The Intriguing Problem Of The Younger Dryas—What Does It Mean And What Caused It?

OK, so that reply either went in partially or disappeared. To try again:
Of course I’d take a bet ten years out. It’s nearly a sure thing, because you are completely clueless about what measures as far as extrapolating a fit trend is concerned (basically, nothing).
Are you familiar with: Taylor Series, the Stone-Weierstrass theorem, how both together guarantee that you can fit nearly any smooth function with a parabola in a sufficiently local neighborhood, and how that fit will fail for almost all possible smooth functions (all the ones that aren’t really parabolas) once you try to extrapolate outside of that neighborhood and cubic or higher order terms become important, and how even that doesn’t serve to explain the problem with using this in climate science, where transitions appear to be smooth nearly flat variations punctuated by sudden jumps on a decadal plus timescale (Hurst-Kolmogorov transitions).
Did you even glance at the Koutsoyannis reference I linked? Of course not, or I wouldn’t be explaining the problem to you, as his figure 1 shows how a parabola for a short sample interval becomes a linear trend for a longer one becomes a sinusoidal for a still longer one and (not shown) how all of this could still be nearly irrelevant patterned noise on a still longer timescale?
I could take a few thousand lines and explain statistics or modelling to you. Or you could take a course and learn what really means for extrapolation versus interpolation. Hmmm, I vote for the latter.
rgb

EARTH’S MAGNETIC FIELD – A weak magnetic field is associated with cooling of the earth, while a strong magnetic field is associated with a warming earth. Somebody had this backwords. It has been shown that earth’s geomagetic field increases by a factor of 3 to 5 times in strength during interglacial periods.
The above all make sense if one things an increase is cosmic rays/more clouds plays a big role in the climate. The weaker the solar magnetic field is ,along with earth’s magnetic field, the more cosmic rays will be allowed to enter into earth’s atmosphere .Hence a cooling during periods of waek magnetic fields both on the sun and earth..
The SURPRISE , for lack of a better word is the sun has SPURTS of activity within the minimum most likely ,and that is when charged particle burst from the sun are greatest, and with earth’s magnetic field already in a weakened state ,their effects are much greater then would be otherwise. I think the charge particle burst to earth’s magnetosphere ,play a role in an increase in geological activity here on earth, if the bombarment of the particles occurs when the sun is mostly in a solar minimum state.
This is why in large part there is always an increase in geological activity here on earth, in association with prolong solar minimum periods. Look at the Dalton,Maunder minimums as an example .

Okay,
I’ll bite.
The answer is… Drumroll….
Cosmic Ray incursion and deformation of the heliosphere resulting in varying levels of cosmic energy absorption throughout the solar system.
Why hasn’t it happened since?
Well, we don’t know enough about interstellar space to predict timing of these things.
But, if this theory were true, it could happen again tomorrow.

Eske Willerslev, a geogeneticist at Copenhagen University, studied genetic diversity, geographic ranges and diet of several megafauna from 50,000 years ago till the Holocene. The picture that emerged is that the animals which becme extinct at the Holocene has already been in decline for at least 20,000 years. This decline was largely attributable to reducing habitat and climate change (deepening glaciation). In the final centuries before extinction competition caused by overlap between different megafauna with similar diet (as shown by gut and fecal material) and also overlap with humans and human hunting, both may have contributed to the final extinction. However the species in question, e.g. mammoth, wild horses, wooly rhinos, had already been in decline for thousands of years.
http://www.nature.com/news/2011/021111/full/news.2011.626.html

This change in solar activity to very low levels , which started in 2005 and should last for sometime to come ,in my opinion will tell us very soon just how much influence or lack of influence, the sun exerts on earth’s climatic system.
Again I say the sun drives earth’s climatic/oceanic systems, therefore if something changes from the source(the sun) that drives the systems , it stands to reason, those changes must effect the systems it drives.
The sun has gone from very active (prior to 2005) to very inactive after 2005. I can’t believe the forces the sun exerted on earth’s climatic system (along with all the secondary effects) will not change due to the significant change in the solar activity.
Again the Dalton and Maunder Minimums lend support to what I am saying.
Time will tell.
One thing I know for sure, and that is, it is not a trace gas with a trace increase that controls the climate. The CO2 hoax, is just that, a hoax.

I cannot find any examples of temperatures rising on the earth when solar activity was low, or
cooling when solar activity was high.
Lief Svalgaard is your man. He’s got a zillion of them. Although it is really pretty easy to do — temperatures rose pretty steadily from the Dalton minimum up to the late 20th century right across significant variability in the solar cycle. Lief also offers evidence — take it or leave it, but he’s pretty serious — that the sunspot count in particular has been seriously abused as a measure of solar activity for the last 200 or so years and that corrected measures show far less correlation between solar activity and temperature (a lot of the correlation that exists comes from changes in the counting algorithm that basically found more spots compared to the earlier measures as the 20th century advanced, in coincidence with generally rising temperatures). If I understand his papers correctly — perhaps he’ll chime in (he usually does, if his name is invoked).
I will note that this isn’t universally accepted — Ushokin and others use radiometric proxies to determine solar activity across the entire Holocene; their results do seem to support the hypothesis. But the main point is that finding or not finding (counter)examples depends in part where (at which solar activity representation and at which global temperature representation, given that THAT isn’t particularly reliable either) you look. It leaves the entire proposition at least somewhat dubious, although hardly disproven.
If you want more evidence, it is hardly plausible that solar activity remained low throughout the entire 80,000 years of the last ice age. It may be an important local modulator of temperature — I suspect that it is, without agreeing that it is quite proven as things stand — and still not be an important global modulator. As, for that matter, may CO_2 be.
rgb

El Gordo says:
June 21, 2012 at 12:55 pm
Okay,
I’ll bite.
The answer is… Drumroll….
Cosmic Ray incursion and deformation of the heliosphere resulting in varying levels of cosmic energy absorption throughout the solar system.
Why hasn’t it happened since?…..
______________________________________
And then there are those who mention the Solar system bobbing in and out of the galactic plane.

…Every 60 million years or so, two things happen, roughly in synch: The solar system peeks its head to the north of the average plane of our galaxy’s disk, and the richness of life on Earth dips noticeably.
Researchers had hypothesized that the former process drives the latter, via an increased exposure to high-energy subatomic particles called cosmic rays coming from intergalactic space. That radiation might be helping to kill off large swaths of the creatures on Earth, scientists say.
The new study lends credence to that idea, putting some hard numbers on possible radiation exposures for the first time. When the solar system pops its head out, radiation doses at the Earth’s surface shoot up, perhaps by a factor of 24, researchers found….
http://www.space.com/10532-earth-biodiversity-pattern-trace-bobbing-solar-system-path.html

That could also explain the wiping out of the mega fauna but not the sudden freezing with buttercups still in the mouth and undigested plant matter in the stomach found in some specimens.

this site is notletting me write in it ,the way I want to.
That aside, I say in order for the sun to have the climatic effects I refer to, the SOLAR FLUX,
a measure of solar activity, has to stay at values of less then 90, and do so for many years.
That only happens during GRAND MINIMUMS , which we are likely to have, going forward.

As Dr. Henrik Svensmark appears to have shown that there is a link between high cosmic radiation activity and cold weather periods on Earth, based on intervals when the Earth and the solar system were going through the spiral arms of the galaxy, where such radiation is high, I think it might be worthwhile to see if the Earth might have been transiting an unusual part of the galaxy during the Younger Dryas event.
If this were, for example, the expanded remains of an ancient supernova, then there may also have been many embedded solid fragments, greatly increasing the probability of impact events during the transit interval.
If true, this would really prove the Svensmark’s theory.
Svensmark: The Cloud Mystery
“Uploaded by rwesser1 on Jul 24, 2011”
100 likes, 9 dislikes; 8,433 Views; 62:46 min
“Henrik Svensmark’s documentary on climate change and cosmic rays.”

Now for getting these results, pray do tell, where did I go and extrapolate outside of my measuring range? I stayed within the 37 years where I measured. I only looked at the other side of the binominals for means and minima which suggests to me that another change at y=0 could have occurred at ca. 42 years BP and wanted to know from Stephen Wilde how he had figured that out.
I did not proclaim that yet as a final result.
Good, because that is an example of what I’m talking about. And if you extrapolate your polynomial fit back 200 or 300 years, does it still fit the data? How about 1000 years, forward or backwards? Of course not, because your function becomes so negative that the Earth’s temperature would be negative, and negative temperatures don’t exist outside of lasers (and then, only for a special interpretation of the term “temperature”). So we agree that your fit has limited predictive value outside of the range in which it was fit. Now to find a reasonable value for that range.
I would argue that a reasonable value is at most four or five years, even more reasonably one or two. Of course, I can’t really tell from your web page (which I’d looked at multiple times generating my previous responses) but it looks like you’re trying to fit only four data points — the temperature anomaly on a 37 year basis. Your method description leaves something to be desired, so let’s go through it. You use 45 weather stations taken randomly “from all over the world”, that weren’t missing a lot of data. This is a bit self-contradictory — “randomly” means that you use a random selection method to choose weather stations out of a very large pool. However, you obviously preselect them by eliminating from the pool stations with inadequate or missing records (a non-random process). Finally, the “all over the world” part is perplexing — random selection from most pools would yields a sample — especially one with only 45 elements — that is highly biased towards populated areas which and are hardly uniform. Generally, to select stations from all over the world uniformly one has to have a hightly NON random selection process as some parts of the world have only a handful of stations where others, e.g. the United States, are extremely dense in them.
And then there are the oceans, which cover 70% of the Earth’s surface and are completely unrepresented by “weather stations”.
So when you say random, do you mean:
a) Handle all weather stations with data problems — in particular infilling (a process I absolutely hate, BTW — there are right ways and wrong ways to handle missing data and this is a very wrong way, as it is basically making stuff up in such a way that your error estimates are all horribly wrong from that point on — if you make error estimates).
b) From what is left, select a large number of sites from geographical locations at different latitudes and longitudes that yield a uniform probability of covering the planet with a random selection process;
c) Roll dice (or use a random number generator) to select out of this pool.
or do you mean that you just did b), looked over a bunch of sites and grabbed one “at random” that didn’t have data problems. I ask because this latter process cannot be described as random. Unless you have an objectively random selection process, you should not use the term, and you should recognize that it is an opportunity for bias to creep in and/or for your work to be challenged. It has long been demonstrated that humans are almost incapable of generating random elements in any selection process. If you try to write down a random string of digits, it will almost certainly fail many tests of randomness. We make lousy random number generators!
Second point — why are you working so hard? Woods For Trees will let you slice and dice and fit the last 40 years of data a few thousand different ways, all online, with far, far more data (and hence much more statistically meaningful results).
Third point — I have no real idea of how you made your table, not from your description. In particular, how did you compute the number in each column labelled “Last X years”? You call them the “average change in temperature per year”. What does this mean? How do you compute it? In words it sounds like you are (per station):
a) Forming the 37 year mean (from all of the data over all 37 years?).
b) Subtracting the mean from the data to form the anomaly (for all 37 years) to form the temperature anomaly on a 37 year basis.
So far this is all good, or at least reasonable, and very similar to what Roy Spencer does with his truly global, properly selected and averaged UAH lower troposphere data which is far, far better than yours, and which he also sometimes presents with a fit (as I noted) with a smooth curve for fun. You can see his data here:
http://wattsupwiththat.com/widget/
and even put a widget on your personal webpage that automatically updates the graph every month (Anthony doesn’t re-plot this page quite often enough, sadly, so it gets a bit out of sync with the actual widget:-(.
Then you do not describe adequately how you form the average change per year during the period indicated. Do you:
c) Take the anomaly for 37 years ago, divide by 37, and make that your first column, anomaly for 32 years, divide by 32, and make that second column, etc?
c’) Take the anomaly for 37 years ago, subtract it from the anomaly for 36 years ago, divide by 1, make that the first column, etc?
c”) Something else?
The second of these is really “the slope” at 37 years ago, but what you are probably doing is akin to the first. I won’t criticize this at this point (although I could) except to note two things: Why 37, 32, 22 and 12 years? Why not 37, 36, 35…? Why vary the interval? Why not coarse grain the average so that the same number of samples contribute on both sides of a centroid, that is look at the change per year over a seven year interval and center the slope on the middle year? Both of these are serious flaws in your approach as year 37 MIGHT have been (and in fact was) anomalous in some way (in particular, it might have been sitting on or in the El Nino bobble clearly visible in the UAH data). Picking particular years in the set is yet another form of bias, intentional or unintentional.
So then (AFAICT) you end up with the four columns of data, one entry per “randomly” selected, infilled station. Let’s focus on just one thing — the means. Your final product is:
37 32 22 12
0.015 0.013 0.014 -0.019
You fit these four points — again, if I understand your method correctly, which I may not — to:

and claim an .
Here I’m a bit dismayed. First of all, none of these points have an error estimate or a standard deviation (whatever that might mean, frankly, for the method you are using). Goodness of fit estimate are thus almost completely meaningless. Of course you have the data and might have generated the s.d. for the means for each column, in which case you could have fit the data and generated Pearson’s , which would have been at least moderately meaningful. Or perhaps you are fitting the entire table of data all at once in a routine that internally generates and (if so, what routine or toolset are you using?) How do you justify fitting the mean anomaly in this way given that different stations have very different temperature anomaly ranges (so that you automatically give the greatest weight to the stations with the greatest anomaly)?
Here is where your using different intervals to compute the slope comes back to haunt you. When you divide by 37 you suppress the variation in the data by a factor of 3 compared to dividing by 12 (if you use method c) above). So the 12 year data is incorrectly weighted in computing error estimates.
In the end, your table of data means above is obviously not quadratic, it has two local maxima and two local minima. Indeed, it could be fit exactly with a cubic — so why not use a cubic? Perhaps because the cubic gets warmer in the past? But how do you reject that possibility, since you have no a priori reason to choose a constant, linear, quadratic or cubic or fourth, fifth or fiftieth order polynomial (except that with only four points cubic is already perfect).
Then you do something really odd. You assert that since your fit has zeros — two of them — “cooling has already begun” at the year corresponding to a zero in the fit range. Cooling compared to what? The 37 year mean? You note that (since your quadratic is upside down) there is another zero 42 years ago. Was the earth “cooling” 47 years ago? 60? 100?
Of course not. And this is why your entire result is nonsense, or if not nonsense, nothing one cannot get faster, more accurately, and without the egregious claims by simply looking at the UAH data, well plotted, or looking at your own data for just the annual anomalies, well plotted, and without “selecting” particular years. Just plot the full dataset (scatter plot) or the means, per year, and please — compute the s.d. per year and include it as an error bar on the means for the year. Then fit your whatever to the full dataset — to the anomalies themselves, not to the “slopes” averaged over some variable interval. Compute for the fit, as it will be very revealing — if the error bars are as wide as the range of variation of the data from end to end of the plot, will be very small and so will . If the error bars are small, will be much larger as a smooth low order polynomial fit will not interpolate the data particularly well, but a linear may be significantly different from 0, indicating a meaningful trend on the interval.
Look, statistical analysis, and fits, aren’t “magic”. They can’t squeeze information out of noise, and most of the content of any given graph of temperature or (especially) temperature anomalies is noise! It is also very much a case of GIGO — garbage in, garbage out. There’s no reason to think that a quadratic fit to anomalies (even the strange ones you compute) should be more meaningful than a linear fit, or for that matter a constant over the interval. Every year the anomaly will, after all, be different from the mean and if you generated a completely random distribution of anomalies for the years involved with a given width around the mean, for a short enough interval you could most definitely fit a low order polynomial to the result, even when you know that the result has no meaning. It would have very, very small polynomial coefficients — just like yours does! Those small coefficients are a warning — you could be (and mostly, probably, are) fitting noise, not trend.
That’s why most people stop with computing a linear trend over short baseline data. Even that is usually meaningless, but that’s your best chance for meaning. Quadratic curvature? Difficult to infer, especially when the means themselves are manifestly not quadratic over only 4 samples, and believe me, if you fit a cubic to the data you’d get completely different quadratic and linear and constant coefficients, a sign that the expansion is not systematically convergent. Getting a linear trend clearly distinguishable from no trend is usually considered — incorrectly — to tell you something about what might happen if one continues the trend to extrapolate the future outside of the interval.
It’s incorrect because for short sample baselines, in climate science, the curves are never nice linear/smooth low order functions. Well, except for CO_2 concentration at the top of Mauna Loa — that’s a major exception to many rules. They are noisy as all hell, fluctuate up and down and sideways, have completely different “trends” as one makes small changes in the baseline of the averages. What you are doing is no different from what Hansen et. al. do/did when they drew a simple exponentially diverging fit to the temperature anomaly and predicted — oh so confidently — that the temperature right now would be some 0.2 to 0.4 C warmer than it is, extrapolating the trend from the 80s and 90s into the 2000s. Oops.
Why commit the same, stupid, oops?
rgb

ROBERT, do you see where I am coming from? I am saying once we are in a particular regime then relatively small short lived solar variablity, won’t necessarily correspond to temperature changes, because of the stronger negative feedbacks ,inherent in the climate system once it is in a particular regime.
Only extreme changes in the items that control the climate (sun for example) can bring about enough of a positive feedback, to overcome those inherent neg. feedbacks to switch the climate into another state with a different range in temperature.
So far nothing has come close to doing this up to this point, since we left the DALTON MINIMUM climate regime, and went to what we presently have. Only minor random variations have taken place since,causing random up and down temperature noise ,due to the random small oscillations in earth’s climatic system. Those(some examples) being convection rates in the tropics, to minor solar variations, to enso,to volcanic activity, to the arctic oscillation phase, pdo/amo etc etc. All strong enough to cause minor temp. changes ,but far to weak to bring earth’s climatic system to a threshold, which could bring it to another climate regime.