Good Scientists Embrace Competing Working Hypotheses
How Much Does the Arctic Iris Effect Explain Recent Warming, 1930s Warming and Dansgaard-Oeschger Events?
During the last Ice Age, Greenland’s average temperatures dramatically rose on average every 1500 years by 10°C (+/- 5°C) in a just matter of one or two decades, and then more gradually cooled as illustrated in Figure 1 below (8 of the 25 D-O events are numbered in red on upper graph; from Ahn 2008). These extreme temperature fluctuations between cold “stadials” that lasted about a thousand years and warm “interstadials” lasting decades are dubbed Dansgaard-Oeschger events (D-O events). These rapid temperature fluctuations not only rivaled the 100,000-year fluctuations between maximum glacial cold and warm interglacial temperatures but D-O warm events coincided with expanding Eurasian forests (Sánchez Goñi 2008, Jimenez-Moreno 2009), northward shifts of subtropical currents along the California coast (Hendy 2000), and shifts in belts of precipitation in northern South America (Peterson 2001).
Just 25 years ago most climate researchers were hesitant to accept initial Greenland ice core evidence that suggested such abrupt D-O warming events (Dansgaard 1985). But as other Greenland ice cores verified their reality, it was clear that the only mechanism realistically capable of producing such abrupt warming was the sudden removal of insulating sea ice that allowed ventilation of heat previously stored in the Arctic as Dansgaard (1985) had first proposed. Still that raised the question ‘what caused the sudden loss of insulating sea ice’?
Changes in CO2 concentration are unlikely to have had much impact on D-O events (3rd graph from the top in Figure 1). CO2 concentrations did fluctuate by about 20 ppm during a third of the D-O events (red numbers), but could have contributed directly to no more than 0.4°C or only 30% of the largest warming events. In contrast during 68% of the other D-O events (not numbered), abrupt warming occurred while CO2 was declining. Thus D-O rapid warming and cooling events seem independent of any CO2 forcing.
Abrupt D-O warming and cooling suggested to some researchers (Broecker 1985) that the Atlantic Meridonal Overturning Circulation (AMOC) turned “on” and “off”. Based on the misleading belief in the existence of a simplistic “ocean conveyor belt” (Wunsch 2007), researchers incorrectly interpreted a lack of deep-water formation as evidence of a lack of warm water flowing into the Arctic. However based on increasing proxy evidence (Rasmussen 2004, Ezat 2014), it is now understood that the inflow of warm Atlantic Waters never “shut off” but continued to enter the Arctic and warmed the subsurface layers. As seen in Figure 2 (from Itkin 2015) the upper layer of fresh water and the halocline insulate inflowing warm Atlantic water from the overlying ice. Together the thick sea ice and polar mixed layer simply “turn off“ any heat flux from the ocean to the air, thus maintaining cold stadial air temperatures. Coincidentally if the salty Atlantic Water cannot be sufficiently cooled by the cold Arctic air, then formation of North Atlantic Deep Water is reduced as well, in agreement with proxy evidence during the last Ice Age.
Although climate models have failed to simulate D-O events, models can be manipulated to shut off poleward heat transport by prescribing ad hoc floods of freshwater. Whenever models employed freshwater “hosing”, the models rported intrusions of North Atlantic waters were prevented prevented from cooling and sinking, which shutoff deep water formation, and thus the “ocean conveyor belt” resulting in contrived cooling. That scenario became the reigning paradigm and researchers began searching for evidence of a flood of freshwater, to justify why nearly every model employed freshwater “hosing” to explain abrupt climate change. But evidence of the required freshwater flooding has yet to be found and a growing wealth of proxy evidence suggests there was as much freshwater during stadials as there was during interstadials. Even the notion of freshwater floods from an armada of melting icebergs was not consistent with the timing of D-O events (Barker 2015). Suggestions of a freshwater shutdown of the Atlantic Meridonal Overturning Circulation is most likely a figment of the models’ configuration.
Other researchers suggested drivers of past and present rapid temperature change were likely to be very similar (Bond 2001, 2005), and recent findings are now supporting that notion. More recent explanatory hypotheses for D-O events are gaining widespread critical acceptance and do not require any massive floods of freshwater nor a shutdown of the AMOC (Rasmussen 2004, Li 2010, Peterson 2013, Dokken 2013, Hewitt 2015).
When sea ice prevents heat ventilation, the inflow of warm and dense Atlantic Waters continues to store heat in the subsurface layers. As heat accumulates, the warm Atlantic Waters became more buoyant, upwells and melts the insulating ice cover. The loss of an insulating ice cover “turns on” the heat flux to the atmosphere causing a dramatic rise in surface temperatures to begin a D-O interstadial. Although details of hypothesized D-O mechanisms vary slightly, they all agree on the ability of growing and shrinking sea ice to affect the heating and cooling of the northern hemisphere. I refer to this sea ice control of Arctic Ocean heat ventilation as the Arctic Iris Effect.
The signature of an Arctic Iris Effect is the opposing temperature trends in the ocean versus atmosphere. When ice is removed, warmer air temperatures coincide with cooler ocean temperatures. When ice returns cooler air temperatures coincide with a warmer ocean. The thicker the sea ice, as during the last Ice Age, the longer the period between ventilations such as the D-O events. Thick sea ice is less sensitive to small changes in insolation and/or natural variations of inflowing Atlantic Waters. As discussed in Hewitt 2015 decreases in the freshwater layer that separates sea ice from the warm Atlantic Waters are also likely critical contributors to D-O events. For example as the Laurentide Ice Sheet grew, sea levels fell shutting of the inflow of fresher Pacific water through the Bering Strait, coinciding with an increased frequency between D-O events from 8 thousand to 1.5 thousand years.
Peterson 2013 suggested that in addition to thick multiyear sea ice, ice shelves were critical for maintaining the longer cold stadials by better resisting small oscillations of increased inflow of Atlantic Water. Likewise with reduced multiyear sea ice during our present interglacial, and shallow shelves where warm water can be more readily brought to the surface, much smaller changes in insolation and/or Atlantic inflow can more easily initiate ventilation events. With smaller time spans between each ventilation event, less heat accumulates and warm spikes are more muted (1°C to 2°C) compared to 10°C +/- 5°C during the D-O interstadials. Over the past 6000 years, decades of rapid ice loss resulted air temperatures in 2°C to 6°C warmer than today quickly followed by centuries of colder temperatures and more sea ice (Mudie 2005).
The 20th century ventilation events have produced only a 1°C to 2°C increase yet the signature of the Arctic Iris Effect is still observed. In 2001, Dr. Vinje of the Norwegian Polar Institute reported on the opposing temperature effects as ice retreated in the Nordic Seas. Between 1850 and 1900 the subsurface temperature of ice covered Nordic Seas rapidly warmed by 0.5°C between 1850 and 1900 despite very little change in atmospheric temperature. This was followed by a rapid atmospheric warming for which Vinje reported “The warming event during the first decades of this century [1900-1930] is characterized by a significant decrease in the Nordic Seas’ April ice extent, an increase of ~3°C in the Arctic surface winter temperature, averaged over the circumpolar zone between 72.5° and 87.5°N, and an increase in the Spitsbergen mean winter temperature of as much ~9°C.”
During this warming event the subsurface ocean temperature was lower than normal.
“An increasing preponderance of positive ice extent anomalies, with an optimum in the 1960s, was observed during the period 1949–66, concurrent with atmospheric cooling in the circumpolar zone of ~1°C, a fall in the Spitsbergen mean winter temperature of ~3°C, and an increase in the mean winter air pressure in the western Barents Sea of ~6 hPa.”
During this atmospheric cooling event the temperature in ocean waters were higher than normal.
Similarly the most recent Arctic warming again reveals the fingerprint of the Arctic Iris Effect. There was no atmospheric warming in Arctic when there was an insulating cover of multiyear sea ice. Measurements between 1950 and 1990 reported a cooling Arctic atmosphere prompting researchers to publish, “Absence Of Evidence For Greenhouse Warming Over The Arctic Ocean In The Past 40 Years”. They concluded, “This discrepancy suggests that present climate models do not adequately incorporate the physical processes that affect the Polar Regions.”
Abruptly rapid Arctic warming began in the 1990s with an initial loss and thinning of Arctic sea ice when the Arctic Oscillation shifted wind directions so that below-freezing winds from Siberia pushed multiyear ice out of the Arctic. Rigor 2002 correctly pointed out, “One could ask, did the warming of SAT [Surface Air Temperatures] act to thin and decrease the area of sea ice, or did the thinner and less expansive area of sea ice allow more heat to flux from the ocean to warm the atmosphere?”
They concluded, “Intuitively, one might have expected the warming trends in SAT to cause the thinning of sea ice, but the results presented in this study imply the inverse causality; that is, that the thinning ice has warmed SAT by increasing the heat flux from the ocean.” [Emphasis Added] That conclusion has been further supported by recent analyses of ocean heat content by Wunsch and Heimbach 2014, two of the world’s premiere ocean scientists from Harvard and MIT. They reported the deep oceans are cooling, suggesting the oceans and atmosphere are still not in equilibrium and oceans are still ventilating heat from below 2000 meters that was stored long ago. Also in their map illustrating changes in the upper 700 meters of the world’s oceans (their Figure shown below), we see the entire Arctic Ocean has cooled between 1993 and 2011, as would be expected from the Arctic Iris Effect. Keep in mind that the warm layer of Atlantic water on average occupies the depths between 100 and 900 meters.
The Earth’s Energy Budget
The Earth’s energy budget depends on a balance between absorbed solar radiation and outgoing infrared radiation. While some atmospheric scientists have focused on a possible energy imbalance created by 2 watts/m2 generated by rising CO2, widespread regions of the ocean absorb and ventilate over 200 watts/m2 of heat each year. As illustrated in Figure 3 (from Liang 2015), the oceans absorb heat (blue shades, in watts/m2) along the equator and over the upwelling zones along the continents’ west coast. Intense tropical insolation and evaporation creates warm dense salty waters that sink below the surface storing heat at depth. Changes in insolation, tropical cloud cover, and ocean oscillations like El Nino affect how much heat the oceans absorb or ventilate. Excess heat absorbed in the tropics is transported poleward. To gain a proper perspective on the importance of heat transport from the tropics to the poles, currently Polar Regions average 30°C colder than the equator. If there was no heat transport, the poles would be 110°C colder than the tropics (Gill 1982, Lozier 2012).
On average, the greatest ventilation of ocean heat happens where heat transportation is most concentrated: along the east coast of Asia over the Kuroshio Current and along east coast of North America along the Gulf Stream. Additionally large amounts of heat are also ventilated over Arctic’s Nordic Seas region, a focal point of the Arctic Iris Effect. A comparison of temperature changes at varying ice core locations from southeast to northwest Greenland, points to this North Atlantic region as the main source of heat ventilated during each D-O event (Buizert 2014). Likewise modeling work (Li 2010) shows that reduced ice extent in this region exerts the greatest impact on Greenland temperatures and snow accumulation rates. And it is in this same region that Vinje 2001 reports the greatest reduction in ice cover coinciding with the rapid changes in Greenland’s instrumental data. While CO2 warming theory would predict the greatest rate of Greenland’s warming to occur in the most recent decades, the Arctic Iris effect would predict a greater rate of warming in the 1920s because thick sea ice from the Little Ice Age would have caused a greater accumulation of heat. Indeed Chylek 2005 reported, “the rate of warming in 1920–1930 was about 50% higher than that in 1995–2005.”
Climate Model Shortcomings
In 2008 leading climate scientists at the University of East Anglia’s Climatic Research Unit published Attribution Of Polar Warming To Human Influence. As seen in their graph below, their models completely failed to account for the observed 2°C Arctic warming event from 1920 to the 1940s, (illustrated by the black line labeled “Obs” for observed). This was a warming event that climate scientists called “the most spectacular event of the century” (Bengtsson 2004). Their modeled results of natural climate change grossly underestimated the 40s peak warming by ~0.8° C, and simulated a flat temperature trend throughout the 20th century as illustrated by the blue line labeled “NAT” for natural. More striking when the models added CO2 and sulfates, the modeled results (red line labeled all) cooled the observed early 20th century warming event further. Despite their failure to model natural events they concluded, “We find that the observed changes in Arctic and Antarctic temperatures are not consistent with internal climate variability or natural climate drivers alone, and are directly attributable to human influence.”
More accurately said, their results only demonstrated that CO2 driven models failed to account for natural climate change, the Arctic Iris Effect and ventilation of ocean heat during the 1930s and 40s.
By all accounts the recent warming of the 1990s and 2000 was likewise a ventilation event that also cooled the upper layers of the Arctic Ocean. The failure to model ventilation driven heat events led to incorrectly attributing recent warming to increasing concentrations of CO2. That failed modeling further led to explanations that reduced albedo effect allowed greater absorption of summer insolation, warming the Arctic Ocean and amplifying temperatures. But observations show the upper Arctic ocean has cooled. Like the 40s peak, it is likely 1990s/2000s ventilation, similarly contributed a minimum of ~0.8° C to the recent rise in Arctic temperatures, and probably much more as the greater reduction in sea ice extent has allowed for much more ventilation.
If climates models are correctly configured, they should be able to reproduce both D-O events and the 1940s ventilation events. We don’t expect model perfection, but turning a massive warming event into a below average cool period is unacceptable. When the modeling community simulates the Arctic Iris Effect more accurately, only then will their attribution of polar warming to human vs. natural factors be trustworthy!
In the past the return of cooler ocean surface temperatures and more sea ice has always been a much slower process than the abrupt warming events. When sea ice is reduced, the winds are suddenly able to mix the ocean’s fresher upper layer with the saltier lower Atlantic Waters disrupting the insulating halocline. But once the halocline and upper layers of freshwater are restored, the cooling is rapid. Until then all the natural factors - lower insolation associated with reduced Atlantic inflows of the past, cooler Arctic seas, a negative North Atlantic Oscillation, and increasing multiyear ice – all suggest the current ventilation event will soon come to a close.
In contrast, those who attribute Arctic warming to rising CO2 predict a continued sea ice death spiral. Some suggest global warming is slowing down the poleward flow of Atlantic Water, but also suggest CO2 warming will offset any cooling effects of that slowdown (Rhamstorf and Mann 2015). Within the next 2 decades, nature should demonstrate how well these competing models and competing interpretations extrapolate into the future. Good scientists always embrace 2 or more working hypotheses, and the Arctic Iris Effect provides a well supported hypothesis with testable results that challenges the robustness of CO2 warming theory.
posted September 1, 2015, edited January 2016 & April 5, 2017
follow on twitter @