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.
Really hard to believe this analysis hasn’t been done before now.
RC Saumarez
BTW, and I should have mentioned in the Earth Sciences aliasing is very often something that one has to accept. You cannot exhaustively sample strata for example. Without doing so you never really know whether you’re sampling at a high enough resolution to capture the highest frequency components and whether aliasing is therefore an issue . But it’s quite common to compare wireline responses between different petrophysical logs in order to garner interrelationships. In short, sometimes you have to be a pragmatist. Furthermore, there is also natural processes which smooth some petrophysical properties more so than others. This isn’t laboratory science!
RCS says: “A good example of aliasing is wagon wheels in movies. As they start to go round, they appear to move in the correct direction, they then speed up and suddenly appear to reverse direction and then slow down until they appear to stop. This is because the frame rate is too slow to capture the movement and the reversal of motion is aliasing (the true frequency gets represented as a negative frequency).”
That’s a very good example of how devastating the problem can be. It’s not just a case of not being too pedantic and making do with rough data.
You earlier post showing how it can create false trends is also very relevant.
Climate may not be wagon-wheel bad but this problem seems never to have been addressed in climatology. Probably because they make up most of their own methods and ignore existing systems analysis and sig. processing knowledge.
@cd
I missed one of comments.
Averaging does not eliminate aliasing.
Given high frequency data, there can be high excursions and profound control activity, yet the average over a time record can be near zero – thus wiping out the interesting data. (averaging is a filter with a frequency response of -1/jw).
I don’t know how well these calculations can be performed. The CERES site is down (?Tea party deniers) and I can’t trawl through it. What I am saying is that one has to be really careful with this sort of analysis and absolutely nail down the possible sources of error, which is a pretty non-trivial in this case.
ps: In medicine one usually get good signals, except from images but I can see how the problems arise in Earth sciences. However we have the problem of highly non-linear, non-stationary systems to deal with – generally dismissed as “variability”! I expect you encounter the same problems. in Earth sciences.
Willis, thanks for the flux density plots.
Both the equinox months now look pretty flat for both hemispheres, with a slight max around 10 deg C. Interesting. This would add a little hysteresis to the system.
This does not make the poles more important globally, it just means the effect is more pronounced at the poles in their respective summers and is still nearly nothing when not illuminated. That makes sense to me.
NH has a clear local minimum near the pole near 3-4 deg C, which is interesting.
Also the two separate traces in NH are perhaps even more clearly defined now. (Sorry, I forgot you’d taken out land). It would be interesting to see what this is due to. My instant guess is Atlantic vs Pacific. There must be some useful information in there being two different response curves. Maybe much bigger equatorial component in Pacific.
Bill Illis says:
The 30C barrier is too evident to ignore.
What is it about this sea surface temp that cause cloud cover to explode?
Once positive feedbacks kick in it’s catastrophic change time.
A good example of +ve f/b is a classic household light switch. You push it slowly up to a point, then “snap” it bangs home against the limiting negative feedback: the casing of the switch.
It may that there is a fairly broad range of localised temperature variation (hot and cold zones ) that allows quite fine variations in the percentage of sea area affected. This will give a smooth control. If there was very little variation it would tend to be all-or-nothing, on/off regulation with large hysteresis and poor control.
One way to ensure a feedback control cct provides smooth regulation is to add some noise when comparing the input to the threshold. This adds some jitter and smooths the transition response.
The +/ve f/b would make for a very steep transition which has to happen somewhere along the scale 0 – 100 degrees. The statistical distribution of SST smooths the transition. I don’t think this implies any magical property of water at 30 deg C that we’re missing.
BTW I have seen some temps touch 32 in western Pacific in plotting ARGO this week.
Short answers is +ve feedbacks make things go snap, crack, explode or shatter. Expect rapid change in measured variables.
Willis, could you clarify what you’re plotting here in relation to the the first-cut and second-cut articles. Is this just reflected, reflected incomeing plus effect on outgoing, or that latter with solar correction?
The V shaped summer plots show a +ve feedback in cold and temperate zones leading to -ve f/b in tropics. ie in temperate latitudes warmer days => less cloud => even warmer. Tropics as you’ve detailed before are the opposite.
The cloud feedback would seem to drive SST towards 26 degC, at least in summer. Quite clever, it provides both +ve and -ve f/b to control both extremes of temperature. Neat trick.
RC Saumarez
As i said i understand what aliasing is. I write commercial software where signal processing can be an important part of the workflow. And i never suggested you can eliminate aliasing by averaging and hence the point about natural smoothing.
You’re setting laboratory standards to often very extreme analytical conditions it is not a controlled envirnment. If we took your view we wouldnt have seismic imaging.
Dec-Mar-June-Sept-Dec makes a really interesting mental film loop. Would love to see these plots “animated” like that.
Btw, what are the specific areas to the right of the dotted line? They seem to be mostly NH June-Sept. Red Sea?
@cd
I didn’t mean to be condescing or question you. I agree that you have to make all sorts of allowances and approximations in real life. Nevertheless, basic signal processing theory generally appears to be correct and when one ignores it, one can come unstuck, as I have discovered on many occasions
I’m simply saying that you cannot buck basic theory too much. Aliasing is aliasing and the question is does it matter or are its effects negiigable? . Having done a fairly simple-minded simulation of a CS with the dynamics of cloud feedback and ocean heating/coooling (linear only, which the real system isn’t), I would predict l;arge errors, even including the sign of the feedback.
As I pointed out, one has to analyse the process carefully and get an idea what errors are involved. What has not been specified here is the exact structure of the data, but the CERES data appears to have been decimated to 1 monthly samples, as is the HADCRUT data, which I feel is likely to give large errors.if the feedbacks operate on a shorter timescale of hours.
Willis, I find your analyses insightful and compelling. Keep up the good work!
Willis, I just cranked up the colour contrast and saw your plots more clearly.
NH March has two clear lines too.Definitely worth finding if that’s geographical.
The 10 deg max is a bit stronger that I first thought. It looks strong enough to present a barrier. In March there’s negative cloud feedback at high lat. pushing temps back towards just above zero.
Once through the 10 deg C barrier, +ve f/b pushing towards 26.
Even in June the local minimum will give some neg f/b pushing temps down.
This idea definitely reveals some interesting and useful info.
Stephen Wilde: In itself it is not an adequate explanation for global climate variability over centuries.
That’s for sure. It is a partial description of how part of the climate system works now. I doubt that it says anything useful about the last few centuries, but it has a serious implication for how a doubling of CO2 and possible increase in surface temperature might be understood in the next few decades.
Mike Jonas: In summary, you appear to be finding seasonal and regional effects, for which there may be seasonal and regional drivers other than absolute sea-surface temperature, and which therefore may have no implications wrt multi-year global temperature changes.
True enough, but in the classical hypothesis testing tradition, Willis has formulated a hypothesis and performed a series of tests based on the available evidence. So far, the evidence supports/doesn_discredit his hypothesis. In the standard language, “More research is needed to eludicate the mechanisms and explore alternative hypotheses.” No matter how you cut it, these are challenging results for the standard CO2-based theory of future global warming.
Matthew R Marler says:
October 7, 2013 at 10:17 am
A doubling of CO2 from its late LIA level (supposedly) around 285 ppm globally to 570 would have an effectively immeasurable impact on average temperature. We’ve already gotten to 400 ppm with (allegedly) a resulting increase of at most less than a degree C, & the next 170 ppm will have relatively less effect, since it’s logarithmic. A more than negligible effect relies on assumptions about positive feedbacks not in evidence, indeed demonstrably false.
“””””……SasjaL says:
October 6, 2013 at 4:52 am
Clouds affecting temperatures are well known to us living far north. A cloudless winter night is noticeably colder than a cloudy one……..”””””””
Nearly everybody always gets it backwards:- Clouds affecting temperatures are well known to us living far north…. A cold winter night is noticeably less cloudy than a warm one.
And the daytime before the cold night, is cooler than the daytime before the warm night, and in both cases it will be colder just before sunrise, than it was before the sunset.
Warm daytime Temperatures in the presence of some humidity, are THE CAUSE OF both the warmer night, and the clouds. The clouds do NOT cause the warmer night. And the warmer the day, the warmer the night, and the higher will be any clouds that form.
Greg says:
October 7, 2013 at 8:06 am
I have plotted temperature vs amount reflected.
w.
Well I don’t disagree with Willis’ conjecture; seems obvious to me.
But there’s more to it than that.
More cloud formation (and conditions) imply higher atmospheric water vapor concentration. Atmospheric water vapor (H2O) plus O3, and also CO2 to a smaller extent, also absorb significant amounts of incoming solar radiant energy , particularly in the case of H2O and O3.
This absorbed solar energy, never reaches the ground as short wave solar spectrum energy, even in the absence of local clouds, so it is lost from the ocean energy storage sump.
Yes there’s an increased isotropic thermal radiation emission from the warmer atmosphere, but only half of that is directed towards the surface, and when it reaches the ocean surface, which most of it does, it simply promotes more prompt evaporation, rather than warming the bulk of the ocean.
RC Saumarez, you keep talking about the problem with “aliasing”. Perhaps you could explain how that aliasing would affect a scatterplot such as the one that I have done above. What would your putative “aliasing” look like if it were actually there? How might it be detected if it existed in my plot?
So far, you’ve brought up your “suspicions” and your “concerns” and your “potential problems” and what you “feel is likely to give large errors”, which seems to be your forte—lots of worry backed by no facts.
Because what you haven’t brought up is one single scrap of actual evidence that your suspicions and concerns have validity. You just wave your hands and say trust me, I’m a signal analyst …
So you’ll have to excuse me if I’m not as impressed as cd seems to be with your line of patter. Can you show that aliasing is an issue in my figures above?
Because if you can’t, then you’re just another concern troll. Look, aliasing can indeed be an issue, as you point out.
But I doubt greatly whether the rapid rise in reflected sunlight when the ocean approaches 30°C is an artifact or a result of aliasing. And since that’s all I’ve shown in this particular thread … why are you going on and on about aliasing?
As I’ve said before to you more than once, RC, you need to either put up or shut up. Either show that aliasing is in fact a problem in this very analysis, or go be a concern troll somewhere else. I’m getting bored with you playing Banquo’s ghost, full of foreboding. Your schtick is getting old. You need to actually demonstrate that your concerns are valid … because your habit of endlessly repeating that you are so concerned and suspicious and worried ad nauseum? That spiel is not going anywhere at all.
Put up or shut up, RC. Either show that aliasing is a problem with this very analysis, or go peddle your line of grave concern somewhere else.
w.
RC Saumarez
I wasn’t offended but I think we might have been talking cross-purposes. I think you raised an excellent point, and not an obvious one, certainly not the first thing that came to my mind. But, you must admit, at face value there does seem to be something about these plots – even if it could be an artifact. Let’s assume Willis is comparing bulk characteristics. Surely an obvious analogy is biogeographical zones often explained in terms of broad, regional climate. Of course you could argue that if you look at higher temporal and spatial resolution the picture is much more complex and causality is less certain because of other micro-issues interfere. It’s only after upscaling that one sees that there might be a relationship – even if not an exact one. By upscaling, I mean using something akin to a simple summary statistic (I’m not concerned about higher frequency components that cannot be adequately sampled and yes is likely to lead to issues of aliasing; an experimental limitation). Surely, you can see that there might be some value in such an approach with all the stated caveats; as a first pass at least.
Perhaps you could offer a better experimental design – the current practical limitations.
Thank you Willis.
It’s interesting that I can have a reasonable dialogue with people on this blog. I was actually referring to your earlier posts.
I am certainly not a troll. I, and others who contribute to this blog have a highly educated knowledge of signal processing and, while you may be reluctant to accept it, might see some problems with the way you handle data.
I would like you to explain exactly how you handled the CERES data and explain why. If I think there is a problem, I will explain it to you.
RC,
Willis said he plotted the data (no signal processing), in this case there would be no alias issues caused by Willis. The sampling done by the Satellite is nothing like sampling done for digital music or digital film. Additionally the CERES team has taken the time to check their data with other observational systems to eliminate Temporal Sampling errors (which would be aliasing errors in certain types of signal processing).
You can read more about CERES here:
https://climatedataguide.ucar.edu/climate-data/ceres-ebaf-clouds-and-earths-radiant-energy-systems-ceres-energy-balanced-and-filled
v/r,
David Riser
@David Riser,
Thanks, The issue is not with the CERES data itself, but the way that it is used, in conjunction with other data to derive feedback.
Obviously the CERES data is sampled, and I agree that immense precautions are taken to get high quality data. I have huge admiration for the engineers and scientists involved in the program and their work is superb.
@Willis 11:34am, RE: concern on aliasing.
What are the data being plotted?
We have on the Y-axis, Terawatts/grid cell. I didn’t realize it until now but are you normalizing by the size of the grid cell? I suspect not, otherwise why give those units? Therefore the trend we are seeing at the low temperatures is partly convolved with the variable size of high latitude cell sizes as well as temperature.
That aside, The Y-Axis is a measurement from CERES from a sun-synchronized orbit. How many times per day does a grid cell get measured? At what times per day? How wide is the CERES measurement? Possibly in the northern latitudes, we see an average reading from many passes each day, but at the equator, we get an overhead pass on one grid cell out of 20 and the other 19 are oblique. Again, what are the times of the CERES data by latitude? I can envision that 30N and 30S are measured 15 minutes apart in a 90 minute orbit. 45N and 45S are measured 22 minutes apart. If the thunderstorms are blowing up during this hour, it might make a difference.
What is the time of the SST measurement in each grid cell? Do they coincide with the CERES? Is SST and the CERES measured from the same satelite? If so, how does one know the SST under the clouds? Is the SST a broader average reading?
But I doubt greatly whether the rapid rise in reflected sunlight when the ocean approaches 30°C is an artifact or a result of aliasing.
The quantization of the effect, may very well be affected by the sample timing. If CERES is a solar noon pass, then your plots may be underestimating the solar reflectivity from afternoon cloud development. If it is a solar 3 pm pass, then it is overstating the reflectivity. If at 60N we are seeing 40 grid cells on a pass instead of 20 at the equator. So are we getting 3 readings over a 3 hour period at 60N instead of 1 reading at the equator?
The aliasing concern derives from my personal lack of knowledge of what CERES measures, how it measures, and when it measures it compared to the SST measurement.
Along the lines of the strobicopically stopped wagon wheel, what would we know of the Earth’s weather if the GOES-13 satellite only took pictures at 1500Z each day?
@Willis Eschenbach.
You keep on telling me to put up or shut up. I have several times but the problem seems to be that you can’t understand the answer. Your inability to understand Autocorrelation functions is a case in point. Since its very difficult for me to post mathematics and graphs her, why don’t we approach this in a different way and make this an exercise for the reader?
Wtite a program to simulate your hypothesis that increased radiation leads to higher surface temperatures, this leads to higher albedo which has a negative feedback on radiation.
We’ll keep it dead simple and assume that the increase in water temperature is a first order system (which it isn’t but for the purposes of this discussion it doesn’t matter). We will also assume that cloud formation is a first order process (which it isn’t). Make the time constant of the sea, say 20 days and the time constant of cloud formation 2 days. B y keeping everything linear, this makes it very easy to calculate in the frequency domain.
Now simulate the process. Then decimate the data so we have “observations” at monthly intervals.
Now generate a different model which is exactly the same but the clouds do not have a negative feedback effect, i.e. they are just derived variables. Decimate the data in the same way.
Using this “data” show FORMALLY that one model contains feedback and the other doesn’t.