Inside the Sea ice Anomaly Oscillation (SAO) – Part 1

Guest essay by Craig Lindberg

In a recent post, A Relationship Between Sea Ice Anomalies, SSTs, and the ENSO? (, I introduced the Sea ice Anomaly Oscillation (“SAO”) – an observation that there is an oscillation in the sign and magnitude of the changes in the relationship between the hemispheric sea ice anomalies over time. This post takes a look at one possible internal mechanism of the SAO: sea level pressure (“SLP”).


The SAO is the moving, trailing-356-day regression of the Northern Hemisphere sea ice anomaly vs. the Southern Hemisphere sea ice anomaly (“NH” and “SH” respectively) as calculated from the daily data available at Cryosphere Today / Arctic Climate Research at the University of Illinois (Figure 1).

The SAO appears to have a 32 year period that is almost exactly half that of the Atlantic Multidecadal Oscillation (“AMO”) and also that of a similar index calculated for the North Pacific I call the PMO (the detrended mean 20N-65N, 100W-100E SST anomaly). Minimums, maximums, and zero crossings of the SAO all appear to coincide with certain related aspects of the relationship between the AMO and PMO as detailed in my previous post.

In general, when the SAO is negative, the hemispheric sea ice anomalies tend to move in opposite directions; that is, when the anomaly increases in one hemisphere, it tends to decrease in the other (either up or down). When the SAO is positive, the opposite is true.


Figure 1 – the Northern and Southern Hemisphere sea ice anomalies and the SAO. Any point in the SAO represents the sign and magnitude of the relationship between the NH and SH sea ice anomalies for the preceding 365 days.

The SAO and SLP

Looking into the relationship between the SAO and SLP, I used the NP Index, and a similar index calculated for the North Atlantic (the “NASLP”) – the mean N. Atlantic SLP (0-70N, 280-360E).

The NCAR website ( describes the NP index as “the area-weighted sea level pressure over the region 30°N-65°N, 160°E-140°W. The NP index is defined to measure interannual to decadal variations in the atmospheric circulation. The dominant atmosphere-ocean relation in the north Pacific is one where atmospheric changes lead changes in sea surface temperatures by one to two months. However, strong ties exist with events in the tropical Pacific, with changes in tropical Pacific SSTs leading SSTs in the north Pacific by three months.”

The first and most obvious thing you see in the monthly NP Index is the similarity to the NH sea ice anomaly over the past seven years or so. With a four month lag, the two line up very closely (2007.00 – 2013.58: r^2 = 0.39). A similar relationship appears to exist between the NH anomaly and the NASLP (r^2 = 0.19), Figure 2.

With both the NP Index (N. Pacific SLP) and the NASLP (N. Atlantic SLP), it appears that after the SAO turns positive at around 1996, the correlation between the NH sea ice anomaly and the SLP gradually changes from negative to positive. In addition, Figure 3 appears to show that when the SAO is negative, the NH anomaly correlates fairly well with the inverse N. Pacific SST (Sea Surface Temperature), and when the SAO changes to the positive phase, SLP becomes the main driver.


Figure 2 – the NH sea ice anomaly compared to the NP Index lagged 4 months, and the mean N. Atlantic SLP lagged 3 months. Both the North Pacific and North Atlantic SLP appear to be negatively correlated to the NH Anomaly in the negative phase of the SAO and gradually change to a positive correlation after entering the positive phase of the SAO.


Figure 3 – the NH sea ice anomaly compared to the inverse mean N. Pacific SST (top) and the NP Index (bottom). The NH sea ice anomaly appears to be directly correlated to the NP SST when the SAO is negative and the NP SLP when the SAO is positive.

Another observation that is less obvious, and perhaps less expected, is that when smoothed with a 13-month centered simple moving average filter (no lag), the NP Index correlates fairly well with the SH sea ice anomaly. As can be seen in Figure 5, the sign of the SH anomaly correlation appears to be inversely related to the phase of the SAO. When the SAO is negative, the correlation coefficient is positive and vice versa. Correlation coefficients: 1980-1996 (SAO negative) = 0.39, 1996-2012 (SAO positive) = -0.09, and 2012-2014 (SAO negative) = 0.34).

Unlike with the NH anomaly which appears to have a SAO-correlated relationship with both the NP Index and the NASLP, the SH sea ice anomaly appears to have such a relationship with only the NP Index. In the case of the SH anomaly and the NASLP, the two appear to be uncorrelated or perhaps slightly inversely correlated across the entire record. Correlation coefficients: 1980-1996 = -0.08, 1996-2012 = -0.10, and 2012-2014 = -0.67. This imbalance may contribute to the oscillation of the SAO.


Figure 4 – the SH sea ice anomaly compared to the NP Index (top) and the NASLP (bottom) both smoothed with a 13-month centered SMA filter. The SH anomaly correlation to the NP Index appears to follow the phases of the SAO with the SH anomaly being directly correlated to the NP Index when the SAO is negative. A similar relationship is not apparent with the SH anomaly and the NASLP. The insert shows the SH anomaly compared to the inverted smoothed NP Index to illustrate the direct correlation to the inverted NP Index when the SAO is positive.

In addition to relationships between SLP and the hemispheric sea ice anomalies that are correlated to the trend-modeled phases of the SAO, there are also clear visual similarities between the SAO itself and the 13-month smoothed NP Index (Figure 5).


Figure 5 – The NP Index compared to the SAO over the duration of the SAO record.

The SAO, SLP, and the ENSO

In my original SAO post, I noted a possible relationship between the SAO and the ENSO – as seen in the historical N. Atlantic and N. Pacific SST patterns and also in the similarities between the SAO and ENSO indexes such as the NINO3.4 when the index is inverted at the SAO inflection points as illustrated in the bottom chart in Figure 6. As such, it is probably not surprising that the NP Index also appears to have a relationship to the (inverted) NINO3.4 (Figure 6 top).


Figure 6 – the NP Index compared to the inverse NINO3.4 (top), and the SAO compared to the NINO3.4 inverted at the SAO inflection points (bottom).


It appears likely that SLP is an important driver in the sea ice anomaly changes. If so, the logical next question is what is driving the changes in SLP? If GHGs are responsible for the decreasing NH sea ice anomaly as some would have us believe, Figures 2 and 3 above would seem to suggest that GHGs must be significantly influencing SLP. In my opinion, this is not the case. Rather, I think the evidence supports the natural cycles theory. For one thing, the pattern of the SAO – which also appears to be a pattern of the ENSO and SLP – can be seen in the SSTs of both the N. Pacific and N. Atlantic. In the case of the N. Atlantic, the pattern goes back at least 100 years (see my previous post for specifics).

Data, sources, and the methodology used can be found in an Excel file here:

0 0 vote
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
February 20, 2014 12:44 pm

Excellent post! I had noticed last year what appeared to be a correlation with lunar phases and sea ice growth and diminishment. It struck me though, that the lunar phases could only be one part of the mechanism as it is not consistently seen that the sea ice reacts the same with every lunar cycle. Yet it certainly lined up with a portion of it. Notably the last 3 Arctic sea ice trend changes/shifts all occurred at the beginning of a new moon, as seen by watching the NSIDC Arctic trend. It can be seen that as the Arctic sea ice moves towards it,s maximum that each new moon correlates with a pause in the continued growth of the sea ice. I had noted an opposite reaction in Antarctica sea ice growth patterns in May of last year. What now puzzles me about what I think that I am observing is that I had used NSIDC trend data to see the shifts, but when I took a look using the Cryosphere Today page then I could not make the same connection. Is that caused by the NSIDC smoothing their data points for their graph? The Cryosphere shows all the daily steps, so that it shows every shift in the movement of the sea ice.

February 20, 2014 12:46 pm

The last portion of the first comment is referring to watching the changes in Antarctica, not the Arctic.

February 20, 2014 1:53 pm

Does anyone else see a distinct sine wave in these graphs? Aren’t the blue and red halves virtual mirror images, with the red now trending back down to the baseline? Or is it just me?

February 20, 2014 1:54 pm

The NOAA link I posted, not this article’s sea ice graphs.

February 20, 2014 2:15 pm

Looking at the WUWT sea ice page one can see that the maximum Arctic sea ice extent for 2014 is approaching soon and that it will probably be 1 million square kilometers less than 2013s peak amount. The anomaly is found in the space around Svarlsbard. Also the DMI Arctic temperature graph looks much warmer this spring than in recent springs. Especially compared to 2013, So I would not predict another record late Tannana ice break-up this year. I do not know what this means, maybe just a result of the shift of the polar vortex.

February 20, 2014 4:20 pm

Zek203 says:
February 20, 2014 at 2:15 pm
“Also the DMI Arctic temperature graph looks much warmer this spring than in recent springs. Especially compared to 2013,”
Yes, all of the arctic air ended up in Minnesota. This is the result of the breakup of the wintertime polar vortex which just happened to coincide with the weak solar cycle, and also the recent solar magnetic polarity switch from antiparallel to parallel to the earth’s field.

February 20, 2014 4:55 pm

The SAO is the moving, trailing-356-day regression ???
Did you mean 365 day? or average of trailing current month and preceding 11.

February 21, 2014 2:37 am

So it is just possible we have tidal link to all of this? A pattern based on a 3 or 4 times multiplier of the Saros cycle at 18.6 years would fit and have the required physical mechanisms to create the linkage?

February 21, 2014 2:39 am

P.S. It would be nice to have the horizontal axis labelled with the Year as it gets quite difficult to compare this to other work without.

February 21, 2014 5:16 am

RichardLH, they are labeled, it’s just hard to see in the smaller images. If you click on them, it’s easier to see in the larger version.

February 21, 2014 6:11 am

Craig says:
February 21, 2014 at 5:16 am
“RichardLH, they are labeled, it’s just hard to see in the smaller images. If you click on them, it’s easier to see in the larger version.”
Thanks. I had missed that because I always expect the Axis to be labelled down low – not in the middle. Sorry.

February 21, 2014 6:14 am

Craig: So why the periodicity (quasi sine wave) in the plots? Are you proposing any external mechanism?

Arno Arrak
February 21, 2014 8:20 am

I really get annoyed when speculation about causes of northern and southern sea ice different behavior repeatedly crop up. The explanation is simple: the difference is caused by different physical processes involved. Arctic sea ice is not controlled by any imaginary greenhouse warming or by any mysterious long term cycle. Its cause is warm Gulf Stream water carried into the Arctic Ocean by North Atlantic currents. I have had a paper out on this since 2011 but these climate scientists, both real and warmist, simply don’t do their homework. To make matters worse, Anthony did not like my English grammar and turned down my offer to post it in some form. To make it very simple, Arctic warming that controls the ice melt started suddenly at the turn of the twentieth century. Prior to that there was nothing but slow, linear cooling for two thousand years. Checking the Keeling curve extension for 1900 shows that there was no sudden increase of atmospheric carbon dioxide at that time. This rules out greenhouse warming as a cause of this sudden warming. Radiation laws of physics simply do not allow it. After considering other possible causes I decided that the only thing that makes sense is a rearrangement of the North Atlantic current system at the turn of the century that started to carry warm Gulf Stream water into the Arctic at the turn of the century. This is the only way you can get warm water into the Arctic on a broad front. The warming paused for thirty years in mid-century, then resumed, and is still going on. Direct measurement of water temperature reaching the Arctic in 2010 showed that it was higher than at any time during the last 2000 years. The pause in warming was not just a pause but an actual cooling at the rate of 0.3 degrees per decade. Most likely it involved a temporary return of the previous flow pattern of currents. Since what has happened in nature can happen again it is not impossible for the current Arctic warming some day to reverse itself. But that is speculation. You can download my paper from Judith Curry’s web site.

Ulric Lyons
February 21, 2014 5:17 pm

With the main source of warmer sea water to the Arctic being through the Fram Strait, it would make sense to compare the northern sea ice anomaly with the North Atlantic Oscillation. The bottom line is that positive NAO conditions are dominant in global warming phases, while during negative NAO conditions there is increased poleward warm water transport and sea ice extent reduction. Global circulation models predict increasingly lower Arctic pressure with increased warming, so they predict the opposite of Arctic amplification.

February 21, 2014 7:16 pm

RichardLH, every curve in this post is the SAO fit. It is a sine wave with a period of about 32 years and almost exactly half that of the AMO and PMO (the exact specifics are in a table near the end of my previous post). The min, max, and zero crossings all appear to be related to the relationship between the AMO and PMO. I don’t have an opinion at this point on whether the AMO and PMO are the mechanism or if another mechanism is driving both..

February 22, 2014 3:18 am

Craig: I now the AMO, PDO and a load of other things have the same ~60 year periodicity to them. I have been pointing that out for some time now. I have, as yet, drawn no conclusions as to which is the driver or, indeed, if they are all responding to some other external factor.

Brian H
February 23, 2014 1:59 pm

In Fig 5, the “visual match” is close in ’94 & ’95, but no better than a drunkard’s walk before and after. “Pray, what is the reason for that?” (Lewis Carroll)

%d bloggers like this: