Seeing the Wood for the Trees: Interactive climate analysis for everyone

In America we have a saying, “Seeing the forest for the trees…” which is used to suggest seeing the bigger picture. In Britain, the similar saying is “Seeing the wood for the trees…”. I’ll explain this in a moment, please read on.

Last week, when Basil and I posted Part 2 of our series on seeing a solar imprint in the HadCRUT temperature record, it spawned a lot of interest, debate, replication, and criticisms. One of the criticisms from Tamino surrounding the concluding results presented in figure 5 showed how using a different statistical test nullified the results. Thats good, falsification of results by alternate methods is what science is about, and we are now reworking our approach using some different data and methods to see if we can create a robust conclusion that stands up to many different statistical tests.

In the days following the posting of the essay, there was a lot of unnecessary name calling, but out of all the negativity, I’m happy to report that something very positive has emerged in the form of one Mr. Paul Clark, an audio specialist and C++/Linux programmer in the UK.

Our essay gave him an idea, and he ran with it. He ran far and wide.

Basil and I have spent the last week helping him refine it. Mr. Clark, like many people, had questions about climate science and data analysis, and felt that sometimes it is just too complex for the layman to get their mind around the way it is often presented.

I’ve often felt that “playing around” with something technical can often be a useful learning exercise. Paul felt the same way and took our idea of pulling signals from the noisy temperature data and turned it into a splendid interactive signal processing website called

Below, you can see it displaying the signal from our original essay, replicated via FFT and other steps.

Paul writes on the main page:

This is the graph that kicked this off; Basil Copeland and Anthony Watts at Watts Up With That are using filtering techniques from economics to look for a solar cycle imprint on global temperature.

Hang on, I thought, this needs a low-pass filter in the frequency domain, and I want to be able to play with different filters and link to the results easily from blog comments. Maybe other people would like to do that too…

The ‘analyse‘ tool and this website are the result of that thought, and this is the graph that resulted, which more or less replicates their filtering technique by a different method:

HADCRUT3 with Fourier low-pass filter at harmonic 18, differentiated

Please visit the site, and see the examples page. More data sets and filters will be added, and you should consider this version 1.0, meaning errors may be present. With your help, we’ll find them. Try it out, but most of all, learn while having fun.

Web driven digital climate signal analysis is now available for the common man, the amatuer scientist, and the researcher. Thanks Paul!

0 0 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
Mike Bryant
April 8, 2008 5:21 pm


Mike Bryant
April 8, 2008 5:33 pm

The “Short-term cycles in temperature compared with CO2” example is really illustrative of CO2 following temperature (not vice versa). Who would’ve thought that it would happen so quickly?
REPLY: I’m not convinced that is real just yet. More research and verification is needed. Remember this interactive website is version 1.0, and still needs to be tested to make sure artifacts or phase shifts aren’t introduced as part of the processing.

April 8, 2008 5:38 pm

Way to go Paul! I don’t want to steal your thunder, but how are you and Basil doing with the revisions? Waiting for the revised version to see what Tamino (Hansen’s Bulldog) jumps on next.

Mike Bryant
April 8, 2008 5:43 pm

Al Gore, “Now, the relationship between CO2 and temperature is quite complex…”
No it ain’t man just look at the graph…

(Gary G) Otter
April 8, 2008 6:00 pm

The ‘HADCRUT3 with Fourier low-pass filter at harmonic 18, differentiated’ link is not working, Anthony.
REPLY: Fixed, thanks

Pamela Gray
April 8, 2008 6:40 pm

I sent an email asking Paul to add sun flare data so we can play with that as well as sun spot number. Very cool site. Felt like a kid in a candy store.
REPLY: It might also be fun to add the price of wheat.

Evan Jones
April 8, 2008 6:53 pm

Paul Clark: Nice, very nice. I’m green with envy.

Pamela Gray
April 8, 2008 6:57 pm

Take a look at:
When sunspot numbers are peaking the earth is shielded from cosmic rays due to the magnetic field of the sun. When sunspot numbers are down cosmic rays bombard the earth. If our greenhouse cover gets thinned/destroyed by cosmic rays during low sunspot numbers wouldn’t it cool during that time period? Look at the peaks and valleys of the cosmic ray measurements. Am I looking at a fairly close reverse correlation to temperature change here?
REPLY: Sorry for the delay, this one ended up in the spam filter. I’ll look at the data.

Alan S. Blue
April 8, 2008 7:39 pm

Very nice!
Are there online sources for these sorts of data sats too?
Volcanic eruption data, hurricane intensity, carbon dioxide production numbers, GDP – US, Gross Planetary Product, Aerosol data, Soot production data, size of ozone hole.

April 8, 2008 7:41 pm

This is just one more fine example of the flattening of the scientific world, along with blogs like WUWT and CA, as well as the surfacestations and galaxyzoo projects (just to name two). It’s a nice bit of work and only will get better.
One feature to add, although maybe I missed it, is the ability to import your own time series, although that may weaken security. Spectral and cross-spectral analysis would be cool too.

Pamela Gray
April 8, 2008 7:58 pm

I would look at ozone in general, not just the hole. It opens and closes in a cyclic manner, and has nothing to do with hair spray. I am wondering about the entire blanket, not just the moth holes.

April 8, 2008 9:05 pm

I always wondered what your site was all about. Very interesting. BTW, I hate Windows Vista, too. It just sucks!

Evan Jones
April 8, 2008 9:30 pm

GFG (and anyone else who hasn’t done this):
Be sure to check out the Weather Stations section (right-hand side of the page, CATEGORIES heading). That is the true gold of this site: The sad (yet comical) plight of the weather stations. Check out all the chapters (56 and counting) and you will see what is the likely root of the recent spike in temperatures. Hint: it’s not an invisible gas–it’s not “invisible” anything!
I–promise–you you will find what is contained there to be deliciously shocking, astounding, and outrageous.

April 8, 2008 10:08 pm

An interesting thing the analysis tool turns up…
Try a 12-month running mean on Hadley data from 2001 onwards like so…
Note the “sawtooth pattern”. Now try it for 19 and 20 month means like so… and which flattens out the sawtooth pattern. Does this mean anything?

April 9, 2008 12:45 am

Thank you all for your kind comments, and thanks in particular to Anthony & Basil for encouragement and positive criticism during that slow-moving part between a proof of concept and a usable product!
(Glad to see the graphs work well with the SnapShot mouseovers – that was the idea of the slightly odd layout of that page)
I *am* collecting requests for new datasets to add, which will be the fun part… Anthony & Basil have already suggested quite a few that I haven’t had time to add yet – it will come, though.
It would be useful if you could post them here, rather than mail me, so that others can help refine them; I will go through this article and check them all out even if I don’t respond individually. I need the URL of the dataset (which the site will fetch automatically), but also the kind of context and reference information I have at It turned out that for the first few datasets hunting down and writing up the reference information took almost as long as writing the import filter!
Some specific replies:
Mike: Yes, it was perhaps rather naughty of me to include the CO2/temperature one in the examples, because we’re still discussing whether it’s valid or not – but part of the point of this is that we can all play and understand what is happening enough to understand it. I’m not a climate scientist, just a hacker, so I rely on others to help work out if something is significant. It was certainly a surprise to me that CO2 should react so quickly, and at such detail, BUT the amplitude of the signal compared to the annual oscillation and the overall upward trend is really tiny.
Pam: Thanks for your mail – do you have references to sources of sun flare/ozone time series?
Alan: Likewise – Where can I find that econometric data etc.?
Gary: I had considered import of your own data; it shouldn’t weaken security if it was based on a cut-and-paste form, but it’s rather tricky to integrate into the site how it works at the moment: Let me ponder that one further. You can generate frequency spectra now, but I’m trying to work out how to convert that into wavelength spectra with so few harmonics – any help/pointers in that regard would be appreciated.
Walter: I did notice your request for a 19/20 month analysis a few weeks back, and it was one of the impetuses to do this, so anyone can try this kind of thing out. There is definitely a 19/20 month signal in the last 7 years, but I have no idea what it might be! A 6-month running mean shows it quite clearly:
Thanks again to Anthony for hosting this great open space for this kind of discussion.

Luís de Sousa
April 9, 2008 3:04 am

This website is impressive!
Any chance to get an aa-index time series among the base data? The agreement with the low-passed temperature should be quite patent.

April 9, 2008 6:53 am

Great job Mr. Clark! An exceptional tool for everyone.
Would be great if individual station data from gisstemp could be imported.
I’ve always been curious about the top of Mt Washington New Hampshire.
They have a record going back to 1933 I believe.
Anyways, wonderful job!

Anthony Isgar
April 9, 2008 8:37 am

If anyone out there can correct me, please do!
Here is the way I understand ozone and was taught it in High School AP Chemistry class by perhaps the best chemistry teacher ever 🙂
Oxygen is a diatomic molecule. In review, this means that naturally it wants to be paired with one other oxygen atom, and float around as O2. However, when radiation hits this free floating O2 molecule, it gets excited, and latches on to an extra Oxygen atom, forming O3 (otherwise known as Ozone).
Now, the Earth is spinning on its axis. This means that the atmosphere (and the planet to some degree) bulges out at the Equator and thins at the poles. In addition, the Sun shines more on the equator and less on the poles because of how our planet faces. To me this says that both of the ingredients to Ozone formation (Oxygen and radiation) are at a minimum at the poles. So it is logical to see holes at either pole.
In regards to the hole getting larger, this can be explained I believe in the same manner that the current warming is explained. As the sun’s magnetic field gets stronger, it blocks off more of the cosmic solar radiation from reaching the solar system. So the main source of radiation for the poles is reduced even further, making the hole grow larger. It would be interesting to see if anyone set up radiation detectors at the poles to see if the radiation level has dropped while the hole grew.
Second thing I am confused about is if my above premise is correct, then isn’t it the free floating O2 that protects us from the radiation by absorbing it and forming Ozone? Not the Ozone protecting us? So basically, we will always have an Ozone layer unless we run out of Oxygen? In which case, I think we would have larger problems then an ozone layer. 😛

Anthony Isgar
April 9, 2008 9:07 am

And by the way, Dupont was the company that “discovered” that Chlorofluorocarbons destroyed the Ozone layer. Conveniently, they discovered it right when their patent for their coolant fluid ran out and they already had a replacement ready.
Interesting coincidence don’t you think?

Robert Wood
April 9, 2008 9:12 am

A note of caution, running means are meaningful, in so far as they distort the signal.

April 9, 2008 9:50 am

Paul Clark –
There are a lot of nice paleo-oceanographic datasets at Especially interesting are the dO18 measurements in deep-sea cores. These have very long time scales and spectral analysis will show evidence of the Milankovitch orbital cycles.

April 9, 2008 10:09 am

Anthony and I are working on an update. But don’t expect anything dramatic. From here on, I think the progress will be incremental. The basic pattern in the warming rate holds up, as we can see from Paul’s Fourier transform. The length of the warming rate cycles go in and out of phase with the solar cycles, suggesting some other forcing at work. A couple of commenters have suggested influence from the 18.6 year lunar tide cycle, and we are looking into that. But the bidecadal peaks in the warming series — the ones that are highest, and which seem to correlate to odd numbered solar cycles — are too long to be simply 18.6 year lunar tide cycles. They actually turn out to be somewhere between the 18.6 year lunar tide cycle and the 22 year Hale cycle.
I personally think that the most significant finding is the consistent way in which the amplitude of the warming rate varies from one cycle to the next. When attention is drawn to the numerous reports of decadal and bidecadal rhythms in climate data and proxies as evidence of a solar influence on climate, those skeptical of that influence dismiss this evidence because there is supposedly no known physical basis for it. But we know that solar magnetic pole reversal influences the cosmic ray flux, and that a mechanism has been proposed by which this would influence terrestrial climate through cloud cover formation. So I think it is facile to so casually dismiss the sun as a causal factor in explaining decadal and bidecadal rhythms in terrestrial climate.
I’m even leaning in the direction that a certain kind of absence of evidence may actually be evidence of something. I’m aware of studies (tree rings, I think) that purport to find 22 year periodicities, but not 11 year periodicities. Well, what if the solar influence is stronger every other solar cycle? Might it not then be possible that in the alternating cycles, the influence is less, and the signal doesn’t always show up?
Well, enough for now. We hope to have an update soon.

Tony Edwards
April 9, 2008 11:54 am

Anthony Isgar (08:37:26) :
One additional aspect of the dreaded “hole” over the Antarctic is that there isn’t a similar hole over the Arctic, but this has been covered in the past by an author, whose name eludes me, pointing out that there are several active volcanoes in the Antarctic. These not only produce ozone damaging gases, but have an excellent delivery system, both by simple convection and by volcanic emissions. Other than that, I entirely agree with your stinks master.
On the Dupont aspect, have you read Kary Mulliss’ book, “Dancing naked in the mind field”? Great fun.

April 9, 2008 12:45 pm

Basil, don’t forget to consider harmonics when looking at curious/unexpected frequencies.

April 9, 2008 2:38 pm

Basil said:
“They actually turn out to be somewhere between the 18.6 year lunar tide cycle and the 22 year Hale cycle.”
Gary said:
“Basil, don’t forget to consider harmonics when looking at curious/unexpected frequencies.”
Also, if radio frequency ideas are used, look at the “hetrodyning” principle (two signals mixed across a non-linear device). The atmosphere is considered non-linear for mixing of radio waves (sum/difference, or 40.6/3.4 using the 18.6 and 22 year cycles).
Maybe this will make sense to someone else (it does to me because I teach electronics).

Pamela Gray
April 9, 2008 5:24 pm

Paul Clark
Here it is. The whole thing is filled with data, some of it ready to cut and paste into spread sheets and then search for correlations. Like finding a gold vein in a mountain. It also has excellent descriptions of what all the stuff is that comes out of the sun that you might be able to site and cut and paste. I have found lots of other really cool data sets on this site that are more earth bound. Explore!

Pamela Gray
April 9, 2008 7:04 pm

I keep staring at that cosmic ray data. When the sun is full of flares, its strong magnetic field shields the earth and our greenhouse ozone blanket from the affects of cosmic ray destruction of that blanket (those rays just can’t escape the magnet). A really active series of cycles would lead me to think that the blanket is an extra thick woolly one that keeps us toasty warm. Then when the sun is quiet, those cosmic rays reach all the way to earth, right down to the ground, eating away at our blanket, creating holes and thin areas. All our nice warm air escapes out the roof. And if the sun is really quiet over several cycles, we are sitting in a frig with just our underwear on.
Maybe the thicker the ozone layer is, or the longer it stays thick, the warmer it gets. CO2 is trapped as well in such a thermally encased greenhouse. In fact all kinds of stuff are trapped inside the greenhouse. But the ozone layer is what is causing the earth to stay toasty. Then when the blanket thins during minimums, the temperature decreases. It may also be that the temperature decreases unevenly since the blanket is worn thin in an uneven pattern.
Since the sun has been ramping down over the past few years as it heads toward minimum, more and more rays are getting to our ozone layer and even through it (which can easily be measured and is at several sites around the globe). Maybe that is why we haven’t seen an increase in temperature.
Let’s add cosmic ray data to the playground.
Side note: Does CO2 escape with the warm air. Is it somehow destroyed by some aspect of the rays as they penetrate through the thinned blanket?

April 9, 2008 7:12 pm

I know you guys are busy with developing a response to hansen’s bulldog’s stats, but you may be interested in a blast from the past when even NASA thought it was all about the sun, although note the rather forlorn disclaimer at the bottom of the second link;

Pamela Gray
April 10, 2008 7:00 pm

Link to the satellite info, launched in 2002(?) that measures, indirectly, ozone:
More stuff about ozone, including very cool maps:
Possible ozone data:

Pamela Gray
April 10, 2008 7:25 pm

You are gonna love this. December 1992 ozone layer was VERY THIN (like 7 to 10% BELOW normal)! It was the beginning of Solar cycle 23 (July 1992) and several scientist were as worried as a newborn’s mother on just how thin the ozone layer had become by December of 1992. Anyone care to take a guess at what global temperatures have done over time since these various scientists got their knickers in a twist over our blanket? Hint: the ozone layer thinned BEFORE temperatures started to back away from hell on earth.

%d bloggers like this: