From the University of Hawaii ‑ SOEST:
Climate researchers discover new rhythm for El Niño
El Niño wreaks havoc across the globe, shifting weather patterns that spawn droughts in some regions and floods in others. The impacts of this tropical Pacific climate phenomenon are well known and documented.
A mystery, however, has remained despite decades of research: Why does El Niño always peak around Christmas and end quickly by February to April?
Now there is an answer: An unusual wind pattern that straddles the equatorial Pacific during strong El Niño events and swings back and forth with a period of 15 months explains El Niño’s close ties to the annual cycle.
This finding is reported in the May 26, 2013, online issue of Nature Geoscience by scientists from the University of Hawai’i at Manoa Meteorology Department and International Pacific Research Center.
“This atmospheric pattern peaks in February and triggers some of the well-known El Niño impacts, such as droughts in the Philippines and across Micronesia and heavy rainfall over French Polynesia,” says lead author Malte Stuecker.
When anomalous trade winds shift south they can terminate an El Niño by generating eastward propagating equatorial Kelvin waves that eventually resume upwelling of cold water in the eastern equatorial Pacific. This wind shift is part of the larger, unusual atmospheric pattern accompanying El Niño events, in which a high-pressure system hovers over the Philippines and the major rain band of the South Pacific rapidly shifts equatorward.
With the help of numerical atmospheric models, the scientists discovered that this unusual pattern originates from an interaction between El Niño and the seasonal evolution of temperatures in the western tropical Pacific warm pool.
“Not all El Niño events are accompanied by this unusual wind pattern” notes Malte Stuecker, “but once El Niño conditions reach a certain threshold amplitude during the right time of the year, it is like a jack-in-the-box whose lid pops open.”
A study of the evolution of the anomalous wind pattern in the model reveals a rhythm of about 15 months accompanying strong El Niño events, which is considerably faster than the three- to five-year timetable for El Niño events, but slower than the annual cycle.
“This type of variability is known in physics as a combination tone,” says Fei-Fei Jin, professor of Meteorology and co-author of the study. Combination tones have been known for more than three centuries. They where discovered by violin builder Tartini, who realized that our ear can create a third tone, even though only two tones are played on a violin.
“The unusual wind pattern straddling the equator during an El Niño is such a combination tone between El Niño events and the seasonal march of the sun across the equator” says co-author Axel Timmermann, climate scientist at the International Pacific Research Center and professor at the Department of Oceanography, University of Hawai’i. He adds, “It turns out that many climate models have difficulties creating the correct combination tone, which is likely to impact their ability to simulate and predict El Niño events and their global impacts.”
The scientists are convinced that a better representation of the 15-month tropical Pacific wind pattern in climate models will improve El Niño forecasts. Moreover, they say the latest climate model projections suggest that El Niño events will be accompanied more often by this combination tone wind pattern, which will also change the characteristics of future El Niño rainfall patterns.
Citation: Stuecker, M. F., A. Timmermann, F.-F. Jin, S. McGregor, and H.-L. Ren (2013), A combination mode of the annual cycle and the El Niño/Southern Oscillation, Nature Geoscience, May 26 online publication at http://dx.doi.org/10.1038/ngeo1826.
h/t to Dr. Leif Svalgaard
I am having trouble posting – It will not post my comments but when I try posting it again it says that it is already posted. Nothing appears. I have tried clearing my cookies and restarting my programs but to no avail.
Greg: Goodman
This is what I have trying to tell Bob and others for years, however, most [not you Greg] have politely either ignored my comments or dismissed my research papers and work. I have tried to pointing people to the work of others [e.g. Claire Perigaud] that support a Lunar tidal explanation for the ENSO phenomenon but this has failed to produce any interest.
Thank you Greg for your support of the Lunar Tidal explanation!
If you are still interested, please read:
Long-Term Lunar Atmospheric Tides in the Southern Hemisphere
Ian R. G. Wilson and Nikolay S. Sidorenkov
The Open Atmospheric Science Journal, 2013, 7, 29-54
http://www.benthamscience.com/open/toascj/articles/V007/TOASCJ130415001.pdf
Greg Goodman,
PDS of trade winds: from chirp analysis of autocorrelation fn:
http://climategrog.wordpress.com/?attachment_id=281
Note that the peak at 7.462 year = (22.386 / 3) years
“Note that the peak at 7.462 year = (22.386 / 3) years”
Ah, I have seen this one come up in a number of spectra , I have never been able to understand where it came from. That is worth considering.
That is interesting. I’m probably too used to looking for even harmonics. Dependant on the magnitudes, 1st+2nd+3rd, could reflect a ramp or sawtooth, probably an exponential decay too. Odd only harmonics can produce an even triangular form.
A word of caution: if you take the full set of possible combinations, just restricting it two bodies you have a pretty full set of numbers. Then if you add in all odd and even harmonics and subharmincs you have enough numbers to start a new arithmetic system.
UN 172.688147785753 55.7803996866069
JN 12.7825513249407 11.0658269913118
SN 35.8731500775881 24.9920895752896
JS 19.8587480110924 8.45706696610071
JU 13.8043643320595 10.3994330961582
SU 45.2791575917216 21.8324184332597
VE 1.59868506535693 0.380877791467258
EM 2.13533151680291 0.652874201610664
The danger is that there will always be one which is “quite close”. What range of number would NOT fit into that kind of scheme ?
This is where the danger of falling into meaningless numerology creeps in. Some fairly strict self discipline needs to be applied.
Having said that, I think the 4,43 looks convincing . All the 3.x and 5.x pairs seem pretty solidly either side of 4,43. and it hardly moves. The 2,45 seems pretty well rooted as well.
Greg, indeed Dickey & Keppenne (1997) revealed more in their graphs than what they wrote about in the text. Serious climate enthusiasts can learn more from that one paper (if they’re careful to read between the lines) than they can learn from the entire collection of climate blog articles that presently exist.
Let’s at least be careful to not have misunderstandings. Misunderstandings are a waste of time.
i) Above you’ve made some generalizations about wavelets that actually aren’t true. Wavelet extent is an adjustable parameter. You suggest it must be narrow — this is not true. It can be set as narrow or as wide (including temporally-global) as one wants. The full range should be explored. Essential: Take due care to distinguish between resolution, grain, extent, & span.
ii) Also, please be careful interpreting my comments at the Talkshop about the 9 year signals. I’m certain you have the wrong impression, in part because I cut corners communicating out of sheer necessity since I’m orders of magnitude short on time. This is a topic we’re just beginning. This may take many months.
iii) Finally, the lunisolar framework looks like a good candidate for a component of the resonance framework being strummed, but the sun is doing the strumming. I suspect some (maybe most) will misunderstand what I’m saying in this paragraph. A number of misunderstandings will fade over time. Certainly Ian shouldn’t interpret silence as lack of appreciation, understanding, & interest.
A final note:
From the work I did last week on the Steinhilber+ TSI, I advise watching for 1500 year modulation of annual variation attractors and 2300 year modulation of semi-annual variation attractors (see attractors illustrated in Figure 3a & 3b of Dickey & Keppenne (1997)).
I’ll just keep picking away at all of this stuff casually as rare free time permits.
When I can I’ll comment on your 2 & 12.8 year sea ice notes.
Regards.
Paul,
Assuming that I understand (at least in part) what you are saying about the annual and semi-annual variations, then I am in agreement with your statement:
“…the lunisolar framework looks like a good candidate for a component of the resonance framework being strummed, but the sun is doing the strumming…”.
I have always claimed that the Luni-Solar tides are (by and large) a modulator of much larger solar driven changes in climate.
Greg, using temporally-global summaries you’re looking at a short univariate record that’s neither stationary nor well-constrained in aggregate by universal laws. I note that you make no mention of one of the more interesting features of TWI: the ~1998 changepoint. I hope you will make the effort necessary to understand it before you put any stock in the notion that you’ve precisely characterized temporally-global waves using this sample.
It appears we’ve taken this as far as we can for now.
All the best.
Ian: I thank you & Nikolay Sidorenkov for another stimulating contribution. If I ever have the right combination of time & resources (this luxury is nowhere to be seen on the current horizon), I’ll take a look at the patterns using a related but different method. I’ll be paying attention to see if you influence Warren White’s future work. I wish you serenity as you wait in territory that’s presently beyond the vision of most.
Sorry, just don’t buy the Scientific American analysis. Watch a few Ninos develop. They begin as salients of warm water carried westward from Peru by the various reflecting waves. Last year started with several furtive salients that didn’t develop. The most promising one in September was aborted when the warm water was mysteriously replaced by cold.
Do you seriously think the warm water source for these salients is the Pacific warm pool relaxing thousands of miles away. How would this relaxation of the warm pool and flattening of the thermocline cross the Pacific leaving no surface signature?
The waves consist of either a pair of low shoulders and a high center, or a pair of high shoulders and a low center crossing the Pacific. When they hit the continent on the opposite side they reflect with opposite sign. Either way a railroad track or a single line of warm surface water should be observable crossing the ocean. It is not. The warm salients always begin off Peru and travel west. In 1997 the salient continued across the Indian Ocean with even a weak sympathetic signal in the Atlantic.
Paul Vaughan: “i) Above you’ve made some generalizations about wavelets that actually aren’t true. Wavelet extent is an adjustable parameter. You suggest it must be narrow — this is not true.
”
No, that is not what I said. I don’t now how you manage to read that into what I said. Such misunderstandings are indeed a waste of valuable time. I realise it is variable and you used a variety of wavelet lengths to create the plots at TS using KNMI (which unfortunately no longer render what you did).
You concluded that the 9 year signal I had detected by using spectral analysis of the whole record was not a persistent feature since it was not universally present in the plots where you had used a shorter wavelet. Indeed the shorter the wavelet used, the less persistent the feature was.
What I’m saying is that that is a spurious conclusion since in periods where such a cycle is counteracted by destructive interference it will not be visible to wavelet analysis based on a short window.
That is the key point and the cause of your spurious (IMO) conclusion.
“ii) Also, please be careful interpreting my comments at the Talkshop about the 9 year signals. I’m certain you have the wrong impression, in part because I cut corners communicating out of sheer necessity since I’m orders of magnitude short on time.”
It is true that you are often terse to the point of obscurity. Free time is a problem.
You comment at TS seemed clear enough however. You said you thought the 9y cycle would be very significant if it was persistent. You then did the various wavelet plots with different windows and concluded it was not persistent and thus no more than an interesting feature.
I’m saying that this is a spurious conclusion based on misinterpretation of wavelet analysis results.
In the absence of evidence to the contrary I remain convinced that it is a significant as you suggested it was until misleading yourself and the rest of those reading my article with what appears to be a spurious conclusion.
As and when you have time to reconsider what you posted, please feel free to contact me via climategrog.wordpress.com
The presence of a strong circa 9 year signal in most major basins is significant
http://climategrog.wordpress.com/2013/03/01/61/
It is also one one of the main features that gets wholesale abuse by the Hadley regridding and processing and is the main reason I prefer ICOADS for SST when doing frequency analysis.
http://climategrog.files.wordpress.com/2013/03/icoad_v_hadsst3_ddt_n_pac_chirp.png
The following plot shows a remarkable similarity between North Atlantic and _South_ Paciic
http://climategrog.files.wordpress.com/2013/03/autocorr_icoads_n_atlsp.png
The apparent circa 10 year periodicity is in fact the superposition of 9 and 11 years signals as can be seen from the drops in autocorrelations at around 40 and 80 years.
It is a combination of the corruption of the data in “reanalysis” datasets and the failure of trivial analyses to resolve these features that has lead to the rejection of both solar and lunar hypothesis as major drivers of global scale climate variation.
Greg said:
It is a combination of the corruption of the data in “reanalysis” datasets and the failure of trivial analyses to resolve these features that has lead to the rejection of both solar and lunar hypothesis as major drivers of global scale climate variation.
Reply: Greg, your comment is right on the money. The climate systems here on Earth are responding (in part) to outside forcing(s) from both the Sun and Moon. The problem we have in interpretating what is going on is that different parts of the system (atmosphere pressure, sea-surface temperatures, up welling of deep ocean waters etc) respond in different ways – some respond directly to the forcing(s), while others respond to an admixture of the forcing(s) and the annual (seasonal) cycle.
Greg, you’ve repeated your misinterpretation of what you are calling my “conclusion” about the 9 year signals. Also, the Talkshop readers have not been misled to dismiss the 9 year signal — quite the contrary. You’re making false assumptions about what I’m thinking. We’re more on the same page than you realize, but we differ in one key way: You have time to chase spatiotemporally-turbulent interference patterns in local metrics that are not well-constrained in a global sense whereas with my limited time & resources I strategically focus my attention intensely where there’s simple crystal clarity — i.e. on globally-well-constrained metrics such as EOP (Earth Orientation Parameters). If you reread my Talkshop comments, you’ll see clearly that I suggest someone with a lot of time & resources on their hands explore further the spatiotemporally-local details of the 9 year signals that are undeniable at the globally-well-constrained scale. I appreciate the preliminary exploratory contributions you have made in this area. Such contributions are neither necessary nor sufficient to prove solar-terrestrial coherence. That’s best done with metrics that are well-constrained globally by the laws of conservation of angular momentum & large numbers, as I’ve shown repeatedly in the past. However, the contributions you are making may be necessary or at least critically supplementary for other interesting purposes. I will continue monitoring your stimulating explorations, which are a welcome contribution to an often dull, politically-overcharged-&-misdirected climate discussion. Thanks sincerely for sharing your notes.
Bill Illis, thanks for sharing the seasonal west/east/combined trade wind summaries — this has triggered a valuable review of dozens of climatology animations plus SOI seasonal variability patterns. This has reminded me that I need to get back to some very clear semi-annual patterns I noticed in the Indo-Pacific Warm Pool (via KNMI data) after Harry van Loon dropped by with some valuable tips several months ago. The warm pool has a very special type of semi-annual stability that exists nowhere else on Earth so far as I have been able to tell to date.
I have grabbed the blurred preview of their figure2 , sharpened it and added annotations to clarify the text.
http://climategrog.wordpress.com/?attachment_id=282
They have plotted the first three EOFs it seems PC1 PC2 PC3.
PC1 seems dominated by three peaks around 0.2 0.25 and 0.3. That is the circa 3,4,and 5 year periods already discussed above as being all manifestations of the lunar 8.85 / 2 being split by something at around 22 years in period.
PC2 had additional peaks at 0.45 0.65 and 0.75 . Using this as 1-f and inverting we find: 22m, 34m and 4 year periods. Obviously very approximative guessing the log scale off a blurred graph but ballpark this is indeed 20 months , 30 months and the central 4.43 y peak.
In fact the three peaks in PC2 are a simple reflection of the three peaks in PC1 due to the (1-f) calculation.
using: 1/(1/pc-1/p2)*12, 1/(1/pc+1/p2)*12)
1.00 * 4.43 years => 15.50 + 9.79 months
1.00 * 5.42 years => 14.71 + 10.13 months
1.00 * 3.75 years => 16.36 + 9.47 months
They have correctly labelled the side bands as 1+f and 1-f this gives a difference of 2f between the peaks hence my earlier suggestion of a factor of two. Now I have decrypted the graph, I see their numbers are correct. Apologies to the authors for doubting 😉
So in view of what we have already shown it seems that this paper has extracted a series of patterns and frequencies that all have their origin in the interaction of the annual cycle with the [b] lunar perigee precession cycle [/b] modulated by a circa 22 year period .
First time I’ve seen a paper supporting luni-solar climate published in Nature-Geoscience.
I wonder if they realised !!
Re. lunar atmospheric tides:
What we would need are quantified mechanisms. Ocean and earth tides are gravitational, so that the moon’s influence is twice that of the sun’s. Atmospheric “tides” are driven primarily by solar radiation; secondarily by gravitation. True lunar atmospheric tides are so feeble that they are driven primarily by sea tide displacement–not by gravitational attraction of air.
Which is to say that lunar cycles can have nothing but negligible effects on weather. All who claim otherwise have some explaining to do. –AGF
I’ve been working at getting a clear understanding of these “beats” and it requires going back to basics to distinguish two similar treatments that are often confused with one another.
Here’s the link I intended to put in earlier
http://www.animations.physics.unsw.edu.au/jw/beats.htm
amplitude modulation:
the basic trig. identity that is used to relate modulation to interference patterns is this:
cos (a+b) + cos (a-b) = 2 cos a * cos b
this rearranges to gives amplitude modulation
cos a * cos b = 0.5 * (cos (a+b) + cos (a-b) )
now since spectral analysis splits things into a series of additive components , we have to interpret what we find in the spectrum as being the sum and the difference of the physical frequencies that are modulating.
Superposition (beats).
in the other direction for superposition of two signals:
cos (f1t) + cos ( f2t) = 2 cos ((f1t + f2t)/2) * cos ((f1t – f2t)/2)
which if we rename things using a and b
cos (a) + cos ( b) = 2 cos ((a + b)/2) * cos ((a – b)/2)
Now if we had perfect sampling the two forms are mathematically identical and again what we would find in the spectra would be the left-hand side , the two additive signals. However , what happens to the modulated “beats” on the right in the climate system may well mean that the faster one get smoothed out , or our sampling interval and all the averaging and data mangling breaks it up . The longer “beats” frequency may be all that is left and then it is what comes out in the spectrum. Also,if one part (usually the faster one) gets attenuated by other things in climate it may still be visible but the mathematical equivalence in gone and we detect the two, now separate frequencies.
So there we see that we can determine whether we are seeing amplitude modulation in the climate data or additive superposition of two signals. The two are very similar and that is where all the confusion comes from.
Now as I noted above from the paper’s figure 2, taken from the “observed” data, we find PC2 had peaks at 0.45 0.65 and 0.75 . That is periods of around 27m 18.5m and 16 months. Again all _very_ roughly but whatever we do, we must end up with figures of that order, not half those values.
using: 2/(1/pc-1/p2)*12, 2/(1/pc+1/p2)*12) which will give us SUPERPOSITION rather than MODULATION, which is what the authors musical analogy would suggest they intended.
1.00 * 3.75 years => 32.73 + 5.29 months
1.00 * 4.43 years => 31.00 + 5.39 months
1.00 * 5.42 years => 29.43 + 5.49 months
Now since the sun passes over the equatorial regions twice a year there is a strong six monthly signal in climate and it is likely that something around 5.4 months will get absorbed or simply lost in the noise. Since a lot of climate science focuses on “anomalies” where the seasonal average is removed, anything on a sub-annual scale is either removed or totally distorted. This means all that is detectable is longer period “beats”. This must now be interpreted using the above formula that produces a period twice as longer as the form used for modulation.
The remaining period with values around 29 -31 months are similar to what is suggested for the somewhat variable quasi-biannual oscillation : QBO .So QBO can be seen as the superposition of ENSO variations superimposed upon the annual cycle once the 5.4 month component has been lost.
This is almost certainly the peak in the paper’s PC2 curve just short of 0.5 on the frequency scale. Looking again at PC1 the rightmost peak is probably at about 0.4 which would also match 2.5 years = 30 months. Since QBO is strongly present in many tropical signals , especially atmospheric ones, this is not surprising.
Furthermore, the two major peaks probably correspond to 5.42 and 3.75 years. with possibly a much weaker 4.4 being the reason for the poor resolution of the two peaks (it was very small in the trade wind data I looked at)
http://climategrog.wordpress.com/?attachment_id=281
What is interesting and significant here is that this links QBO , with its awkward variability, back to the variability of ENSO and its bimodal origins in lunar and solar periods.
Now it would be nice to have Stuecker et al’s PC1 and PC2 in numeric format to get a more precise evaluation but I think this paper actually provides some strong evidence of the solar-lunar influence, once correctly interpreted.
PS: The sea tide bulge and “true” atmospheric tidal bulge move retrograde; the relevant (solar) atmospheric “tide” is prograde, like average sea and air currents generally. –AGF
” The [lunar??] sea tide bulge and “true” atmospheric tidal bulge move retrograde;”
retrograde with respect to what , the moon?
http://www.nature.com/ngeo/journal/vaop/ncurrent/full/ngeo1826.html
“…At the top of the vertical QBO domain, easterlies dominate, while at the bottom, westerlies are more likely to be found. (Wikipedia)
Missing Data In Domain: missing data present
”
So UCAR/NCAR climate guide’s one paragraph on this gives its source as being Wikipedia !!
I love the warning that missing data is “present” .
Moonshine has more influence on climate than tides. Tidal dissipation has sufficient energy to heat the ocean two thousandths of a degree C per century (assuming no loss). The notion that the precession of the moon’s orbit influences climate is less scientific than Hansen’s sea level predictions. Invoking LOD evidence evinces total misunderstanding of the physics. Fortnightly tides strongly affect LOD–reversibly–miniscule angular momentum is lost by conversion to heat. And of course the strength of fortnightly tides varies according to lunar precession, showing up on the LOD record as an 18 year cycle. But virtually none of this energy is available to the earth’s climate system–the process is almost entirely reversible.
Reasonable interpretation of atmospheric coupling requires that ALL tidal effects be modeled out of LOD history, leaving nothing but genuine weather bereft of lunar cycles. –AGF
Greg Goodman says:
May 29, 2013 at 11:35 am
” The [lunar??] sea tide bulge and “true” atmospheric tidal bulge move retrograde;”
retrograde with respect to what , the moon?
==================================================================
Conventionally the terms “prograde” and “retrograde” when speaking of planetary atmospheres mean with respect to the solid planets they circle, the classic case being Venus. –AGF
“But virtually none of this energy is available to the earth’s climate system–the process is almost entirely reversible. ”
I don’t think anyone is suggesting an energy transfer from the angular momentum of Earth-Moon system is responcible to heating the planet (other than the miniscule frictional loss).
However, that the 18a cycle is visible in the LOD record means it is either displacing significant amount of water on that time scale or significant transfer of water from oceans to atmosphere is happening in a way synchroneous to the lunar cycle. Possilby both.
Either the moon is affecting climate or the earth’s climate is affecting the precession of lunar perigee. Which do you consider the most likely?
First let me say my 1126 post is BS — both tides are retrograde relative to earth rotation whereas overall air and sea currents are prograde.
——————————————————————–
The fortnightly tide cycle entails equatorial water displaced toward the poles (the earth speeds up) and returning toward the equator (the earth slows back down) with negligible loss of energy. The 18 year cycle determines how much water moves toward and away from poles, hence the LOD amplitude of the fortnightly tide–still reversible–no energy available for weather. Once tides are accounted for wind speed (primarily) and sea current strength (secondarily) control LOD–also reversibly. If the jet stream speeds up the earth slows down, and when the jet slows down the earth speeds back up.
The point being, don’t confuse tidal processes with weather–they both show up in LOD but the causes must be sorted out. So no, neither does the moon affect climate nor does climate affect the angle of the moon’s orbit. There’s no statistical evidence for such effects when LOD is properly interpreted, neither is there any possible mechanism evident for such a process. –AGF
Oops, typo in formula I used. I saw something unexpected with 5m4 months and forgot to check back on it. Those values should be near 20 months.
2/(1/pc+1/p2)*12)
1.00 * 5.42 years => 29.43 + 20.26 months
1.00 * 4.43 years => 31.00 + 19.58 months
1.00 * 5.42 years => 29.43 + 20.26 months
So the peak in the middle of the “1-f” band is the other part of superposition of ENSO and annual cycles.
So it appears that Stuecker et al are taking the right approach but have failed to distinguish two fundamentally different interference patterns in their EOF of observational data.
Their broad “1-f” band is not necessary. Only the peak closest to one year is explicable as amplitude modulation. This is the 15 month value. This is due amplitude modulation in the physical domain , not “beats”.
The beats phenomenon (superposition) is also present and when correctly calculated accounts for the other two peaks they found in PC2 in the grey band they labelled “1-f”. Values around 20m and 29 months. These are in fact (1-f)/2 values.