Guest post by David Archibald
Bob Tisdale’s post on ENSO on 19th November prompted me to see what I could find in the Southern Oscillation Index (SOI) data. The SOI is calculated from the monthly or seasonal fluctuations in the air pressure difference between Tahiti and Darwin. Sustained negative values of the SOI often indicate El Nino episodes. These negative values are usually accompanied by sustained warming of the central and eastern tropical Pacific Ocean, a decrease in the Pacific Trade Winds, and a reduction in rainfall over eastern and northern Australia. Following is a graph of the SOI on a monthly basis from 1876 to 2010. The major El Ninos are discernable, otherwise it looks like a lot of noise.
The graph following shows the cumulative monthly SOI from 1876 to 2010.
The SOI does tell a story. It was non-trending for the last of the Little Ice Age and then from 1920 went into a long La Nina-dominated trend that ended with the Great Pacific Climate Shift of 1976. The planet warmed into the 1930s at the beginning of this trend, but then cooled, as it should have down in a La Nina-dominated trend, from the 1940s to the 1970s.
The subsequent El Nino-dominated trend from 1976 to 1995 was almost three times as fast as the rise. The Climategate emails show that Phil Jones was aware that global warming ended in 1995. The end of the El Nino-dominated trend in 1995 might be the physical cause of that cessation of warming. The SOI has been non-trending since.
This might have been a very neat story if the world had cooled instead of warmed into the 1930s. The 20 years of El Nino-dominant trend from 1976 to 1995 produced the late 20th century warming that got so many people hot and bothered. The story told by the SOI also reinforces how important the Great Pacific Climate Shift of 1976 was. The climate system turned on a dime for some as yet unknown reason.
Those who say, “So what?” to this are missing the point.
Of COURSE this is just observing an artifact. ALL climate measurements are artifacts. There is not a direct measurement of a cause in any of it. We measure temps and pressures and precip and SSTs and Arctic ice and tree rings and ice cores and coral reef growth, and all of them – without exception – are artifacts. They are all symptoms. Saying “So what?” could be applied to the temperatures just as well.
The causes are the various energy outputs that reach us from the Sun, whatever heat comes out of the Earth, the tilt of the Earth, the laws of chemistry and physics, the rotation of the Earth, the configuration of the continents and mountain ranges, and apparently the Milankovitch factors. Numbers 4, 5 and 6 in that list are constants; the rest are variable. Warmers would say human activity should be included, but thus far that has really only been the pea under the Princess’ mattress.
Just as “everything is derivative” in out culture, everything we measure in climate is an artifact.
I applaud Archibald for finding a new perspective on this. If ENSO is important, it behooves us to find the significant measurements to make. Since we don’t understand climate yet, it is obvious we haven’t found the right things to measure and interpret yet. Our collective light bulb has not yet illuminated us yet.
So why not rearrange the data a bit, look at it in other ways? I think history would show that we don’t know which ways of observing are important until we get smacked in the face by them. Most of the ways we are looking at it now are not working, are they? Not yet. Maybe fairly soon, but not yet.
Synchronized chaos, folks. There are many factors involved and they don’t all work together at the same. When they do however, that is when climate shifts take place. See Tsonis.
Are we now witnessing a synchronized event as described by Tsonis?
Bob Tisdale says:
December 9, 2010 at 8:16 am
How? There’s a time lag (a couple of years at minimum) between the release of heat from the Pacific Warm Pool until the time that any residual warm water could be carried down to the ACC, then caught up into the Benguela Current in the South Atlantic, and carried north into the North Atlantic. If memory serves me well, an ENSO signal takes 8 years to travel around the Antarctic in the ACC.
Hi Bob, good question.
The only answer I can come up with at this stage, and it’s more than likely that I’m wrong, is rather than thinking of it as a “body of warm water” meandering for thousands of miles, I think of it as a channell of flowing water. If this flow slows then thousands of miles at the other end (the North Atlantic end) it also slows thereby allowing the sun to warm the surface. But for this to be true, there has to be an upwelling zone in the North Atlantic. I believe Thermohaline Circulation models show that there is, off the coast of Nth West Canada Sth of Alaska.
Is just a thought.
Feet2theFire:
December 9, 2010 at 9:11 am
Well put sir, bravo.
Ian L. McQueen says:
December 9, 2010 at 7:39 am
It is my understanding that it is the Trade Winds that blow the warm surface water westward during a La Nina, permitting the cold subsurface water to rise.
Hi Ian, here is the way John L Daly explains it…my bolding
The El-Niño Southern Oscillation is the result of a cyclic warming and cooling of the surface ocean of the central and eastern Pacific. This region of the ocean is normally colder than it’s equatorial location would suggest, mainly due to the influence of northeasterly trade winds, a cold ocean current flowing up the coast of Chile, and to the upwelling of cold deep water off the coast of Peru.
Feet2theFire says:
December 9, 2010 at 8:29 am
I hope this is useful to you. At the John L Daly site there is hours and hours of reading about the SOI or ENSO. Most of it is by a man called Theodore Landscheidt.
His equations and charts are way over my head but the gist of it that I could gather makes sense. He claims a high degree of accuracy in predicting ENSO 2-3 years out.
make a start here if you wish http://www.john-daly.com/sun-enso/revisit.htm
Bob Tilsdate says: http://i54.tinypic.com/33di5qr.jpg
You say you filter the data with a 121 month running mean. You recognise this is a filter, do you not worry the frequency response of this filter may give misleading results or cause you to miss something?
What mean did you use here? Just a straight box car running mean?
I have done some work of filter frequency responses, the number and magnitude of side lobes are frightening.
Can you link to the data you plotted here? I’d be interested in trying some other filters.
regards.
Re: DR
For uncritical consumers, your line of influence may be hazardous. Take a minute to think about how nonstationary spatiotemporal modes alias. When the assumption of randomness fails, Simpson’s Paradox bites HARD with SHARP teeth. Suggested: Consider the possibility of strange nonchaotic attractors and switch to multiscale phase-aware data exploration (ethically resisting misguided conventional mainstream pressure to underpin meaningless statistical inference with absolutely untenable assumptions). Regards.
Bob Tisdale,
With anomalies, we can see more variation around the average annual cycling in the KOE, but temperature-precipitation relations are driven by absolutes. Seeing this graph [ http://i56.tinypic.com/29e0pvp.jpg ] in absolutes might add even further insight. Certainly we do need to be aware of where the big amplitudes are when looking at aggregated summaries (e.g. global averages).
Your notes on KOE dovetail with what I’ve been saying about solar cycling & earth rotation rates. For example, see recent notes here [ http://wattsupwiththat.com/2010/12/07/massive-solar-filament-eruption-captured-by-sdo/#comment-547001 ] in response to a comment from vukcevic on the ~1925 X pattern, where seasonal heating/cooling patterns crossed in the NH.
You had an article awhile back where you succinctly illustrated how the North Pacific is warm during the cold phase of PDO (contrary to widespread misunderstandings about PDO, which are seriously interfering with progress in these discussions). I think readers might benefit from a link to that article. (I’ve lost the link.)
–
I agree with Bob Tisdale & Bill Illis that atmospheric teleconnections are the main ingredient. Nonsensical notions about long term ocean lags ignore the fact that oceans have no problem cooling down given a few months without much insolation (which varies seasonally & with interannual cloud cover). People are quick to point out the disparate heat capacities of water & air. That matters on diurnal timescales. Ask any sea-kayaker. The ocean might not be colder at night than during the day, but it most certainly is cooler in winter than in summer. (Decades not necessarily required, just a change in insolation persisting for at least many days.)
–
David Archibald, you might next explore the effect of recentering your anomalies on your integral of SOI. The base period matters. Most agencies are using base periods centered on the recent warm period. The early 20th century “issue” with your graph goes away if you recenter to something less biased by the late 20th century. I also suggest using -SOI [i.e. negative SOI] instead of SOI so that up corresponds with warm persistence and down corresponds with cold persistence. The next step (e.g. for anyone trying to understand what vukcevic is hinting at) is to look at the integrals by month to explore variations in seasonal persistence. I recommend doing so via a color-contour plot (available in Excel).
Best Regards.
Bob Tisdale, in reading your recent article …
Tisdale, Bob (2010). The ENSO-related variations in Kuroshio-Oyashio Extension (KOE) SST anomalies and their impact on Northern Hemisphere temperatures.
http://bobtisdale.blogspot.com/2010/12/enso-related-variations-in-kuroshio.html
… I noticed some musing on dynamics of ~1988 when AO persistence switched decidedly to positive & AO/NAO coupled for about a decade with global patterns (like GLAAM & LOD’). Interannual NPI coupling with interannual AO/NAO was briefly reversed at the beginning of this interval (rare during the 20th century). I detected this using a new algorithm that represents correlations of adjacent derivatives in the complex plane. Subsequently I found this supporting evidence:
Schwing, F.B.; Jiang, J.; & Mendelssohn, R. (2003). Coherency of multi-scale abrupt changes between the NAO, NPI, and PDO. Geophysical Research Letters 30(7), 1406. doi:10.1029/2002GL016535.
http://www.spaceweather.ac.cn/publication/jgrs/2003/Geophysical_Research_Letters/2002GL016535.pdf
Note their speculation on a temporary change in the number of spatial modes. (This could certainly help with explaining why EOF1’s explain so little of the total variance.)
Data exploration using a high-resolution complex wavelet suggests that the change was driven by an external factor: the sun.
Looking at this 1876-2010 SOI graph, I have a simple and perhaps stupid question.
The AGW mantra includes the claim of ever more ‘disrupted’ and extreme climate swings, yet that alleged trend does not appear to be evident. If anything, it looks like the opposite trend.
So what’s up with that, if anything?
Paul Vaughan & Bob Tisdale
Couple of months ago I took values of the geomagnetic field along Kuroshio-Oyashio currents semi-loop and got an interesting graph:
http://www.vukcevic.talktalk.net/PDOz.htm
(Paul might recall the graph from past discussion).
P. Solar says: “You say you filter the data with a 121 month running mean. You recognise this is a filter, do you not worry the frequency response of this filter may give misleading results or cause you to miss something?”
I’m using the filter in this instance to show that there are long-term changes in frequency and amplitude in the NINO3.4 data—that there’s more to the data than just noise. And I use it to show that the data switches from positive to negative at times that coincide with warming versus cooling epochs. That is, El Nino events dominated from 1910 to the mid 1940s (global temperatures rose), La Nina events dominated from the mid-40s to the mid 1970s (global temperatures fell), and El Nino events dominated after the mid 1970s (global temperatures rose again). I’ve used other spans for the running mean with similar results. And when I use the 121-month running mean in a post, I usually preface it with something along the lines of, it’s the same filter used by the NOAA ESRL for their AMO data.
You asked, “What mean did you use here? Just a straight box car running mean?”
The value at month 61 is the average of the NINO3.4 SST anomalies for months 1 through 121. The value at month 62 is the average of months 2 through 122. Etc.
You asked, “Can you link to the data you plotted here?”
Starting here:
http://climexp.knmi.nl/selectindex.cgi?someone@somewhere
Click on NINO3.4 in the second row, which is the HADISST1-based NINO3.4 SST anomaly data. On the next page, scroll down to “Manipulate this time series” and enter 1900 & 2011 in the “select years” fields. Click on “select”. On the next page, scroll down to the third graph, which is anomalies, and click on “raw data”. There’s your data. (From the looks of the graph we’re discussing, I believe that I also used 1950 to 1979 as the base years, which you would select after you limit the years. But as you’re aware, the base years will just shift the curve up and down.)
Paul Vaughan says: “Seeing this graph [ http://i56.tinypic.com/29e0pvp.jpg ] in absolutes might add even further insight.”
The GISS data is not available in absolute temperatures, only anomalies. Also, the differences between TLT and surface temperatures are great. For example the average TLT for the No Hemisphere north of 20N from 1995 to 2009 is approximately -6 deg C, while the KOE SST is about 16.5 deg C. Plotting them togther on a graph would serve little purpose.
And thanks for the link to Schwing et al.
vukcevic says: (December 9, 2010 at 1:37 pm)
Thanks.
Ian L. McQueen says: “Caveat: Though solar insolation or cloud cover do not cause ENSO events, they may well modulate the strength of these events. High insolation during El Nino strengthens it, low insolation weakens it. Vice versa for la Nina.”
Do you have a link to a paper or data that supports this? Cloud cover follows the warm water east from the Pacific Warm Pool during an El Nino event, so DSR over the central and eastern equatorial Pacific decreases during an El Nino. But DSR also drastically increases over the Pacific Warm Pool during an El Nino.
I’ve been reading this thread with great interest, from the original plot by David Archibald through to all the comments, explanations and queries about cumulative plots of time series. Over the last 16 years I have made literally thousands of cumulative sum plots on virtually every climate related time series that I’ve been able to find, or which have been sent to me. The “technology” is very simple indeed, as has been shown in various postings above. Interpreting the plots is something that one acquires with experience, and I don’t want to embark on a “lecture” about it here, but I would be very willing to go into some of the “subtleties” if pressed!
I have cusum plots of /loads/ of assorted climate indices, and these readily confirm the main points made by David. The 1976 shift in the PDO is a nice example, and as I’ll claim below, it can be “quantified” in a temporal sense to a particular month or possibly two. Climate step changes in Alaska, derived from the several readily available station data sets turn out to occur a small number of months after the PDO shift.
The great plus point for cusum analysis is that it produces a pattern, or trace, or fingerprint, that virtually instantly describes the large-scale behaviour of a time series. Attempting to do this from standard data plots, or plots “enhanced” by smoothing processes, is generally rather difficult. For instance, it is almost impossible to identify a step change from a conventional plot, much less attach a good estimate of the date of the step, and still less its magnitude. All these are quite trivial when examining a cusum plot.
What I have found over the years is that climate data are suffused with step changes, which seem to punctuate long stable periods. By long I mean perhaps 20 months up to two centuries. Of course the departures from a straight line cusum (a period of stability) increase with the length of the series, and with the time resolution one chooses to use. There is also the problem of dealing with subjectivity in relationship to choice of periods that one might believe to be “stable”, but once you’ve looked at several cusum plots you’ll realise that the major features of the plots are usually so obvious that you’ll wonder why other people haven’t noticed them! It is possible to use standard (though often fairly computation intensive) methods to identify step changes from cusum plots. I have tried this but with my limited facilities (and faculties) it is a real chore. For academic purposes I guess one would have to go about it that way, but from the practical aspect I’d expect that it is not worth the hassle. The differences would be trivial, I would bet.
If anyone is interested in the sorts of things that cusums reveal I would be happy to correspond via email. I could then send plots illustrating loads of fascinating things that climatologists seem to have overlooked – such as the NW Atlantic step change in the autumn (fall?!) of 1922, or the extraordinary behaviour of Central England temperatures.
Robin
Thanks for the reply Bob.
I would like to respond to a number of points made in this most interesting thread.
I am willing to expand on any of the points that I am about to make.
But will not do so now, in the interest of brevity.
I have found that rainfall is a better indicator of climate than temperature.
At least, when both are expressed as annual totals/ averages (rain/temperature).
There appears to be no long term tred in the rainfall at a number of individual Australian locations.
Rainfall like othere aspects of climate is chaotic, in the mathematical sense.
Hurst examined records of 841 years of river Nile flow and from that developed R/S analysis, which he measured on a scale of zero to one.
A Hurst number of 1, indicates a series that is fully auto-correlated, which means that if the series starts to wander say upwards, then it will continue in that direction until an external force pushes it back down again. The share market is an example of series with high Hurst numbers.
A series with a Hurst value of 0.5 is truely random – Einstein’s Brownian motion is an example.
Series with values approaching zero are trendless and have strong tendence to revert to the mean.
I have found the rainfall at locations such as Sydney & Mrlbourne, which have long rainfall records (150 years plus), have Hurst values very nearly zero.
Rainfall in Australia fluctiates wildly on a multi decile basis, but there is no long term trend.
Now having said all that quite briefly, I turn to my main surmise.
It is possible that all the trends that you are all calculating, are merely long term fluctiations.
You cannot find the principal driver, because there may be none.
The spinning earth, its yearly passage around the sun, fluctiating magnetic force line between earth and sun (11/22 year sun spot cycles) and so on, all kick things around and around – AMO, PDO ENSO and so forth respond – all trying valiantly to catch up.
Perhaps what goes around just comes around again and again and again; chaotically and on multi different time scales.
Until there is some basic change in the system and another ice age or hot house era emerges.
That also could just be due to the random coincidence of independent directionless forces coming togther haphazardly.
We humans keep looking for pattern, causes, directions, purposes, when none may in reality exist.
Bob Tisdale replying to P. Solar: “[…] the base years will just shift the curve up and down.”
…by a different amount for each month – and each of these 12 amounts can vary substantially as one adjusts the base period. In some types of analysis, this substantially changes what turns up. I advise making sensitivity analyses a habit.
–
Robin Edwards, Looking at integrals & derivatives is among the first things I do to learn the nature of a new time series. I suggest organizing some of your top highlights into a webpage or perhaps a guest post. The note you mention about the North Atlantic dovetails with findings of Harald Yndestad. He suggests there was a fundamental change in the N. Atlantic sometime around 1923 (based on signal analysis, including wavelet analysis of a variety of time series).
–
vukcevic, I saw your note on the other thread. I suspect the 45 year pattern relates to alignments of the QBO with the year. It doesn’t take more than a few minutes to write a little program in Excel that will figure out stuff like that for any period you want to enter. The program will also illustrate why periods jump around – for example as one sees in the LOD time series. Everyone needs to realize that due to north-south, polar-equatorial, & continental-maritime terrestrial contrasts (for a few simple examples), nonstationary signals won’t necessarily sit conveniently piled in global FFT power spikes.
~ Bob
I’ve grabbed the data you indicated but it does not really look anything like your plot. Could you cast an eye over this snippet and tell me if that’s the data you intended to point me too.
2009.2500 -0.190559
2009.3334 0.304281
2009.4166 0.574663
2009.5000 0.800828
2009.5834 0.694503
2009.6666 0.778035
2009.7500 0.997765
2009.8334 1.57316
2009.9166 1.80899
2010.0000 1.40437
2010.0834 1.15557
2010.1666 1.12739
2010.2500 0.615990
2010.3334 -0.235387E-01
2010.4166 -0.538576
2010.5000 -0.785873
2010.5834 -1.25070
2010.6666 -1.46854
thanks.
Bob Tisdale, there’s also a decided upturn in the integral of AAO/SAM ~1992 that I suspect (based on multivariate analyses) is related to the upturn in AO/NAO a few years earlier. A question I’ve been meaning to ask you: Where was the ~1940 upturn in SST most distinct? (It seems to me I remember seeing a graph of some section of the Indian Ocean where ~1940 spiked uniquely in a long record lacking other prominent features…)
AusieDan, people see chaos in places where it does not exist (perhaps proving your point). Have a look at this for some stimulating ideas:
White, W.B.; & Liu, Z. (2008). Non-linear alignment of El Nino to the 11-yr solar cycle. Geophysical Research Letters 35, L19607. doi:10.1029/2008GL034831.
https://www.cfa.harvard.edu/~wsoon/RoddamNarasimha-SolarENSOISM-09-d/WhiteLiu08-SolarHarmonics+ENSO.pdf
Russian scientists are arguing that we should be looking for a strange nonchaotic attractor (which would still be very hard to figure out).
Ian L. McQueen says: at 7:39 am
“This is because of the cold ocean currents flowing up the coast of Chile, upwelling near Peru.”
“The upwelling is the result of the thermohaline circulation, . . . ”
E.M.Smith says: at 7:10 am
YES and so on . . .
I’m late to this post and want to express an interest in the – more than usual – ocean circulation comments.
I appreciate seeing such comments (and links) because lack of this sort of knowledge often leads to very elementary mistakes. One I’ve seen a couple of times is an expression of surprise when someone finds out that sea level isn’t the same across the world or even for the same ocean.
More concern for the ocean — currents, the size, timing, and shape of the various basins – seems to be required. For example, does not the surface water along the west coast of South America move away from the coast following the principle outlined in Ekman transport?
http://en.wikipedia.org/wiki/Ekman_transport
So, the pressure system west of South America contributes to the development of the South Eastern Trades. From the link just above:
“Ekman transport is a factor in coastal upwelling regimes which provide the nutrient supply for some of the largest fishing markets on the planet. Wind in these regimes blows parallel to the coast (such as along the coast of Peru, where the wind blows North). From Ekman transport, surface water has a net movement of 90 degrees to the left in such a location. Because the surface water flows away from the coast, the water must be replaced with water from below.”
Now I’m off to read a few of those links others have thoughtfully provided. Thanks.
AusieDan:
Re: patterns vs. no patterns in rainfall data, I did a study some time ago of the rainfall over the Melbourne area. Taking the Bureau of Meteorology (BOM) data for the grid box centered at 38.0 deg S: 145.0 deg E, I did a seven period, centered, non weighted, moving average of the annual totals. When plotted, this showed a distinct periodic form averaging roughly an eighteen year period. There were some quite large variations in amplitude, but no noticeable trend. It sounds as if the lack of trend is consistent with your broader observations, however, a closer look at the cyclic pattern revealed an intriguing possibility.
I superimposed a nominally 18.0 year period sine curve, and by adjusting parameters, closely matched the magnitude to a mean of 900 mm/yr, +/-55 mm. Scanning periods showed that it was not 18.1 yrs.. Rhodes Fairbridge has identified an 18.03 year cycle in tides affecting the Canadian-American east coast from Newfoundland down to the Gulf of Mexico, which he attributed to the Saros cycle of solar-lunar nodes alignment, associated with the occurrence of lunar eclipses. Could it be that the basic cycle also has a tidal influence here? I found that setting the period to 18.03 years, with the first high point at 1902.5, produced a very close temporal match.
Looking at recent events, an obvious departure from the idealized sine wave occurred around the time of the 1998 El Nino, falling to about 110 mm/ yr below an idealized low in 2002.5, but then instead of then rising, stayed low until around 2008. This has been taken as evidence of a changing climate, but since 2008 rainfall has increased considerably, and the effect of the current La Nina looks like restoring the pattern back into line with the long term 18.03 year period cycle. There is an ideal high at 2011.6, and if the current weather pattern holds it seems very likely that will be achieved. Was the recent dry period really not a sign of long term climate change, but rather an unusually low excursion in a natural and normal cyclic pattern?