Guest Post by Willis Eschenbach
I have put forth the idea for some time now that one of the main climate thermoregulatory mechanisms is a temperature-controlled sharp increase in albedo in the tropical regions. I have explained that this occurs in a stepwise fashion when cumulus clouds first emerge, and that the albedo is further increased when some of the cumulus clouds evolve into thunderstorms.
I’ve demonstrated this with actual observations in a couple of ways. I first showed it by means of average photographs of the “view from the sun” here. I’ve also shown this occurring on a daily basis in the TAO data. So I thought, I should look in the CERES data for evidence of this putative phenomenon that I claim occurs, whereby the albedo is actively controlling the thermal input to the climate system.
Mostly, this thermoregulation appears to be happening over the ocean. And I generally dislike averages, I avoid them when I can. So … I had the idea of making a scatterplot of the total amount of reflected solar energy, versus the sea surface temperature, on a gridcell-by-gridcell basis. No averaging required. I thought well, if I’m correct, I should see the increased reflection of solar energy required by my hypothesis in the scatterplots. Figure 1 shows those results for four individual months in one meteorological year. (The year-to-year variations are surprisingly small, so these months are quite representative.)
Figure 1. Scatterplots showing the relationship between sea surface temperature (horizontal axis, in °C) and total energy reflected by each gridcell (in terawatts). I have used this measurement in preference to watts/square metre because each point on the scatterplot represents a different area. This approach effectively area-averages the data. Colors indicate latitude of the gridcell. Light gray is south pole, shading to black at the equator. Blue is north pole, shading to red at the equator. Click to enlarge
So … what are we looking at here, and what does it mean?
This analysis uses a one-degree by one-degree gridcell size. So each month of data contains 180 rows (latitude) by 360 rows (longitude) of data. Each point in each graph above is one gridcell.That’s 64,800 data points in each of the graphs. Each point is located on the horizontal axis by its temperature, and on the vertical axis by the total energy reflected from that gridcell.
The main feature I want to highlight is what happens when the ocean gets warm. From about 20°C to maybe 26°C, the amount of solar energy reflected by the system is generally dropping. You can see it most clearly in Figure 1’s March and September panels. But from about 26° up to the general oceanic maximum of just above 30°C, the amount of solar energy that is reflected goes through the roof. Reflected energy more than doubles in that short interval.
Note that as the ocean warms, the total energy being reflected first drops, and then reverses direction and increases. This will tend to keep ocean temperatures constant—decreasing reflections allow more energy in. But only up to a certain temperature. Above that temperature, the system rapidly increases the amount reflected to cut down any further warming.
Overall, I’d say that this is some of the strongest evidence that my proposed thermoregulatory system exists. Not only does it exist, but it appears to be a main mechanism governing the total amount of energy that enters the climate system.
It’s very late … my best regards to everyone, hasta luego …
w.
[UPDATE] A commenter asked that I show the northern and southern hemispheres separately. Here is the Southern Hemisphere
And the Northern. The vertical lines are at 30.75°C, nothing magical about that number, I wanted to see the temperature shift over the year and that worked.
Large file 64mb: hhttp://tinyurl.com/k88q3h6
“Radiative Flux Assessment”
from “Appendix A :
Data Product Descriptions and Supporting Detail for
Data Error Sources and Estimates”
In short:
Willis;
The aliasing is not affecting Your good work to show evidence of your claim ” main climate thermoregulatory mechanisms is a temperature-controlled sharp increase in albedo in the tropical regions”
A temperature controlled non-linear albedo feedback hypothesis is refreshing compared to all other simplified linear forcing and feedbacks theories.
lynx, just what part of that document says: “The aliasing is not affecting Your good work”??
This has got nothing to do with aliasing of the CERES data as such.
The problem is decimation and the temporal scaling of the proposed feedback mechanisms.
Time Sampling :
For time sampling errors, the ERBS spacecraft orbit samples the entire 24 hour diurnal cycle every 72 days, or close to 5 times per year. Sampling studies were carried out using 3-hourly geostationary data subsampled over the tropics at the ERBS orbit times to determine diurnal sampling errors for monthly means. Wielicki et al. (2002a,b) showed that the 20°S to 20°N monthly mean ERBS WFOV SW flux error is 1.7 Wm-2 and LW flux diurnal sampling error is 0.4 Wm-2. They also showed that use of orbit precession cycle means of 36 days for the tropics (72 days for 60°S to 60°N) dramatically reduce time sampling errors.
====
So if you’re doing 20°S to 20°N monthly mean , moderate errors. If you’re using individual point values…. expect significant errors.
Also if changing from 30 to 36 means “dramatically reduce time sampling errors” that means that there is a very large diurnal sampling error. Perhaps like the ones I pointed out.
Stephen Rasey says:
October 8, 2013 at 1:33 pm
Thanks, Steven. Man, I don’t find anything like that. Here’s what I find:
Also, it doesn’t pass over a given gridcell at exactly the same time of day each time. I say that because the satellite makes 14.575 complete revolutions in 24 hours …
So far from the situation you envisioned, of a gridcell being sampled every other day, 15 samples per month, and all the samples at the same times …
… in fact we have about 720 samples per gridcell per month, at different times of day.
Your comment is a perfect example of doing it the right way. Unlike RC Saumarez, who never gives us anything remotely falsifiable, you’ve given us actual numbers so we have something to talk about. It turns out your claims weren’t true, but that’s just science. At least you’ve put up your numbers and advanced your argument, well done.
w.
RC Saumarez says:
October 8, 2013 at 10:08 am
Contrary to your claim, latex for putting in mathematical equations is available to everyone. Start it with the dollar sign ($) and the word “latex” with no space in between. Everything from there to the next $ sign will be interpreted as latex.
is what you get from
\displaystyle { \bold{ \beta = \frac {y} {x} } }
Regarding graphs, put them up on the web somewhere, and then post a link to them. As long as you start it with http colon slash slash wordpress will convert it to a link. But I’ll go you one better. You put in the link, and I’ll convert it to an in-line graphic for all to see. Can’t say fairer than that.
Your move,
w.
@Willis Eschenbach 5:29 pm
From your SOURCE
[!!!] 10:30 am! The day is just warming up! This is the “CLOUDS” Experiment !?!?
It turns out your claims weren’t true, ….
You think so, huh?
Let’s go back to your pasted quote:
That came from the paragraph on ERBE, a low earth orbit (576-589 km), inclination of 57 degs, period 96.30 minutes, from 1984-2005.
How can that that system give “complete global coverage every one hour?”
Stephen Rasey says:
October 8, 2013 at 11:19 pm
Not so. The quote has nothing to do with ERBE. It came from the site I linked to above, and it refers to the CERES instrumentation. Here’s the full quote (emphasis mine:
Regards,
w.
RC Saumarez says:
October 8, 2013 at 3:26 pm
“Temporal scaling of the proposed feedback mechanisms?” I’ve said over and over that this is not about feedback. Nor have I discussed “feedback mechanisms”. As a result, I don’t have a clue what you mean by “temporal scaling” of “feedback mechanisms”.
Look, RC, all I’ve done here is taken the reflected shortwave data from the satellite and plotted it against the Reynolds SST data. What “feedback mechanisms” are getting temporally scaled in that process, and where?
w.
“It has a low 20-km resolution with a limb-to-limb swath width and complete global coverage every one hour.”
Since the US govt. is “down” at the moment we’ll have to rely on Wankypedia. It tells us 98.8min orbital period and 14.5 orbits per day. Not quite “every hour”.
It goes down one side and up the back. It has an instrument scanning limb to limb in longitude so, in some sense it is scanning nearly the whole surface but clearly not in an even vaguely consistent way. How does it’s image of the limb compare to what it sees overhead?
“Complete coverage” has to taken in context, and it’s a pretty distorting context.
It always passed the equator at 10.30 local time. So each orbit it passes a bit further west , that is what is meant by sun synchronous.
So scan ahead longitudes are seeing earlier in the day; behind scanning is seeing later. Now we start to understand rib cage pattern. It’s the scan angle in longitude aliasing the daily temporal build up of cloud.
As I said above the data needs breaking down to see where these patterns come from. Once we understand these artefacts they may hold useful information.
So each latitude has hourly snap-shots of longitude scan that goes back and forth in local time. Within a reasonable range before it becomes too oblique, this gives a time series that covers the onset period Willis is saying is the key to his governor.
The tropical ‘uptick’ is almost certainly similar lines just very compact and unreadable as shown because of the reduced range of temperature in the tropics.
Now if Willis would deal with the data issues and investigate the aliasing patterns I’ve pointed him too, rather than regarding the whole thing as a ego contest, he may be able to extract some very useful information that may have a very direct bearing on his hypothesis.
I don’t know what he hopes to gain from his did-didn’t argument with RC but I’d suggest his time better spend on the data.
Greg,
You don’t need to rely on wiki. Read the link below. This is from a reputable university. The folks responsible for CERES take great pains to ensure the accuracy of their data and unless your willing to replicate their ground truth experiments you don’t have a leg to stand on as far as “alias” errors go. Read the part under the guidance tab.
Also this is not a signal processing issue, Temperature is a relatively slowly changing metric at best and hourly sampling of such a low frequency event is more than sufficient I would think. This was born out by several comparison studies.
RC also seems to think the CERES folks are very careful with their data.
https://climatedataguide.ucar.edu/climate-data/ceres-ebaf-clouds-and-earths-radiant-energy-systems-ceres-energy-balanced-and-filled
v/r,
David Riser
Willis,
No you don’t have a clue what any of this means. That is the problem.
The al;iasing problem, in a nutshell, is that by averaging and decimating data, you destroy phase information. Without this you cannot tell whether you are dealing with a closed or open loop system.
You appear to think that proper analytic theory doesn’t matter. In fact it does because toy start doing things that lead to wrong conclusions.
If you did my little exercise, assuming that you can, you would understand this immediately.
@David Riser,
I am certainly not saying that the CERES data is bad data. I have the greatest admiration for the people behind it. Nevertheless, data which is sampled, which the CERES data is, is potentially aliased, simply because there are things going on in clouds with a higher frequency content. You get the same problems in imaging. Everybody knows this.
The issue here, in the last three posts. is that the data has been decimated and used with decimated data of land and sea temperature to make deductions about feedback. This is where the aliasing problem comes in because it mangles the phase information, without which you cannot determine whether you are dealing with a closed loop or open loop.situation.
The problem is that our author, by his own admission, doesn’t know any maths and certainly doesn’t know anything about signal processing, so trying to have a discussion on any of these problems is like talking to a brick wall because he simply doesn’t understand what the issues are.
RC Saumarez:
re your post at October 9, 2013 at 4:05 am
This is the fourth WUWT thread where you have attacked Willis with bombast and insults instead of reasoned arguments and/ or evidence. Furthermore, you persist in making assertions which you refuse to justify, for example this
But Willis has repeatedly demanded that you justify that assertion or stop wasting space on threads by iterating it. For example, he demanded it here
http://wattsupwiththat.com/2013/10/06/evidence-that-clouds-actively-regulate-the-temperature/#comment-1440695
But you continue with insults, assertions and nothing else. Please desist.
Richard
richardscourtney
RC’s point is rather elementary – do a search for sampling theory/Nyquist frequency. Willis is sampling something with a temporal signal so RC’s point is not open to debate.
Whether or not it renders Willis’ post useless is a different matter. But it does require that caveats are added. I think experimental issues such as whether one uses interpolated data points (with associated uncertainty) as actual control data could be added to the list of caveats. Though I don’t think this should be used to dismiss Willis’ recent work given the limitation imposed on him by the experimental setup. I think everyone needs to calm down and recognise that this isn’t a peer-reviewed article and it isn’t an experiment carried out in a controlled environment.
richard I forgot make clear in relation to the Nyquist frequency, that Willis, and by his own admission, his sampling frequency is lower than the frequency of variation in the phenomenon he wishes to measure (akin to cloud cover variation). In short, he’s way under the appropriate sampling rate and he will have an issue with aliasing…after which all RC’s points follow.
@richard Courtney
How do you suggest I justify this? I can do so easily except that this is a medium that does not allow one to do so.
It is well known that the processes involved have relatively high frequency content. This is lost when processed. In particular, phase information is lost. Without phase information, one cannot distinguish between open and closed looped systems.
This might be assertions to you, but it is elementary signal prrocessing and system theory and should be part of the background of anyone who tries to analyse signals.
For example I have written a very simple simulation to illustrate this point. However I cannot present the results as part of a scientific discussion. Since I am unable to make my point, I have suggested that Eschenbach reproduces these calculations and, if he were to do so, the nature of the problem would be immediately apparent. It is not particularly difficult.
These are reasonable scientific points of criticism and science depends on critical thinking. You will note that there are other professional signal processers on this thread who also have reservations about the approach taken here. If we are correct, then WE’s model of feedback is wrong or certainly incomplete.
Cloud feedback is a problem that has exercised many good scientists in the CERES program and elsewhere. It is acknowledged as being exceptionally difficult to analyse with current data. Any claims to have done so should be treated with utmost scepticism.
@richard Courtney.
Since this is an argument that will go nowhere, Since WE has claimed groundbreaking “science” in his analysis, I suggests that he writes the whole thing up as a proper scientific paper and attempts to publish it in a respectable journal;.
cd:
Sincere thanks for your two posts addressed to me at October 9, 2013 at 6:35 am and October 9, 2013 at 6:46 am.
To be clear, I am not disputing your interpretation of what RC Saumarez intends to say. As I said to you in another thread, if he has an explicit criticism of Willis’ argument then he should state it so the criticism can be addressed. But RC Saumarez does not do that and, instead, he flames Willis.
In this thread the nearest RC Saumarez comes to making the criticism you suggest is in his answer to you at October 8, 2013 at 5:54 am
http://wattsupwiththat.com/2013/10/06/evidence-that-clouds-actively-regulate-the-temperature/#comment-1440432
Prior to that, at October 7, 2013 at 11:34 am, Willis had replied to him by asking him for specific explanation of what he meant by “aliasing” and to justify its putative effect.
But RC Saumarez did not provide the needed clarification. Instead, he provided more arm-waving, bombast and insults. Your explanation of what he meant may be right, or not.
Perhaps he does see a Nyquist violation in what Willis has done. If so, then why does he not show the violation, and why did it require you to prompt him for him to mention it?
This is the second thread where you have provided your interpretation of what RC Saumarez means (and your suggestion was of a different meaning in the previous thread). All I am asking is that when RC Saumarez has a specific criticism of Willis’ work then
(a) he states the specific criticism because that would be useful
and
(b) he stops the inane insults because they are annoying and not useful.
Richard
RC Saumarez:
I was writing my reply to cd when you wrote your posts addressed to me atOctober 9, 2013 at 7:07 am and October 9, 2013 at 7:07 am. I apologise that this may have given the impression that I ignored your posts: in fact I did not see them and I am replying to them now.
I ask you to note my request to you in my post at October 9, 2013 at 4:25 am which you have replied; i.e.
And please read my explanation to cd of why I made that request for you to desist.
As for your saying to me
Well, that is another assertion.
You say you have done a “very simple simulation” which shows Willis is wrong but you have not described it, you have not linked to it, and you have not here explained its relevance to the actual study Willis has conducted. Furthermore, you claim that doing those things is beyond your capabilities.
Sorry, but if you think that is cogent then you need to get a friend whom you trust to explain some basic ‘home truths’.
And it is for Willis to decide when, how and where he publishes his work. I feel sure he will give your suggestion all the consideration it deserves.
Richard
richard
I agree that RC can be quite adversarial, but he has stated his case quite clearly in this instance. I don’t think he needs explain himself any further.
If so, then why does he not show the violation
From what Willis has stated, and for the above case, it is quite clear that his data is was sampled below the Nyquist frequency; and hence the issue with aliasing; and hence all the issues that follow.
Where I think RC and others have got this wrong, is that somehow they think that such shortcomings render the complete exercise a waste of time. I think with the right caveats the work is fine for a first pass. But as Roy Spencer pointed out there is plenty of work on this.
All I am asking is that when RC Saumarez has a specific criticism of Willis’ work then
(a) he states the specific criticism because that would be useful
and
He has done so time and again. You need only do a search on sampling theory and you’ll get plenty of articles explaining why you must sample at or above the Nyquist Frequency. It doesn’t take a great leap to see that RC has made valid points. In short, it shouldn’t need any further qualification.
@richard Courtney.
Set up a model. It has a forward sea heating transfer function A(s). It had a feedback of clouds with a transfer function B(s). Assume that these are of the form gain/(1/tau+s).. Let tau sea=20 days and tau clouds=2 days.
Take a band-linited input X(s), with periodic inputs if you so wish. Calulate the sea temperature, Y(s), as
Y(s)= x(s).a(s)/(1+A(s)B(s))
Solve this by substituting s=-jw and form the complex exponential discrete Fourier Transforms of the transfer function and input. Do this at say 2 hourly intervals.
Now take the input, output and feedback signals and decimate them. Compute the spectrum and spectral overlap imposed by decimation.
Perform a further simulation with the feedback removed, i.e.: the clouds do not modulate temperature.
I realise that this is grossly over-simplified but it allows easy solution in the frequency domain. The thermal capacity of tthe sea cannot in reality be represented by a single phase process and cloud formation is non-linear as is its saturation effects. However, analysis of such a model is more difficult. This simulation is of course error free and completely analytic. If there are measurement errors the problem becomes much worse. However, the model is sufficient to illustrate the problem.
Given the two sets of simulated signals attempt to establsh which one is open loop and which one is closed loop. This is inherent in the transfer function. Because you have destroyed the phase relationships due to decimation , this is not possible.
If the choice of time constants in this model is reasonable, it is clear that all phase relations which determine feedback are obliterated.
This is not an assertion, as you constantly assert, it is analysis of the problem and uses well established theory to calculate possible errors, which should be a standard component of experimental work.
I have now laid out the problem clearly and it is a reasonable criticism. If I am wrong refute it.
cd says:
October 9, 2013 at 7:48 am
First off, the “Nyquist frequency” is half the sampling rate. As a result, it is impossible to sample a data set “below the Nyquist frequency”. I suspect that you are conflating the Nyquist frequency with the Nyquist rate. The Nyquist rate is approximately twice the highest frequency of interest.
Now, in this case, the frequency of interest to me is monthly. Sure, I’d love to have samples every ten minutes … but I’m happy to concern myself with monthly variations for the moment.
That means that the Nyquist rate is about two samples per month … and CERES takes about 720 samples per month.
So no, I’m not sampling below the Nyquist rate as you claim.
Next, I’ve asked Richard to provide evidence that aliasing is an issue in this dataset. I know it’s an issue in theory, and can easily be a problem in certain kinds of analysis … but is it an issue in this dataset for this particular analysis?
Both he and you have returned with nothing but theory and handwaving. I don’t give a rat’s patootie about your theories. I’m interested in one question—can you show that this particular analysis is affected by aliasing? In other words, if there is aliasing in a movie of wagon wheels, we can see it and measure it. So … where is the aliasing going on in my analysis? How large is it when you measure it? How can we detect if it is happening? Those are the kinds of questions that need answers, and you’ve not given even one of them.
To date, all either of you have done is flapped your gums. To date, neither you nor RC has shown anything at all about my analysis. To date, neither one of you has demonstrated that sampling hourly and averaging monthly makes the slightest difference IN THE REAL WORLD IN THIS PARTICULAR ANALYSIS. Yes, you both keep saying over that your theories say it should make a difference … but you haven’t said what difference it might be making.
Sorry, not impressed.
w.
RC Saumarez:
Thankyou for your post addressed to me at October 9, 2013 at 8:51 am. It – at last – attempts to say what you mean, and it concludes saying
Firstly, I didn’t “constantly assert” anything about the “analysis” you have now provided. Indeed, I could not have done that because until now you have persistently refused to present it and have provided bombast and assertions instead.
I agree that error analysis is an essential part of any work. Unfortunately your “analysis” does not include a result to which a confidence level can be applied so such error analysis is not possible.
All you have provided are these assertions
Also, it is not possible for anybody to “refute” an analysis which has no result and relies on assumptions of time constants. Please note that all one needs to say is the sample frequency must be at least twice the frequency of the signal to be detected for Nyquist compatibility. So, the only argument required is to show that the signal Willis attempts to determine is less than double the sample rate adopted by Willis.
You have described a reasonable method to determine Nyquist compatibility and there is no reason to refute it. However, it would be reasonable to assess the application of your method to Willis’ work if you were to state that. Unfortunately, you have not provided
(a) the indication of your suggested analysis,
(b) the applicability of that analysis to Willis’ work
and
(c) the effect of that assessment on Willis results.
When you correct those oversights then anybody will be able to assess the validity of your criticism. Until then you have merely added to your arm-waving.
Please note that none of this would have arisen if you had started by asking Willis if he had checked for Nyquist compatibility.
Richard
PS I ponder what grade you would give one of your students whose work included similar oversights.
RC Saumarez says:
October 9, 2013 at 8:51 am
Actually, RC, the problem was already laid out for you—it was to demonstrate how and where your claimed aliasing is affecting my analysis above. All you’ve done is ignore that very real problem, and make up your own problem to analyze.
As a result, all of your claims about your lovely model are meaningless. I don’t care what your model does. Either you can show whether and where aliasing is affecting my analysis, or you can’t—showing that aliasing does or doesn’t occur in your wonderful model complete with a “forward sea heating transfer function” means nothing about my analysis.
w.
PS—I did love your “transfer function B(s)”, it’s so much more refined than your usual BS …