According to a recent article in Eos (Doran and Zimmermann, ‘Examining the Scientific consensus on Climate Change‘, Volume 90, Number 3, 2009; p. 22-23 – only available for AGU members – update: a public link to the article is here), about 58% of the general public in the US thinks that human activity is a significant contributing factor in changing the mean global temperature, as opposed to 97% of specialists surveyed. The disproportion between these numbers is a concern, and one possible explanation may be that the science literacy among the general public is low. Perhaps Sherry Seethaler’s new book ‘Lies, Damn Lies, and Science’ can be a useful contribution in raising the science literacy?
The book is about science in general and about how science often is miscommunicated in the media. It addresses a range of issues, such as how statistics often is misused, how scientific progress is made in general, that the ‘scientific method’ is not always as straightforward as one might like to think, the influence of stake-holders, the importance of knowing the context of the research, relationships between science and policy, and ploys designed to bypass logic. Many of the points made in the book are probably well known for the RC readership – albeit used in different situations to the case studies discussed in the book. There is also some discussion about AGW, amongst other subjects.
One little paradox is that the book claims (p. xx) that it will empower people of all ages and educational backgrounds to think critically about science-related issues and make well-balanced decisions about them. To me, that sounds like a big promise, and after having read the book, I started to wonder whether that statement is just the sort of claims it tries to make people become more skeptical about? Or maybe Seethaler really did succeed after all – because I saw how the arguments in her book could be applied to this promise?
The book touches on AGW, and does in general do a good job in my opinion. However, I cannot avoid bringing up some small details to pick at: The description of the greenhouse effect is not quite correct, as the reader gets the impression that it involves reflecting infrared radiation back to space (p. 84). That is not the case, as the energy from the sun lies mainly in the visible spectrum, and the infra red light from the Earth is a product from the absorption of the sunlight and a re-emittance due to Planck’s law.
Another point that I think is that the book discusses the controversy around AGW, but this can be a bit misleading. If you look in the climatological field, you may not see much controversy, but if you search the web, you may see something that looks like one. But I think that this controversy to a large extent is constructed out of thin air, an impression I feel is supported by Doran and Zimmermann’s, Eos article.
I get the impression that ‘Lies, Damn Lies, and Science’ has much in common with the older book ‘Lies, Damn Lies, and Statistics’, and that they try to convey similar take-home messages.
‘Lies, Damn Lies, and Science‘ gives a nice collection of anecdotes and general tips. The book has a nice index and overview, so it’s easy to find your way through the book. I think the book is very useful for a lot of people – especially students, scientists, journalists, politicians, bureaucrats, and the voters.
jbob,
thanks again for your efforts. like i said, your graph looks very similar to tamino’s annual graph. then, tamino graphed it using 5,10,30 year averages (or something like that) and came up with those “hockey stick” graphs. you didn’t answer my questions about 1)whether you understand what tamino did, 2)whether he “did the math” right and 3)whether you think his approach is valid.
i’m also eager, but will wait as long as it takes, to see what other information you can “get out of these graphs.”
even in the annual graph you and tamino did, i DO see a big jump at the end. but i also see a similar “jump” around 1700.
as far as “leveling off” in the last few years, well, that may be true, but it only looks that way because 1997 was such a crazy unbelievably hot year that it skews the data. many of the years since then have been hotter than many of the years leading up to 1997.
J. Bob writes a frequently asked question:
> Why do we see, wrt our current discussion on CO2, is
> what appears to be a leveling off in the raw temp since 2000?
Because we are primed to think we see patterns in randomness.
Once you read Tamino you know picking 9 years isn’t sufficient to tell anything about a trend in this particular measure; it’s noisy.
http://tamino.wordpress.com/2009/01/15/what-if/
We fool ourselves all the time. Science is very new in the world:
http://calteches.library.caltech.edu/51/2/CargoCult.htm
________
“The first principle is that you must not fool yourself—and you are the easiest person to fool. So you have to be very careful about that. After you’ve not fooled yourself, it’s easy not to fool other scientists. You just have to be honest in a conventional way after that.
I would like to add something that’s not essential to the science, but something I kind of believe, which is that you should not fool the layman when you’re talking as a scientist. I’m not trying to tell you what to do about cheating on your wife, or fooling your girlfriend, or something like that, when you’re not trying to be a scientist, but just trying to be an ordinary human being. We’ll leave those problems up to you and your rabbi. I’m talking about a specific, extra type of integrity that is not lying, but bending over backwards to show how you’re maybe wrong, that you ought to do when acting as a scientist. And this is our responsibility as scientists, certainly to other scientists, and I think to laymen.”
———-
Our ancestors as youngsters saw tigers lurking in the leaves and dappled sunlight — and screeched and climbed trees — perhaps hundreds of times when there was no tiger there. They detected every tiger. The penalty for mistakenly thinking they saw what appeared to be a tiger was a little extra exercise.
Those of their young sibs who, once, missed seeing a tiger did not have children.
http://scholar.google.com/scholar?q=human+pattern+detection
“but it only looks that way because 1997 was such a crazy unbelievably hot year that it skews the data.”
And goes away if you include it in your last average. Which is why it went from “this century” (in 2001) to “the last 10 years” (in 2008).
Pretty transparent cherry-picking there.
And note that now there have been several more very hot years, they’ve gone from “it’s cooling” to “it’s flattening or cooling”.
mark,
this one (recent cooling) is probably about to blow up in their face…
unfortunately, walter, they’ll just ignore any query about it if they’re professional. They don’t have, want, or need their own theory, they just need to tear down the AGW theory.
For those who don’t care about a response, they will continue to use “it’s cooling” in the same way as “the hockey stick is wrong” arguments. They won’t listen or retract, they’ll just repeat the same old story because it lets them think it’s all A-OK. Any evidence to the contrary is not listened to. After all, anyone for AGW just wants to kill the Western economy and have us all living in caves. Just so we can get climate work that pays so handsomely compared to a CEO of an oil company or a geologist working for that company…
mark,
it’s funny (sad) that invariably when you scratch the scientific surface you find ideology and big gubmint conspiracy theory below..
Walter
Check out the posting at
http://www.imagenerd/uploads/yemp_est_3-NmQP2.gif
This shows the “raw” (1 yr ave) data on top, and a “scrubbed” temp plot below. I don’t have to much time right now, but will get back to you later with a more detail explanation of what I did to highlight more of the “signal” ( the lower frequency wave). It looks like about three low frequency waves in the plot. The first is about 10 years long. Will try to tease more info out.
Have fun!! Let me know what you think.
Walter
correction to graph
http://www.imagenerd/uploads/temp_est_3-NmQP2.gif
jbob,
darn it! the link didn’t work. it sends me to a verizon “page not found” kind of page… please try again. i like the idea of underlying “waves” in the record. sounds like music. do you know if central emgland was “sooty” in the late 1800s? wonder if that would suppress temps then?
JBob, first you have a typo temp not yemp. Secondly, what is that graph supposed to show? What physical presence is being teased out by filtering on frequency domain?
I posit that there is NO physical presence being teased out.
Noise is, if it is truly random, white noise in the frequency domain. So all you’ve done is select just as much noise in comparison to the signal as if you didn’t do the filtering.
That is, there will be a certain amount of noise at each frequency, approximately equally distributed over all frequencies. Taking out high frequencies takes out as much signal as noise, leaving you no better off than before. Except your graph has less information in it.
So your work is going to be wasted: it isn’t going to be able to show anything significant S/N ratio is going to be unchanged.
What you COULD try is add up all the same months together for 10 years then see how the September temperatures have changed. Then try 30 years addition.
And, as someone else pointed out, there’s proof that you can’t tell a lot from a 10-year mean, so it’s hardly going to be proof of anything, merely a hint.
jbob…sorry…link still doesn’t work for me.
hank,
great “cargo cult” link.
“it’s funny (sad) that invariably when you scratch the scientific surface you find ideology and big gubmint conspiracy theory below..”
What I found astoundingly blinkered was The Register was completely 100% behind the idea that AGW is a great multi-government conspiracy. Lots of bluster and bollocks about it.
But 11/9 being a single-government conspiracy was laughed at each time.
If governments around the world, thousands of scientists and the entire scientific journalistic integrity can make up AGW so that one small segment of science can get more grant money, whilst the government get more powers to ban cars, why can’t one single government with a few (couple of score) people in the know crash airplanes into buildings and deliberately kill a few thousand USians so that they can pass the PATRIOT act, start a war to remove a country that had abandoned the dollar for the euro, and give trillions of dollars to companies that employ US government officials (unofficially) manage to keep it as secret? Why would they not even try?
walter Try typing
the link in from your browser. That worked for me this morning. http://www.imagenerd.com/uploads/temp_est_3-NmQP2.gif
Will watch for your post.
jbob,
it worked! why would typing it in manually make a difference? anyway… i can see how that second graph, nicely placed below the first one for easy visual comparison, is a “muted” version of the first one. it only seems to show current warming equivalent to one that happened in the 1700s. (what happened in central england in the early 1700s?!) not much “hockey stick” in your graph. i think tamino did something different than just “smooth it out”. he changed the intervals from years to decades – presumably more in line with a climatologist’s long scale view. i’m not sure why/how his results seemed to “mute” the early part of the graph and “amplify” the later temp rise.
anybody know why that would happen with this set of numbers?
Walter – Glad you finally got it. Don’t know what happened. Anyway, after getting rid of the high freq “noise”, some lower frequency trends seem to appear. Are the 8-10 year cycles related to the solar cycles? Don’t know yet, but that will come out of the wash. The interesting thing was the downward hook the temp had recently. This was not expected, and it should provide some interesting discussions.
The method used has been around for many decades, early on in communication, and later in image processing. It involves transforming time domain information into the frequency domain, via the Fourier transform, manipulating the frequency data and transforming it back to the time domain. In image processing, space instead of time is used, and is called spatial filtering. I’m surprised I have seen little if any relating to the global warming thing, but I could be looking in the wrong places. Illustrate of the power in this method, is given in Blackman & Tukey (Tukey was co-discover of the Fast Fourier Transform FFT) in their book “Measurement of Power Spectra”. “We were able to discover in the general wave record, a very weak low-freq peak which would have surely have escaped our attention without spectral analysis. This peak, it turns out is almost certainly due to a swell from the Indian Ocean, 10,000 miles distant. Physical dimensions are: 1mm high, a kilometer long”.
Since this week is special to us, will not be broadcasting much, but will watch for you. You may look if solar activity is related to the 8-10 year peaks that show up in the lower plot.
jbob, you’re right that it is old hat: Dolby noise reduction, for example.
But it does rely on high frequency noise to be prevalent. Dolby B and C pre-amplify the higher frequencies so that the reduction is more effective.
We can’t pre-amplify high frequency noise in the climate.
Again, you haven’t seemed to have described why getting rid of high frequency noise (and high frequency signal) is appropriate to remove the noise of weather from climate.
Fine when your tape noise is mostly high frequency, but even annual is, from a climate point of view) is high frequency.
jbob,
i believe there is generally roughly an 11-yr solar cycle. most scientists attribute the “downward hook” at the end to the fact that 98 was such a “crazy hot” year. compared to that record year, the would-have-been record years in the 2000s have been called a “cooling trend”. (2005, i believe was by one set of instrument’s account hotter than 98.) anyway, if you “mute” or “average out” that record year, the hook probably goes away.
Walter – That last down hook also puzzles me. I reran the data using the last 256 years so as to get a sharp end. The FFT uses fixed blocks of data (2,4,8,…128,256,512,etc.) and I got the same downward hook. If it was just “noise”, it would have been filtered out, as you can see how smooth the curves are. The question is: is this downward hook part of the ~10 year cycle, a longer cycle, or a combination of both? What I have to do is look at some of the other lower frequencies to see what they are doing. So I’ll be down over the Easter week, but will take a look after Mon. What I would also like to do is superimpose the solar cycle data on top of the filtered data, and see what it looks like. It would also interesting to superimpose the PDO on it, what would show up.
I have to apologize at the slowness of this, but having to build this up from scratch takes some time. But it also gets you back to building things up from the simple, and forces you to think what you are doing. So far the Visual Basic coding in EXCEL took about 2 pages.
The problem with the image link was the period at the end of the URL address.
“downward hook”
“last hook”
Sorry, guys, you’re trying to explain a familiar illusion. Excel encourages this kind of thing. You can’t trust Clippy that much:
http://scienceblogs.com/stoat/upload/2007/05/5-year-trends.png
http://scienceblogs.com/stoat/2007/05/the_significance_of_5_year_tre.php#mor
http://atmoz.org/blog/2008/01/29/on-the-insignificance-of-a-5-year-temperature-trend/
Dear Hank – The methods I have used are standard signal analysis procedures. You might want to visit
http://www.dspguide.com
where I got some of the code for the FFT. I would trust dspguide as they are pros, and publish in that area.
jbob,
i was not really “puzzled” by that “hook”. like i said it’s an artifact of an extremely hot 98. i think it would “go away” if longer time-periods were analyzed. curious to see you start “layering” cycles onto the graph.
hank,
great links – especially that “atmoz” one.
“Dear Hank – The methods I have used are standard signal analysis procedures. You might want to visit
http://www.dspguide.com
where I got some of the code for the FFT. I would trust dspguide as they are pros, and publish in that area.”
Yes and everyone knows that digital signal processing is eminently suitable for ascertaining a trend in temperature data…
Oh, no. We don’t.
Please prove that DSP and filtering of some frequencies is applicable to ascertaining a TREND in noisy data.
#323 – If you listen to a radio, or watch a TV, there is you proof. How do you think the noise was removed from the signal?
jbob,
earlier i was wondering what would happen to that “hook” at the end if we took out, or “muted” 98. on another thread, chris colose pointed me to this:
http://chriscolose.wordpress.com/2009/04/07/decadal-scale-coolings-not-all-that-unusual/
it’s interesting because he analyzed the data for the last decade or so and experimented with taking out the extremely warm 98 data point.
“#323 – If you listen to a radio, or watch a TV, there is you proof. How do you think the noise was removed from the signal?”
Only because the noise on that scheme is not white noise. It is prevalent on the higher frequencies.
The weather is NOT the weather channel on TV.
#323- Do you know of any real physical process that has true “white” noise? Generally we used “pink” or “colored” noise (noise not flat over the frequency spectrum), that was measured for the particular process.
Walter – I have spent a little time looking at that down hook at the end. Checking to see if any errors were present. But after cutting off freq. higher then 5 years it keeps showing up. It seems that’s what the math says. More later.
J. Bob, you really should take this discussion over to Tamino’s blog, since he is a professional statistician specializing in time series analysis. If you’ve really stumbled upon a huge hole in how statisticians analyze trends in time series, don’t you think you should work to inform them of this fact? Informing Tamino would be a start.
Of course, doing so would expose you to the possibility that you’re simply wrong but that’s life, eh?
He’s keeps an open thread alive there at all times for posts that aren’t related to his latest ramblings, there’d be no problem at all with you posting your use of DSP techniques on temperature data.
Hi Walter We talked about solar cycles and effects on the weather/climate, etc. Using the
analysis methods we have discussed, here is a summary of what I was trying to come
up with, along with some graphs. A re-cap of getting a least error, linear
estimate of the Hadcet data from 1659-2008, as show in the top illustration of
temp_est_10
http://www.imagenerd.com/uploads/temp_est_10-k214I.gif
The middle illustration shows the error between the actual and estimated temp. This error
Labeled, Input, is used to feed into a wave analyzer to determine what frequencies are present. The curve label Output is a reconstruction of the Input to check the computational procedure. The bottom illustration is the energy (amplitude) present in the frequencies. Like the spectrum visualizer in WinAmp. From the spectral graph, one can see that there is more energy in the lower then at the higher freq. The higher freq are more flat, generally indicating “noise. NOTE:
The spectral frequencies are in years, NOT seconds, for this analysis.
The second figure temp_est_11
http://www.imagenerd.com/uploads/temp_est_11-Ncu6K.gif
shows what happens to the error signal if the high freq. are “cut off” above 0.12 years. This is
shown in the spectral graph, as the line is zero. At this point you can start to make out periodic oscillations, especially in the 0.1 year area. The bottom illustration shows this added to the linear estimation to get a more realistic view of the temp.
Note it still shows a downward trend at the end.
The next figure temp_est_12
http://www.imagenerd.com/uploads/temp_est_12-GOpNo.gif
shows freq. above 0.02 cycles/year cut off, and 50 year wave show up. Also at the end, these
low freq. waves seem to be peaking out and beginning a down trend.
Figure temp_est_13
http://www.imagenerd.com/uploads/temp_est_13-vyhH0.gif
is one of the more interesting. What I did was cut off the freq. above 0.12 cycle/yr, and
the freq. just below 0.06 yr/yr, as shown in the spectral graph. In addition, below the
output line, I plotted the sunspot activity (from http://www.climate4you.com). I think you can see the strong correlation between the activity and the temperature. To me that indicates there is something going on between the sun and our temp. in Central England.
The last plot was interesting, temp_est_14
http://www.imagenerd.com/uploads/temp_est_14-RbvN4.gif
Here I looked at the very long wave that seem to be present., almost like a wave 600 years
long of small amplitude, but still present. This might be related to the ~1000 year cycles, but more data points would have to be taken.
Any way to sum it up, from this small analysis, I would come to a VERY preliminary conclusion that the warm period we have may be ending. This conclusion is based on looking at the downward trend at the end of figs. _11, & _12. This is based on ONLY ONE sample data base, but it’s a start. It would be interesting to expand the sunspot-temp. However that will have to wait. Spring seems to have arrived here, so the outdoor work begins. Will have to put this work on hold until fall.
I hope these graphs are helpful, and I must apologize for not having them in a better form. Hopefully they will convey the analysis idea. Will look into your comments above when I get more time. You comments?
#328 Good idea, but will wait for fall.
Why 0.12 years, Bob?
If climate is a multi-decade issue (to void out and average PDO cycles), surely it should be removing any frequencies quicker than 10-30 years.
…
Claims victory. Disappears. No surprise.
J. Bob, I hope you are not under the illusion that what you are doing is science. At best, you are engaged in numerology until you 1)assess the statistical significance of your conclusions; 2)have some sort of mechanism that at least makes them plausible. You have neither.
well, jbob, i really appreciate your efforts. i have to agree with mark (#330) that you need to look at time chunks of 10 to 30 years. no apologies nec. for the format – it’s great you were able to work out a way to post anything. thanks.
#330 Sorry, the cut off freq was above 0.14 c/yr (7 yr period). Again the reason being, the spectrum looked flat with no breaks, or “groupings”. A little experience helps. The fist “grouping“ shows up at about 0.14 c/yr. Also it’s close to, and includes the solar cycles. As a result from temp_est_11, you can see a cleaner signal, with various harmonics, or waves present. What drives these harmonics ( below 0.14 c/yr), I think is the real question. I picked the 50 year cutoff (0.02 c/yr) to see what showed up. Which will take some time to evaluate. But what seemed to be important was the 60 yr cycle shown at the end, from 1940 to 2000+. To me, this looks like a downturn in the making.
I would also like to see what the Atlantic cycle as a comparison to what I did on the solar cycle figure temp_est_13, as well as the PDO. So there is still some work to do. As such, I never claimed victory, just gave and justified an opinion.
Now getting back to another comment about using science instead of “numerology”. So the mathematician Fourier and his analysis was “numerology”? I wouldn’t say that to loud in a stats class. Let me refer you to the book used by astronomers ( I believe there is a person at Goddard with an astronomy science background), Handbook of Astronomical Image Processing by Berry and Burnell, starting at p. 453.
J. Bob,
Your frequency analysis is interesting and all. What it doesn’t explain is the global shift in amplitude.