Don sent me his AGU paper for publication and discussion here on WUWT, and I’m happy to oblige – Anthony
Abstracts of American Geophysical Union annual meeting, San Francisco Dec., 2008
Solar Influence on Recurring Global, Decadal, Climate Cycles Recorded by Glacial Fluctuations, Ice Cores, Sea Surface Temperatures, and Historic Measurements Over the Past Millennium
Easterbrook, Don J., Dept. of Geology, Western Washington University, Bellingham, WA 98225,
Global, cyclic, decadal, climate patterns can be traced over the past millennium in glacier fluctuations, oxygen isotope ratios in ice cores, sea surface temperatures, and historic observations. The recurring climate cycles clearly show that natural climatic warming and cooling have occurred many times, long before increases in anthropogenic atmospheric CO2 levels. The Medieval Warm Period and Little Ice Age are well known examples of such climate changes, but in addition, at least 23 periods of climatic warming and cooling have occurred in the past 500 years. Each period of warming or cooling lasted about 25-30 years (average 27 years). Two cycles of global warming and two of global cooling have occurred during the past century, and the global cooling that has occurred since 1998 is exactly in phase with the long term pattern. Global cooling occurred from 1880 to ~1915; global warming occurred from ~1915 to ~1945; global cooling occurred from ~1945-1977;, global warming occurred from 1977 to 1998; and global cooling has occurred since 1998. All of these global climate changes show exceptionally good correlation with solar variation since the Little Ice Age 400 years ago.
The IPCC predicted global warming of 0.6° C (1° F) by 2011 and 1.2° C (2° F) by 2038, whereas Easterbrook (2001) predicted the beginning of global cooling by 2007 (± 3-5 yrs) and cooling of about 0.3-0.5° C until ~2035. The predicted cooling seems to have already begun. Recent measurements of global temperatures suggest a gradual cooling trend since 1998 and 2007-2008 was a year of sharp global cooling. The cooling trend will likely continue as the sun enters a cycle of lower irradiance and the Pacific Ocean changed from its warm mode to its cool mode.
Comparisons of historic global climate warming and cooling, glacial fluctuations, changes in warm/cool mode of the Pacific Decadal Oscillation (PDO) and the Atlantic Multidecadal Oscillation (AMO), and sun spot activity over the past century show strong correlations and provide a solid data base for future climate change projections. The announcement by NASA that the Pacific Decadal Oscillation (PDO) had shifted to its cool phase is right on schedule as predicted by past climate and PDO changes (Easterbrook, 2001, 2006, 2007) and coincides with recent solar variations. The PDO typically lasts 25-30 years, virtually assuring several decades of global cooling. The IPCC predictions of global temperatures 1° F warmer by 2011, 2° F warmer by 2038, and 10° F by 2100 stand little chance of being correct. “Global warming” (i.e., the warming since 1977) is over!
Figure 1. Solar irradiance, global climate change, and glacial advances. Click to enlarge
The real question now is not trying to reduce atmospheric CO2 as a means of stopping global warming, but rather (1) how can we best prepare to cope with the 30 years of global cooling that is coming, (2) how cold will it get, and (3) how can we cope with the cooling during a time of exponential population increase? In 1998 when I first predicted a 30-year cooling trend during the first part of this century, I used a very conservative estimate for the depth of cooling, i.e., the 30-years of global cooling that we experienced from ~1945 to 1977. However, also likely are several other possibilities (1) the much deeper cooling that occurred during the 1880 to ~1915 cool period, (2) the still deeper cooling that took place from about 1790 to 1820 during the Dalton sunspot minimum, and (3) the drastic cooling that occurred from 1650 to 1700 during the Maunder sunspot minimum. Figure 2 shows an estimate of what each of these might look like on a projected global climate curve. The top curve is based on the 1945-1977 cool period and the 1977-1998 warm period. The curve beneath is based on the 1890-1915 cool period and 1915-1945 warm period. The bottom curve is what we might expect from a Dalton or Maunder cool period. Only time will tell where we’re headed, but any of the curves are plausible. The sun’s recent behavior suggests we are likely heading for a deeper global cooling than the 1945-1977 cool period and ought to be looking ahead to cope with it.
Figure 2. Global temperature variation 1900 to 2008 with projections to 2100. Click to enlarge.
The good news is that global warming (i.e., the 1977-1998 warming) is over and atmospheric CO2 is not a vital issue. The bad news is that cold conditions kill more people than warm conditions, so we are in for bigger problems than we might have experienced if global warming had continued. Mortality data from 1979-2002 death certificate records show twice as many deaths directly from extreme cold than for deaths from extreme heat, 8 times as many deaths as those from floods, and 30 times as many as from hurricanes. The number of deaths indirectly related to cold is many times worse.
Depending on how cold the present 30-year cooling period gets, in addition to the higher death rates, we will have to contend with diminished growing seasons and increasing crop failures with food shortages in third world countries, increasing energy demands, changing environments, increasing medical costs from diseases (especially flu), increasing transportation costs and interruptions, and many other ramifications associated with colder climate. The degree to which we may be prepared to cope with these problems may be significantly affected by how much money we waste chasing the CO2 fantasy.
All of these problems will be exacerbated by the soaring human population. The current world population of about 6 ½ billion people is projected to increase by almost 50% during the next 30 years of global cooling (Figure 2). The problems associated with the global cooling would be bad enough at current population levels. Think what they will be with the added demands from an additional three billion people, especially if we have uselessly spent trillions of dollars needlessly trying to reduce atmospheric CO2, leaving insufficient funds to cope with the real problems.
Figure 3. Global population.
If it matters, my analysis of HadCRUT data corresponds with Easterbrook’s. Perhaps it was from he whom I first got the idea to do the analysis.
http://users.vianet.ca/~paulak2r/AGW/
My graph ( http://users.vianet.ca/paulak2r/AGW/MyModel.jpg ) does not seem to be that different than his Projection graph, but I appreciate how he added more possible projections from earlier cycles.
John M Reynolds
DB2 (07:57:51) :
Thanks for the link.
It always seems to me that we look at too short a time scale. Just for fun I created this real life example of the sort of temperatures that would be experienced by someone born in Britain in 1660 and living to 70- Average annual temp 8.87c
Some one in 1670 and living to 70 Average annual temp 8.98
1680 9.01
1690 9.05
1700 9.19
1710 9.21
1720 9.17
1730 9.14
1740 9.04
1750 9.03
1760 9.08
1770 9.10
1780 9.07
1790 9.12
1800 9.15
1810 9.13
1820 9.14
1830 9.12
1840 9.10
1850 9.14 (Start of the famously reliable Hadley global temperatures)
1860 9.17
1870 9.21
1880 9.30
1890 9.39
1900 9.40
1910 9.46
1920 9.497
1930 9.60
1940 9.70 (projected to 2009)
1950 9.76 a bit of a guess and assumes current trends continue
1960 9.83 a wild guess and assumes current trends continue
I decided to call the period from 1660 to 1880 as ‘LIA Everyman’ in as much they lived part of their lives during the little ice age, and those born from 1890 to the present day as ‘UHI Everyman’ (although no adjustments have been made to correct UHI Everyman’s well known tendency to exaggerate his (or her) temperatures)
As can be seen the ‘LIA Everyman’ born in the first few decades of the 1700’s would have experienced a life time temperature average some 0.5C below someone living in the modern era. I dont think that either LIA everyman or UHI everyman would even begin to notice the difference.
TonyB
Lief said:
“A neural network is just curve fitting [and with a ‘hidden layer’ to boot] and teaches you nothing.”
I nearly said that (curve fitting) but managed to stop myself because I think it’s almost true but not quite. Formally, what you say is of course correct but in practice there is an important difference, which is that with curve fitting you have to be explicit about your preconceptions (ie actually specify a function to fit to the data) to a greater degree than with neural nets. With neural nets you need to specify independent variables (inputs) but not the functional form. This doesn’t matter much if there is a single independent variable; eg Tglobal = f(TSI), where you do some verbal curve fitting to conclude that this formulation isn’t correct (I agree). But with an increasing number of independent variables our intuition about functional forms evaporates. Usually, faced with not a clue about what’s going on, people resort to a low order polynomial (in climate science mostly first order). I think a neural net has the potential to be much better than that. The “hidden layer” in neural nets that you’ve booted are simply the adjustable parameters, comparable to those of a low order polynomial.
As for neural nets or curve fitting teaching you nothing, I’m sure you don’t mean it. You only work with ab initio theories with no adjustable parameters? (Of course, there is curve fitting and curve fitting but your dismissal was unqualified).
anna v,
“Coupling is the fact that two equations are coupled, that a spring exists between the pendulum bobs. Coupling strength is the spring constant and that is left a variable in their study.”
My point is that in spite of their warning about mistaking synchronicity for coupling they do just that. I don’t think “coupling strength” is “in” their model at all. I think their’s is a phenomenological model with the observations relating to degrees of synchronicity (as measured by d(t)) which they take to have something to do with coupling (reasonably, but imprecisely).
Actually, my guess is that this has to do with (!) Peer Review. I think a Reviewer has commented something like “an interesting paper but the authors confuse synchronicity with coupling” and the authors have responded something like “we accept the criticism and have corrected the manuscript by inserting the passage …” . The Editor accepts the change without referring it back to the Reviewer. (I’ve done exactly that: make a minor change, that doesn’t do much damage to the overall point of the paper, to satisfy a Reviewer even if I thought they were wrong. Shame on me! Shame on Peer Review!)
What do you think of the idea that what the authors have shown is that a climate model with 4 ODEs (one each for the “indices” for PDO, ENSO etc, with nonlinear couplings) might be feasible? A first look at a mathematic model for ENSO (http://www.tahan.com/charlie/research/physics/earth_science/nino/enso-ct/ENSOreport.html)
is not encouraging. It seems that these Os are not well understood at all. But the idea would be not to model the individual processes and their interaction (in the end that would just be a GCM) but the “interactions” between the indices.
I don’t think analogue computers can stage a comeback but there would be one advantage if they did. People could actually see the “model” and see for themselves what happens if they turned this little knob here a little to the left and that one a little to the right. I think there would be lots of jokes.
davidc (13:42:26) :
As for neural nets […] teaching you nothing, I’m sure you don’t mean it.
I did and do mean it in a very strong sense. Imagine that you have a magic box [a neural network]. You tell the box that you think variables A, B, C, …, X, and Y are causative and feed in long sequences of values of A, B, C, …, X, and Y, and observed ‘response’ Z to ‘train’ the network. You then ask the box to predict a new value of Z based of a new set of {A,…,Y} and the box gives you the result Z = 42. What have you learned about the physics of the process? The box is now an oracle and oracles are inscrutable. Without your box you cannot say what Z is going to be. With understanding of the physics or the dynamics, you can always make a ‘back of the envelope’ stab at what Z will be. That is called ‘understanding’ and the neural network doesn’t give you that.
‘Curve fitting’ where you hunt for the appropriate curve can actually teach you something, namely what functional forms might be under the data. With neural networks you don’t even have that as the functions are fixed [given in construction of the network].
Lief,
It was more curve fitting teaching you nothing that I was referring to. I think most (all?) quantative scientific theories have unknown parameters or constants and curve fitting is the natural way to determine them. And if you have rival mechanistic models fitting them to data and evaluating goodness of fit is one way of seeing which model agrees best with observations.
I agree with your comments on neural nets as far as they (your comments) go. My (limited) experience with neural nets is in drug development, attempting to predict what happens when a drug is administered in humans, based on the chemical structure (published in peer reviewed journals; of course, reviewed by people who think neural nets are useful). Modelling the climate system seems to me to have some parallels with my experience: a complex system, poorly understood, lots of data which shows no obvious regularities and with plenty of error. So what I get from Tsonis is that they do a reasonable job of matching past observations with just 4 indices. I don’t believe that would be possible if you really needed to solve Navier-Stokes equations to model climate. That suggests to me that a mechanistic model with just 4 subsystems could work. But no, the neural net gives no hints how to do that.
davidc (21:29:35) :
reviewed by people who think neural nets are useful
A magic box that works can be very useful. If I knew of a Neural Network that could predict the stock market and earn me billions and it worked, I wouldn’t be concerned about that I didn’t understand how it worked, as long as is did. Understanding is a different animal and if that is sought NNs are not the way to go, so the question is really what we are after.
This is a great article and the comments are even better…I just wish I understood what you guys are talking about…
All these sideswipes at Astrology – what is that but blind prejudice – or to quote Newton in response to derision – ‘I, Sir, have studied it, whereas you have not’. Galileo and Kepler might have said the same.
Astrology is the study of cycles of consciousness – individual and mass and has no theories of causation, only correlation and prediction in relation to planetary cycles.
Just because there might be planetary influences on the sun, but no known mechanism doesn’t make it ‘astrology’.
Likewise, Svensmark’s work (now supported by a huge grant from the European Space Agency) found correlations first, then went looking for mechanisms. The jury is still out on whether what Svensmark found in labs in Denmark – due to be replicated when they repair the Large Hadron Collider, really can account for cloud seeding. But his correlations have been confirmed by others (Usoskin at Oulu, Harrison at Reading).
Almost looking at cycles forgets the complexity of time-lags (except Charles Perry at the US Geological Service looking at mid-West hydrology) – people look for immediate correlations. The ocean cycles of pressure and temperature, heat content at depth etc., all exist as pulses with teleconnections through the ocean basins. The warm phase of the PDO (whether driver or consequence of ENSO) has the power to feedback to the jetstream and the track of Atlantic depressions (hence the area of ocean subject to abstraction of heat) and the speed of heat loss is dependent on the depth of the warm water store (the NAO has built up a big store south of Iceland). The jetstream is a standing wave, so if it is jinked over the Pacific, it shifts over the Atlantic and also over Siberia – so some areas can get warmer whilst others are cooling.
I hope to get time to look more deeply at leif’s ‘no sun’ stance. Presumably all the carbon-14 and beryllium-10 data must be due to internal oscillations affecting their absorption and deposition- I can’t see it, not with both correlating to sunspot activity, especially through the Maunder Minimum. TSI might not be the key indicator, but magnetic flux and cloud are still in it for my money, as is the UV variability (Shindell). Am prepared to be persuaded otherwise!
wow, 10xcsn the evidence is GREAT and the global Warmers are out in the cold. The last winter was colder than the one before and this one promises to be even more bitter cold in the Northern Hemisphere.
Al Gore is still going to be trumpeting global warming, even after he is frozen solid in the next ice ages glaciers.
What a laugher.
Oddly at least part of the US Federal government recognizes that its getting colder. The USDA Zone map
http://www.centralfloridagarden.com/topics/hardiness/history.html
has had the zones (temperature zones) move SOUTH since its inception. To put it simply if it was getting warmer any individual zone should be moving NORTH. This is contrary to the “clearly accepted” global consensus of global warming.
The whole belief in human capacity to impact our global environment is somewhat arrogant. Further, spending time arguing about it misses the larger issues of preparing for major warming OR cooling. There are many factors that could (and have historically caused cooling and warming), and those factors can come in quite uncontrollalble form factors such as solar dust clouds, meteors, and of course krakatoa and larger volcanic events. Saying they are historical and therefore not predictable, is simply foolish. We all experienced the images of the Shoemaker-Levy Collision – and one pesky little 1/2 mile asteriod could really ruin our days.
A better model of investment would be to see if CO2 output could actually warm Mars. Safe testbed and if it gets warmer, it gives us another more viable home base. The amounts of funds being diverted for CO2 are huge, and promise to get much larger if the US Gov gets into the “enforcement of CO2 output”. Better to spend those Trillions on research, and testing on Mars, building early warning and early intervention for near earth objects. Better to invest in real polution reduction – CO2 is a natural part of earths life cycle, and its PPM has varied over time.
We can expect some sort of variation in BOTH directions, be it temperature and sea level – Sea level has been more than 100meters LOWER than it is today. Imagine what that would do to international shipping lanes.
http://upload.wikimedia.org/wikipedia/commons/c/c0/Sea_level_temp_140ky.gif
I firmly believe in planning for more extreme variance (increasing AND decreasing). Creating plans and investing smartly, not believing in a CO2 ideology that does not accept scientific debate.
One more item. I am already hearing from the idealogues of global warming that CO2 will cause global cooling also. A perfect ideology. CO2 causes warming AND cooling, protecting all outcomes to be the fault of CO2.
Does ANYONE see this is illogical?
I have some questions, as the people here seem to be generally friendly and not to demeaning. I have rudimentary understandings of these things, so I find some of this debate confusing. The questions are:
1) What is the relationship to heating (seems that would cause evaporation) to precipitation? Is this a correlation at all?
2) Would the distance from the sun matter? It seems that something that hot would swell and contract naturally from it’s own temperature variations. Is the distance between molecules in a heating scenario exponential? This would make minor fluctuations in temperature increase/decrease the size of the sun, and it’s relative effect on our temperature, accordingly, yes? I understand that we are supposed to cool off at night, when the sunlight leaves, so the sun being a factor certainly seems logical.
3) Is there anyway to measure our distance from the sun, and overlay the measurement to temperature changes?
Thanks! Again, please don’t poke fun at me for being ‘ignorant’ of advanced scientific issues, the fact is that, I am. See, I am actually in school to be an accountant, but I find this stuff to be fascinating.
And sorry about the bad grammar. Sheesh, it pays to proofread.