridional transport phenomenon affecting planetary energetics
From the effects of climate shifts it is evident that they affect the global MT system, and particularly the boreal winter MT. As we reviewed in Part III, ENSO is a way of extracting surplus heat from the deep tropics that exceeds the regular oceanic transport system. At the 97CS this need decreases as the Brewer-Dobson circulation (BDC, the stratospheric MT) becomes more active driving more heat out of the deep tropics, causing a cooling at the tropical tropopause that results in more stratospheric dehydration. Also, meridional wind circulation becomes stronger at the expense of zonal circulation resulting in Earth’s rotation deceleration and Hadley cells expansion. As meridional moisture transport to the poles is enhanced with the increase in meridional wind circulation, cloud cover decreases in the low and mid-latitudes, and increases in the Arctic.
In the Arctic the effects of climate shifts through changes in MT intensity are even more evident. At the 97CS, MT to the Arctic was enhanced all year round, but more strongly during the cold season. The rise in heat and moisture advection from lower latitudes results in a reduction in sea-ice cover that augments ocean heat loss, and increases cold season (but not summer) surface temperature. The main effect of winter warming is to increase the radiative loss to space. As we saw in Part III, the Arctic in winter is very special in terms of GHE. The atmosphere is extremely dry, so there is little water vapor GHE. Cloud cover is also quite low during the winter, and the increase in CO2 has the effect of increasing radiation to space from warmer, higher CO2 molecules (van Wijngaarden & Happer 2020).
When there is an intense intrusion event of moist warm air into the Arctic in winter the usual result is a temperature inversion, and despite increased downward longwave radiation, radiative cooling continues from the top of the inversion or the clouds until the advected moisture is either precipitated or exported back to lower latitudes. In essence, more heat transported to the Arctic in the winter must result in more heat lost to space. This conclusion contradicts one of the basic pillars of modern climate theory that states that MT is not a climate forcing since horizontal transport does not affect the amount of energy within the climate system, and therefore is not a cause for climate change. This is the most fundamental of the many mistakes of modern climate theory, as it assumes the top of the atmosphere behaves similarly in terms of GHE everywhere. It does not, as the GHE is very weak at the polar regions, particularly during the long cold season. Transporting more heat from a region of high GHE to a region of low GHE results in more heat being lost at the top of the atmosphere without a compensating gain elsewhere. A change in the intensity of MT towards the winter pole results in a change in the planet’s energy budget as we have shown (Fig. 4.5h).
In Fig. 4.7 the thin grey line is the 7-month average of the monthly mean OLR anomaly in W/m2 from the interpolated OLR NOAA dataset. The thick black line is the 5-year average of the cold season (Nov–Apr) mean. The thick black dashed line is the 5-year average of the summer (JJA) mean. The grey box highlights the Arctic shift in OLR between mid-1996 and late 2005. The time of the Pinatubo eruption is identified. Data from KNMI explorer. The illustration is from Vinós 2022.
OLR in the Arctic is higher during the summer than during the cold season as could be expected from the near permanent summer insolation and higher surface temperature. However, at the 97CS OLR increased a lot more during the cold season than during the summer (Fig. 4.7). Clearly MT became stronger, particularly during the boreal winter. Increased summer transport resulted in more energy storage through enhanced summer melting. Winter refreezing of the melted water returns the summer energy to the atmosphere, only to be lost to space through radiative cooling. Now we understand why Arctic amplification is a winter phenomenon that is not related to global warming, and in fact is where the energy for the “Pause” is going. Arctic amplification is not a GHG effect, but a MT effect that results in planetary cooling. The pause is continuing because Arctic amplification is ongoing. When the pause ends the Arctic should cool and sea ice should grow. As stated previously this could happen by late 2020s to early 2030s, when the next climate shift occurs.
4.7 Meridional transport modulation of global climate
To analyze the multidecadal changes in MT since 1900, the 1912–2008 period has been subjectively divided in three phases of 32 years. Although the different modes of variability do not shift simultaneously (hence the name stadium-wave), the phases so defined describe periods of alternating prevailing conditions in MT well. Starting in the Arctic, where the PV strength determines the polar stratosphere-troposphere winter coupling, the Arctic oscillation (AO; Fig. 4.8a, grey line) is the leading mode of extratropical circulation variability in the NH (Thompson & Wallace 2000). To act as a North-South seesaw of atmospheric mass exchange between the Arctic and mid-latitudes, the AO requires a correlation between its three centers of action —the Arctic, Atlantic and Pacific sectors. The Arctic-Atlantic correlation is known as the North Atlantic Oscillation (NAO), and is strong, the Arctic–Pacific linkage is weaker, casting doubts about the AO being a true annular mode. However, the Aleutian Low and the Icelandic Low have had a negative correlation from one winter to the next since the mid-1970s (Honda & Nakamura 2001). This Aleutian–Icelandic seesaw appears to depend on the propagation of stationary waves and varies in strength with changes in PV strength (Sun & Tan 2013). By calculating the Jan–Feb cumulative AO (Fig. 4.8a, grey line) we can see that until c. 1940 positive AO values (i.e., strong vortex conditions) prevailed, but in the 1940–1980s period negative AO values were more common, only to change back afterwards. The Aleutian–Icelandic seesaw confirms the changes in PV strength with its 25-year moving correlation (Li et al. 2018; Fig. 4.8a black line). When the PV is strong, the mass and heat exchange between the mid-latitudes and the Arctic is smaller, as the PV acts as a barrier to meridional circulation.
In Fig. 4.8, panel (a) shows the polar vortex strength. The black line is the Aleutian Low–Icelandic Low seesaw 25-year moving correlation as a proxy for polar vortex strength. The plot is after Li et al. 2018. The grey line is the cumulative winter (Dec–Feb average) Arctic Oscillation index. The data are the 1899–2002 AO index from DW Thompson. Dept. of Atmos. Sci. CSU (Thompson & Wallace 2000). In panel (b), the black line is the 4.5-year average of the Atlantic Multidecadal Oscillation index. The data are from NOAA and unsmoothed from the Kaplan SST V2. The grey line is the cumulative 1870–2020 detrended cold season (Nov–Apr average) North Atlantic Oscillation index. The data are from CRU, U. East Anglia, Jones et al. 1997. Panel (c) is the cumulative PDO. It is the 1870–2018 detrended annual average cumulative PDO index from HadISST 1.1, and the data are from NOAA. The black dots mark the years 1925, 1946, 1976 and 1997 when PDO regime shifts took place (Mantua & Hare 2002; see Sect. 11.4).
The black line in panel (d) shows the zonal atmospheric circulation index, cumulative anomaly. It is after Klyashtorin & Lyubushin 2007. The grey line in panel (d) is the 1900–2020 inverted and detrended annual ∆LOD. The data is in milliseconds from IERS. Panel (e) is the detrended 1895–2015 annual global surface average temperature, 10-year averaged. The data are from Met Office HadCRUT 4.6. The panel (f) dashed line is the 8.2–16.6-year band-pass of the monthly mean total sunspot number. The data are from WDC–SILSO. The grey line is the 6.6–11-year band-pass of the monthly AMO index. The black line is the inverted 20-year running correlation of the band-pass sunspot and AMO data. The black dots indicate climate shifts, as in c, showing their position with respect to solar minima. The illustration is from Vinós 2022.
The AMO measures SST anomalies that reflect the strength of MT over the North Atlantic. Positive AMO values indicate warm water accumulation due to reduced MT and strong PV conditions (Fig. 4.8b, black line). The NAO is the sea-level pressure dipole over the North Atlantic, and part of the AO. Not surprisingly, its detrended and cumulative value is very similar to that of the AO, but also shows some correlation to the AMO SSTs (Fig. 4.8b, grey line). The decades-long NAO trends cannot be explained by general circulation models as they do not incorporate multidecadal MT regimes. Models consider NAO indices white noise without serial correlation (Eade et al. 2021). Without properly representing MT, climate models cannot explain climate change. Over the Pacific sector, the PDO also measures SST anomalies. A positive PDO indicates warm water accumulation over the equatorial and eastern side of the Pacific, an indication of reduced MT, which moves heat out of the equator and towards the western Pacific boundary so the Kuroshio current can move it northward and transfer it to the atmosphere. The detrended cumulative PDO values (Fig. 4.8c) show that the phases of increased or decreased Pacific MT roughly coincide with those of the Atlantic. Climatic and ecological shifts in the Pacific identified in 1925, 1946, 1976 and 1997 (Mantua & Hare 2002) coincide with times when the PDO shifts from predominantly positive to negative or back (Fig. 4.8c black dots).
The meridional wind circulation is how most of the tropospheric MT is carried out and increases in MT imply increases in meridional circulation and corresponding decreases in zonal circulation. The atmospheric circulation index is a cumulative representation of the yearly anomaly in zonal (E–W) versus meridional (N–S) air-mass transfer in Eurasia (Klyashtorin & Lyubushin 2007). Periods when the NH PV has been stronger and MT over the North Atlantic and North Pacific sectors has been lower (grey areas in Fig. 4.8) coincide with periods characterized by predominant zonal-type anomalies, while periods of weaker PV and higher MT present predominant anomalies of meridional-type (Fig. 4.8d, black line). These persistent changes in predominant atmospheric circulation patterns produce changes in the transfer of momentum between the atmosphere and the solid Earth–ocean affecting the Earth’s rotation speed, measured as changes in the length of day. Periods of increasing zonal circulation correlate with an acceleration of the Earth and a decrease in ∆LOD (inverted in Fig. 4.8d, grey line) while periods of decreasing zonal circulation correlate with a deceleration of the Earth and an increase in ∆LOD (Lambeck & Cazenave 1976). Changes in the rate of rotation of the Earth integrate global changes in atmospheric circulation that support the global effect of MT changes. We must remember at this point that changes in Earth’s rotation rate respond to changes in solar activity (see Fig. 2.5).
Multidecadal changes in MT are the cause of the multidecadal oscillation known as the stadium-wave, and all its manifestations. SST changes in the AMO and PDO are a response to changes in global atmospheric circulation. A reduction in atmospheric meridional circulation and the corresponding increase in zonal circulation mean less poleward energy transport, and since annual incoming energy is near constant and ocean heat transport is only partially dependent on wind-driven circulation, more heat accumulates at each latitudinal band, but particularly in the NH mid-latitudes. This is because sea surface transfer of energy and moisture to the atmosphere is highest at NH mid-latitude ocean western boundaries (Yu & Weller 2007). Land and sea surface heat accumulation resulting from a reduction in MT produces the stadium-wave effects and an increase in the global temperature. When the global average surface temperature anomaly is detrended, periods of reduced (increased) MT correspond to warming (cooling) with respect to the trend (Fig. 4.8e). The modern climate theory explains the 1940–1975 hiatus as due to an increase in aerosols, and the 1976–2000 warming as due to the increase in anthropogenic emissions. These explanations, incorporated into climate models, are untenable in light of the evidence (Tsonis et al. 2007). Although an anthropogenic warming trend is unquestionable, it is evident that the shifts in MT regimes dominate the surface temperature response.
The causes behind the multidecadal stadium-wave changes in MT are unknown. The c. 65-year oscillation is non-stationary. Proxy reconstructions indicate that the AMO had a shorter periodicity and less power during the LIA and a longer periodicity and more power during the Medieval Warm Period (Chylek et al. 2012; Wang et al. 2017). Solar activity modulation of ENSO and Earth’s rotation changes were shown in Part II (Figs. 2.4 & 2.5). As both are a manifestation of MT strength, it is possible that internal variability and external solar forcing are responsible for the current periodicity and strength of the stadium-wave. Alternatively, internal variability in MT might be responding to the warming trend imposed by anthropogenic and natural causes, mainly the increase in solar activity associated with the modern solar maximum. The four climate shifts identified in the Pacific during the 20th century (Mantua & Hare 2002) took place 1–3 years after a solar minimum (Fig. 4.8c & f, dots; solar cycle, Fig. 4.8f dashed line), and the two grey areas and middle white area in figure 4.8, representing alternating MT regimes, span three solar cycles between solar minima. It has been shown that the Holton–Tan effect (see Part I), that relates the tropical QBO phase to the strength of the PV, through planetary wave propagation, is stronger at solar minima (Labitzke et al. 2006), and that the Holton–Tan effect weakened substantially during the 1977–1997 period of reduced MT (Lu et al. 2008). This implies that during winter at solar minima the stratospheric tropical-polar coupling, and the stratospheric-tropospheric coupling are stronger, and they might constitute an appropriate time for a coordinated shift in MT strength that takes effect during the ensuing solar cycle. We shall see if future climate shifts also take place immediately following solar minima. This is the basis of our projection that the next climate shift could take place around 2031–34.
If solar minima are the times when MT shifts occur, one interesting correlation may provide an explanation for the cause of the c. 65-year oscillation pacing. The AMO has a 9.1-year strong frequency peak that is also found in the PDO (Muller et al. 2013). This frequency is readily appreciated in a 4.5-year averaged AMO index as decadal bumps (Fig. 4.8b, black curve). The origin of this conspicuous AMO trait has not been adequately researched, but Scafetta (2010) convincingly proposes a lunisolar tidal origin. The difference in frequency between this reported 9.1-yr tidal cycle and the 11-yr solar cycle is such that they change from correlated to anti-correlated (i.e., constructive to destructive interference) with a periodicity that not only matches the AMO, but is exactly synchronized to it (compare black curves in Fig. 4.8b & f). One can speculate that a constructive or destructive interference between the effect of oceanic and atmospheric tides on the tropospheric component of MT and the effect of the solar cycle on the stratospheric component of MT might result in the periodical change in MT strength that produces the observed climatic shifts. In support of this hypothesis two intrinsic components of c. 4.5 and 11 years are found in the Fourier analysis of the daily NAO autocorrelation series (Álvarez–Ramírez et al. 2011). The 11-year component is phase synchronized to the solar cycle except during solar minima, indicating that NAO predictability increases with solar activity, and became strongly anti-correlated during the 1997 solar minimum, when the 97CS took place. A c. 65-yr climate oscillation that depends on solar activity would explain both the changes in intensity and periodicity over the last centuries as solar activity has been changing. Its 20th century intensity and periodicity are the result of the modern solar maximum, and the non-stationarity of the natural multidecadal oscillation would be linked to solar activity multidecadal variability.
It can be argued that multidecadal oscillations in the climate system should average to zero over multiple periods. Similarly, other factors known to affect MT, like the QBO and ENSO average to zero in similar or shorter timeframes. However, AMO reconstructions show that its values and amplitude have increased greatly over the last two cycles, since about 1850 (Moore et al. 2017). This change in the c. 65-year oscillation suggests that MT is important in modern global warming, since it coincides with the strong melting of glaciers and increase in sea-level rise that started around 1850 and precedes the strong increase in CO2 emissions after 1945 (Boden et al. 2009). Solar activity affects MT and does not average to zero even in very long timeframes because it presents centennial and millennial cycles (Vinós 2022). There has been a long-standing scientific debate about whether there is an important effect of solar activity on climate. Sunspot records show that the average number of sunspots increased by 24% from the 1700–1843 to the 1844–1996 period (see Fig. 1.6). Solar variability is clearly involved in MT variability (see Part II). The effect that solar variability has on MT, and the effect that MT has on the planet’s energy imbalance (Figs. 4.5h & 4.7) settles the controversy on the solar activity effect on climate.
In the next part of this series, the hypothesis of how solar variability affects MT will be presented. It has been named the Winter Gatekeeper hypothesis because solar activity modulates the amount of heat that is transported to the poles in winter, and through it the planet’s energy budget, constituting the main climate change modulator on centennial to millennial timescales, as suggested by paleoclimatological evidence.
The earlier parts of this series on Meridional transport and the Winter Gatekeeper hypothesis:
Part 1: The Search for a solar signal.
Part 2: Solar activity and climate, unexplained and ignored.
Part 3: Meridional transport of energy, the most fundamental climate variable.
32 thoughts on “The Sun-Climate Effect: The Winter Gatekeeper Hypothesis (IV). The unexplained/ignored climate shift of 1997”