Guest essay by Javier*
Arctic sea ice has been on a declining trend since at least 1979, and probably since the bottom of the Little Ice Age. 2007 was a bad year for Arctic sea ice. It got to a low maximum in March, although not as low as the previous year that still holds the record, and then it proceeded to melt an impressive 1.5 million km2 more (15% ice extent data) than a year earlier, reaching values that were not expected to happen until the 2030’s. This gigantic drop, the biggest on record, triggered a concerted campaign on the media that surely didn’t hurt Al Gore’s chances of winning his Peace Nobel Prize over Irena Sedler, a Polish lady that saved 3,000 babies during WWII, that died the next year.
Ever since we have been subjected to a fear campaign over a “dying”, “screaming”, or “in a death spiral” Arctic sea ice. Such campaign chose the polar bear for an icon, as a supposedly immediately threatened species by climate change and more specifically Arctic sea ice melting. It is very ironic that in a letter to the journal Science in 2010 by a group of scientists over their integrity on climate change issues, the editors chose a fake picture of a lone polar bear over a tiny iceberg on an open sea to illustrate it (figure 1).
Figure 1. Announcement (Top) by the Science Journal that the letter to the editors of Science “Climate Change and the Integrity of Science” had been published with a fake picture. The picture (bottom left) was also available with a penguin in case it was needed to illustrate Antarctic melting (bottom right).
The fear campaign increased considerably when in September 2012 the Arctic sea ice took another beating establishing the current low record. That was the year that Greenpeace started its “Save the Arctic” campaign for donations. And while establishing an Arctic natural sanctuary is a worthy goal that deserves the support of all conservationists, the end does not justify the means of collecting money through misrepresentation and fear.
The last episode in this fear campaign has been the curious announcement from the National Snow an Ice Data Center (NSIDC) that “2016 ties with 2007 for second lowest Arctic sea ice minimum“, that has been widely circulated despite being preliminary data. The announcement was dated on September 15th and referred to the minimum reached on September 10th, Although their data shows it taking place on September 7. It is curious that when skeptics are told that the Pause is not significant because it is only 20 years long, the NSIDC is making the headlines with a climate claim based on one day’s data.
But the NSIDC is forced to recourse to daily data because the monthly data supports the opposite interpretation, that Arctic sea ice has been increasing since that fateful September of 2007 (figure 2).
Figure 2. Average Arctic sea ice extent during the month of September between 2007 and 2016 with linear trend.
While for someone worried about Arctic sea ice loss an increase should be better than a decrease, almost everybody would say that a 9 year long trend is not significant. However I’m going to present evidence that suggests that a change of trend might be under way.
So how can we identify a change of trend without having to wait 30 years? Trends are very important in the stock market, so investors have tools to indicate when the chances of a trend change are increasing, and technical analysis of stocks includes a figure, the symmetrical triangle, that looks similar to what Arctic sea ice is showing (figure 3). This figure indicates that the force that was driving the previous trend is debilitating and the new equilibrium of forces is increasingly constraining the values. At some point the triangle is broken and the trend resumed (figure 3 A) or a new trend started (figure 3 B). With this figure it is important to wait for a confirmation of the breaking, because false breaks do happen, as in 2012.
Figure 3. Maximum (March) and minimum (September) Arctic sea ice extent trends according to the Ocean and Sea Ice Satellite Application Facilities (OSI SAF) of EUMETSAT. Purple and blue lines help define triangles of progressively decreasing range variation with two possible scenarios. Scenario A is a downward break with a continuation of the previous trend, while scenario B is an upward break with the confirmation of a new trend. September 2012 constitutes a false break because it was not confirmed afterwards. Source: OSI SAF Ice Graphs.
Technical analysis has a weak predictive power, that is why it is used, and while a symmetrical triangle cannot predict the direction of the breakup or its timing, it does predict a continuation of the figure until the break happens. Based on that, last June when Arctic sea ice set a new record low for May, I had the following to say here at WUWT:
“Arctic sea ice is getting constrained in its variation, waiting for a definitive break of one of the lines. We will know in just a few years. In the meantime it is more probable that we do not have a record low in summer Arctic sea ice this year. Once again the alarmists have gone to the newspapers too soon”. Javier at WUWT, June 14, 2016.
That is the reason they had to rush the “tied with 2007” claim to the media with preliminary data, because the monthly data was going to show different and all the baseless hoopla of June was going to be exposed.
Although I got it right, obviously this is not scientific method. Even if I personally will be convinced that a change of trend has taken place if the triangle is broken towards the up side and confirmed, we will be needing empirical evidence that the situation for the Arctic sea ice has changed. However we already have a consilience of evidence supporting the claim that a change of trend is taking place in Arctic sea ice.
First it is not only ice extent that it is increasing since 2007, but also ice age. A lot of importance has been placed on the disappearance of old ice, so it should be a motive for celebration that since 2007 summer ice survival in the Arctic has been on the increase (figure 4).
Figure 4. Summer Ice Survival has also increased since the 2007 minimum. Source: NSIDC.
In a recent article at Climate Etc., “Is the Arctic sea ice ‘spiral of death’ dead?“, Greg Goodman showed that something else has also changed in Arctic sea ice: Since 2007 the date of the minimum extent stopped increasing and started decreasing (figure 5). It appears that since 2007 the melting season might be ending sooner, and the 2016 melting season is significantly shorter than average. According to Tony Heller, 2016 has the shortest Arctic melt season on record, and the previous record was in 2015. Now here are some climate records that are not being reported by the media.
Figure 5. Variation of date of sea ice minimum also shows a change of trend in 2007. Source: Greg Goodman at Climate Etc.
The shortening of the melting season is not the only evidence besides sea ice extent and age that something new is going on in the Arctic since 2007. Arctic waters main exchange is with the North Atlantic Ocean, and since 2007 the heat content of the North Atlantic has been dropping as fast as it increased in previous decades, having already given up all the heat gains since the mid-90’s (figure 6).
Figure 6. North Atlantic heat content anomaly (0-700m) also shows a change of trend in 2007. Source: Climate4you.
But we do not only have strong evidence from ice extent, ice age, melting season length, and North Atlantic heat content, we also have a theoretical framework that links Arctic sea ice to the Atlantic Multidecadal Oscillation (AMO). Miles et al., 2014 show that AMO proxies and Nordic Seas ice proxies, although not a perfect inverted match, display a high degree of synchrony in their phases since at least the 1570’s with sea ice usually slightly ahead of AMO (figure 7).
Figure 7. Persistent multidecadal fluctuations in sea ice linked to the AMO. Original time series (gray) and multidecadal 50–120 year component (blue) reconstructed from wavelet decomposition: (Top) AMO proxy index, not detrended, 10 year running average. (Bottom) Western Nordic Seas sea-ice extent proxy reconstruction. The numbers in parentheses indicate the amount of variance in the nonsmoothed time series that is explained by the multidecadal component. The color bar in bottom panel indicates periods with reduced ice (red) and cold periods with increased ice (blue) inferred from the wavelet-filtered signal. The reduced ice periods are marked by light red shading and are seen to correspond to warm AMO periods. Source: Miles et al., 2014. Red arrows have been added to mark the two big known Arctic melting events of 1920’s and 1980’s.
As Miles et al., 2014 put it:
“We establish a signal of pervasive and persistent multidecadal (~60–90 year) fluctuations… Covariability between sea ice and Atlantic multidecadal variability as represented by the Atlantic Multidecadal Oscillation (AMO) index is evident during the instrumental record. This observational evidence supports recent modeling studies that have suggested that Arctic sea ice is intrinsically linked to Atlantic multidecadal variability.
Given the demonstrated covariability between sea ice and the AMO, it follows that a change to a negative AMO phase in the coming decade(s) could —to some degree— temporarily ameliorate the strongly negative recent sea-ice trends”.
This last phrase, indicating that despite a found natural correlation between Arctic sea ice and AMO, this can only reduce the trends imposed by the overwhelming effect of the atmospheric increase in greenhouse gases, is designed to bow to the current dominant hypothesis to increase the chances of the article being accepted. It does not affect what the evidence showed in the article demonstrates. We are already seeing since 2007 that the natural variability is strong enough to revert the previous trend, not produce some degree of temporary amelioration.
I have already shown here in some of my comments at WUWT that the correlation between AMO and Arctic sea ice is clear also in modern data (figure 8). It is not clear to me what causality underlines this correlation. According to the Stadium Wave hypothesis of Wyatt and Curry, 2014, both Arctic sea ice and AMO are in the same temporal group of the climate signal that travels the Earth’s oceans, sea ice, and atmosphere with a periodicity of 60-90 years. However in Wyatt’s data, as in Miles’ data sea ice changes appear to slightly precede AMO changes. We should not conclude that Arctic sea ice depends on AMO, but that both appear to change together. Wyatt and Curry conclude:
“But according to stadium-wave projections… this trend should reverse… Rebound in WIE [West Eurasian Seas Ice Extent], followed by ArcSib [Arctic Seas of Siberia Ice Extent] should occur after the estimated 2006 minimum of WIE and maximum of AMO.”
Looks like they also nailed it, like Miles et al., 2014, except that as scientists should do, Wyatt and Curry only bowed to the data.
Figure 8. September Arctic sea ice extent (green) inverted and superimposed over Atlantic Multidecadal Oscillation non-detrended anomaly. Source: Arctic sea ice reconstruction Cea-Pirón and Cano-Pasalodos 2016, AMO graph Trenberth and Shea, 2006, updated to 2011 by NCAR.
To contrast all the evidence that we are witnessing a change in Arctic sea ice trend with the official projections of a continuous decline, I have chosen a figure from NSIDC scientist J. C. Stroeve et al., 2012, as displayed in the “Melting Ice” chapter of “Our Changing Climate” report by the National Climate Assessment. Over that figure (figure 9) I have extended the Arctic sea ice reconstruction based on Russian aerial charts to 1935 by Cea-Pirón and Cano-Pasalodos 2016, that should not be very controversial as it lays squarely within the grey band of uncertainty of the original figure. And I have added an exponential fit in red that was very much in fashion between alarmists around 2012 based on Piomas ice volume modeling, and an AMO model fit in orange using the AMO trend and data from figure 8.
Figure 9. Reconstructed (1935-1979), measured (1979-2016), and projected Arctic sea ice extent between 1890-2090. Source of original figure: NCA. Red dashed line, 1979-2011 exponential fit. Orange dashed line, 1935-2015 sinusoidal AMO fit with AMO trend from figure 8. Red box, satellite data window. Forward projections are completely different, and AMO model is the only one predicting two or more decades of no significant Arctic ice melting or even ice growth.
Based on exponential fit or linear acceleration a group of scientists headed by Peter Wadhams and Wieslaw Maslowki, and parroted by Al Gore, defended a complete melting of Arctic sea ice by 2012-2020. They are completely discredited already in scientific circles, which doesn’t prevent them from reaching to the media and contributing to the fear campaign without being contradicted.
Based on accelerating melting, more mainstream ice scientists like Mark Serreze, director of NSIDC, defend 1 million km2 ice free conditions by 2030.
IPCC defends an Arctic essentially free of ice by 2050, and its followers, like notorious commenter Steven Mosher, accept that claim uncritically.
The CMIP5 ensemble of climate models projects an ice-free Arctic by 2060 in the unrealistic RCP 8.5, and by 2080 in the intermediate RCP 6.0 and 4.5 scenarios, as figure 9 shows.
But the evidence indicates that they are all wrong because they have based their analysis in the satellite window that by chance coincides exactly with the downward phase of AMO and Arctic sea ice. There is evidence that a very big Arctic melting took place in the 1920’s (first red arrow in figure 7), and Tony Heller has made a good job of unearthing it from newspapers reports. It is surely reflected in the scientific literature of the time, but climate scientists appear to be too lazy to walk to the library and use the photocopier in these days of internet. By using only recent data and by blatantly ignoring the evidence of the relationship between AMO and Arctic sea ice they have set themselves for failure. Climate models are essentially demonstrating the ignorance of climate modelers of the very same thing they are trying to model. They have either failed to read the relevant scientific literature or failed to understand its consequences.
While a linear extrapolation of the AMO model suggests that the Arctic could be reduced to 1 million km2 by 2200 AD it is unwise to make climate projections from extrapolation over such long periods. Lets just say that there is a significant probability that Arctic sea ice will not significantly melt over the next few decades and could even grow. Whether that is positive or negative depends on personal opinion on the importance of Arctic sea ice.
Empirical evidence supports the notion that a change of trend in Arctic sea ice took place in 2007. Due to that change, Arctic alarmists have been reduced to make climate claims based on daily data. It is very likely that Arctic sea ice might not significantly melt or even grow during the next few decades. The pause in Arctic melting was predicted in the scientific literature by 2014 before it was evident in the data. The pause in Arctic melting that started in 2007 is hereby officially inaugurated.
*Javier is a first name only. He has also published on Dr. Judith Curry’s site. Both she and I know his full identity. While normally we require guest authors use their full name, Javier’s employment situation is such that he’s likely be penalized in some way for posting at Curry’s and especially here. He has satisfied my need and Dr. Curry’s need for this exception, and is a well published scientist in a non climate related field that requires similar statistical analysis skills.
UPDATE: 10/18/16 by Anthony