Guest Post by Willis Eschenbach
To continue my investigations utilizing the CERES satellite dataset of top of atmosphere radiation, here is a set of curious graphs. The first one is the outgoing (upwelling) longwave radiation at the top of the atmosphere (TOA) versus the sea surface temperature, for the northern hemisphere, at the times of the solstices and equinoxes.
Figure 1. Northern Hemisphere TOA outgoing longwave, versus sea surface temperature. Colors represent latitudes, as follows: dark blue, 10°; red 30°; yellow 50°; sky blue 70°. Vertical dashed line is at 30.75°C. Horizontal dashed line is at 300 W/m2. Black solid line shows the surface upwelling longwave radiation (calculated at emissivity = 0.95). Click to enlarge.
I find this graph both interesting and puzzling.
The first puzzle to me is, why is outgoing radiation in July about 230-250 W/m2 from the pole to the Equator? I mean, the upwelling radiation from the surface (solid black line) increases by 50% from the coldest to the warmest areas … but the upwelling longwave is all about the same regardless of the sea surface temperature. How bizarre!
The second puzzle is that there seems to be a fairly hard limit of about 300 W/m2 of TOA upwelling LW. Not only that, but it doesn’t vary much month to month.
The third puzzle is that even up in the Arctic regions, there’s little seasonal change in the upwelling LW. It only swings about 30 W/m2 at the most variable point, and less as you move away from the poles.
Now, what I think is happening at the warmest temperatures is the same thing that the TOA reflected solar showed in my last post—a significant increase in clouds. Let me explain why more clouds means less upwelling longwave radiation. Remember that this is upwelling longwave radiation. Suppose we have some amount X of upwelling radiation coming from the ground. If we interpose a layer of cloud between the surface and the TOA, the cloud will absorb that upwelling LW radiation, and then re-radiate it, half upwards and half downwards. This reduces the amount of upwelling longwave at the TOA, as we see happening at the warm end of the scale above.
Here is the same analysis, but this time for the southern hemisphere.
Figure 2. Southern Hemisphere TOA outgoing longwave, versus sea surface temperature. Colors represent latitudes, as follows: dark blue, 10°; red 30°; yellow 50°; sky blue 70°. Vertical dashed line is at 30.75°C. Horizontal dashed line is at 300 W/m2. Black solid line shows the surface upwelling longwave radiation (calculated at emissivity = 0.95). Click to enlarge.
Want to know what is surprising to me about the southern hemisphere?
I’m surprised at how little the TOA upwelling longwave changes from season to season. The sun comes and goes … but the southern hemisphere upwelling LW is largely unaffected. Every season of the year it’s about the same, 200 W/m2 around the icy antarctic, rising to 300 W/m2 at about 28°C, and then dropping from there. What’s up with that?
My goodness, the amount there is to learn about this incredibly complex system has no end, I can only shake my head in awe …
w.
Willis,
Are you audaciously suggesting that the Earth’s atmosphere & thus climate, may be pretty much self-regulating as regards heat gain & loss? Who’d a thunk it? All barring Ice-Ages of course!
On a lighter note, the climate here in the PDREU state of UK is starting it’s usual slide into autumnal tones, cooling, damper, (for a change 😉 sarc), At least we had a decent summer compared with the drought that never actually lasted until December of 2012 (& was a wash out), according to our highly paid taxpayer funded chums at the Wet Office! It doesn’t hurt to remind them every now & then, either! They remind me of the guy who loses every bet he makes, until one day he gets it right, then crows about his expert judgement for ever & a day! Ho hum.
“If we interpose a layer of cloud between the surface and the TOA, the cloud will absorb that upwelling LW radiation, and then re-radiate it, half upwards and half downwards. This reduces the amount of upwelling longwave at the TOA, as we see happening at the warm end of the scale above.”
Upwelling LW will hit the bottom of the cloud layer, but because the clouds are mostly opaque to IR, you shouldn’t get much upward radiation. I’m guessing the cloud layer will be colder on the top and warmer on the bottom than you would expect from just the difference in altitude, i.e., the cloud will act as insulation would.
Quite simply, global air circulation changes as necessary to match outgoing longwave with incoming shortwave at ToA over time.
If anything seeks to disturb ToA energy balance then the global air circulation applies an equal and opposite negative response after a period of delay whilst the necessary circulation changes take effect.
I’ve been telling you all that for years and now Willis has noticed the effect.
It is an extension of the Thermostat Hypothesis (not original to Willis) to the global scenario.
The only things that can change total system energy content and globally averaged surface temperature are more atmospheric mass, a stronger gravity field or more ToA insolation.
Everything else just causes circulation changes in both air and oceans – the oceans should be regarded as part of the atmosphere for such purposes.
The sun and oceans affect the circulation in tandem to shift climate zones by up to 1000 miles over the millennial solar cycle.
How far would our emissions shift them?
I’d guess less than a mile.
“The first puzzle to me is, why is outgoing radiation in July about 230-250 W/m2 from the pole to the Equator?”
It’s not so surprising if you relate it to emitting temperature, which in that range is about 255°K (S-B). Most upwelling LW at TOA originates from GHG near the tropopause, and the temperature there is lower than 255K. The tropopause is actually higher and colder near the Equator (than the poles). Less than a quarter comes from the surface, if not cloudy, or the cloud tops otherwise. That’s where warmer SST counts, but it’s only for a fraction.
A more interesting question is how the emitting layer gets enough heat to stay at the temperature it is at. That’s a tribute to poleward heat redistribution.
Willis:
I write to ask a genuine question posed because I am not clear what you are attempting to assess by consideration of upwelling LW alone.
The total radiative output of the Earth will always almost equal the input of solar radiation to the Earth for radiative balance to exist. So, small global temperature changes will only provide a small radiative imbalance while the system adjusts to obtain radiative balance.
Hence, a near constant LW upwelling implies a near constant upwelling SR as observed from space by the CERES satellites. But, so what? The observed upwelling radiation is from the surface and all atmospheric altitudes.
I recognise that you are plotting the LW as a function of SST for different latitudes. And that may indicate something, but I am failing to see what that might be. I note your three “puzzles” especially your first one, but my question is
are you plotting the data in hope that something can be seen or are you seeking some specific information?
Please note that my question is sincere and is not mischievous. It is possible that if I have ‘missed the point’ then others may have, too.
Richard
Ooops!
I wrote
Hence, a near constant LW upwelling implies a near constant upwelling SR as observed from space by the CERES satellites
I intended to write
Hence, a near constant LW upwelling implies a near constant upwelling SW as observed from space by the CERES satellites
Sorry. Richard
Upwelling LW from the tropopause is effectively decoupled from SST because the GHG in the atmosphere and the water cycle distribute the energy through the whole of the atmosphere both vertically and horizontally.
Convective processes cannot get much significant energy into the tropopause, the cloud layer is much lower on average.
The role of CO2 in mediating the movement of LW energy from the surface to the tropopause has been known for at least a century and was accurately defined by military research in the fifties. The energy emissions at the TOA are thermodynamically constrained to match the incoming Sw energy except over short term imbalances. Extra CO2 alters the thermal gradient between the surface and TOA, it would be informative to compare the outgoing LW with the downwelling LW measurements made over the last few decades.
Does anyone know where the EDRO energy budget illustration’s source, I would like to read the article that goes with it
Nice one Willis,
its just amazes me what is hidden in the woodwork !
“I’m surprised at how little the TOA upwelling longwave changes from season to season. ”
Well what would consider a lot. The whole climate discussion is bogged down in arguing about feedbacks measured in single W/m2 units. That level is invisible on the range your plotting.
If you want to see a difference compare your SB line to a linear regression of the data. I suggest you separate tropics , we know they act in a rather special way and see fairly linear relation up to your red colour coding.
Fit linear regression to extra-tropics across the board and you will start to see a seasonal difference in slopes. You may want to consider cropping off < -1deg C where ice phase change will mess up temp relationship.
There's a slight upward curvature too, but linear will get a first stab at it.
The comparison of the slope to the ideal SB line will give you an estimation of the (negative) feedback at play and how it varies with the seasons.
The upward curvature makes it slightly less than linear suggesting the presence of a weaker +ve f/b too.
A strong linear feeback would totally flatten the line. What happens in the tropics, if I'm not mistaken, has to imply strong, non-linear feedback. Though I may have said that before 😉
stuart L says: October 9, 2013 at 2:53 am
“Does anyone know where the EDRO energy budget illustration’s source,”
Link.
If seasonal variations are small, an averaged annual map like provides a good visualisation:
http://en.wikipedia.org/wiki/File:Erbe.gif
Also note:
http://www.giantworlds.org/images/Thermal_Jupiter_cut.jpg
http://cdn.physorg.com/newman/gfx/news/2005/Saturn-hot-spot.jpg
I think I’ve never heard so loud
The Gods are laughing in the clouds.
==================
Nick Stokes says:
It’s not so surprising if you relate it to emitting temperature, which in that range is about 255°K (S-B). Most upwelling LW at TOA originates from GHG near the tropopause, and the temperature there is lower than 255K. The tropopause is actually higher and colder near the Equator (than the poles).
===
Very relevant points. The main GHG being water vapour of course.
I do wonder if the difference between Northern and Southern hemispheres for the appropriate Summer/Winter and Spring/Autumn (i.e. compare N Dec to S Jun & N Mar to S Sep, etc.), given that only geography is then changed, would not reveal the range that geography provides in these figures. That might provide some envelope to the outcomes and provide some sort of global baseline.
Willis, if you would read up on what we already know about this stuff, you would not be so perplexed.
Roy Spencer says:
October 9, 2013 at 4:23 am
————————————————
Dr. Spencer, if you stopped believing that adding radiative gases to the atmosphere would reduce the atmospheres radiative cooling ability, you would not be so wrong 😉
The Upwelling Radiation is so similar because of the equator to poles transfer of energy and the fact that an extra W/m2 of energy at the poles means a greater temperature increase than a W/m2 at the tropics.
The net Greenhouse Effect varies considerably by Latitude in both the net Temperature it adds and in the implied W/m2 of Energy it represents.
For example, the North Pole is on average 80C warmer than it should be just based on the net Solar radiation it receives. (Yes, you read that right, an astounding 80C). This translates into about 190 W/m2 of energy transfer from tropics/greenhouse effect versus the global average of 150 W/m2.
Technically, the implied Greenhouse Effect is actually the lowest at the Tropics as energy is transferred away to higher latitudes. It is -25W/m2 lower than the global average of 150 W/m2.
Implied Greenhouse Effect by Latitude. In effect, there is more consistent outgoing long-wave radiation at all the Latitudes.
http://s7.postimg.org/9ny2kh8vv/Greenhouse_Effect_by_Lat_Temp_C.png
http://s8.postimg.org/5x71x8qn9/Greenhouse_Effect_by_Lat_W_m2.png
I have to say that I find the color curves above almost impossible to read. They appear to be solid bars of vertically AND horizontally distributed points, and I don’t know how to interpret that. So I’ll refrain from commenting.
It might be better to aggregate the data and present it with vertical and horizontal error bars on some scale, I dunno. Or to just present the TOA spectroscopy from which these are presumably derived.
I’m actually working a little bit on trying to understand the feedback involved in a set of very simple single slab ODEs (basically the ones from Petty’s book turned into model ODEs) where I assume that atmospheric absorptivity is a parameter that is monotonically connected to temperature, that is to say, it has “sensitivity”. Originally I was (and to some extent still am) trying to understand why positive feedback in absorptivity doesn’t drive the atmosphere to unit absorptivity at 1.2 T_greybody, or rather, what limits there are on the functional relationship such that this does NOT happen. This seems relevant to the entire discussion on whether feedback from water vapor is net positive or net negative or if net positive, what the limits are on it such that Hansen’s “boiling seas” assertion fails to come to pass. (The latter is pretty easy, even in a single layer model — because the single layer gain is at most 1.2, and 1.2×255 = 306K, or 33 C, which is indeed hot for a mean temperature but far, far short of hot enough to boil anything.)
My secondary goal in this is to look at fluctuation-dissipation in a trivial physical model. One obvious flaw in the GCMs is that they individually fluctuate around their current smoothed GASTA prediction at an amplitude that is nearly twice as great as the actual obvserved amplitude of fluctuation, and fluctuate more more regularly than the actual climate, and have far too steep a rate of fluctuation growth compared to the regular climate. In other words, eyeballing the threads in 1.4 in AR5 SPM, they don’t JUST fail in the mean, they fail badly in their higher order moments. These moments represent their internal DYNAMICS, the way they handle forcings and positive/negative feedbacks. I’d like to try to understand in very simple terms what’s wrong here — it isn’t just “too much forcing” because when they overshoot the GCM mean it falls rapidly back across that mean and often dips almost all of the way back to the actual observed temperature — it goes DOWN too fast and too far. This suggests that they have the wrong GROSS behavior in their feedback terms, in the dissipation modes that cause fluctuations to grow and decay.
I’ve written a little snippet of matlab/octave code so far that actually does a nice job of solving the single slab model, and even gets the right behavior in the limits. In those limits it directly refutes the favorite argument of Joules/Joe Postma, that a properly implemented absorper/emitter atmosphere leads to runaway warming, just as it refutes HANSEN’S argument that (so far) suffiently positive climate sensitivity leads to runaway warming. The model is perhaps oversimplified, but I’m aware of its limitations and my goals are (I think) well within its limitations. My next goal is going to be to add some sort of delta-correlated stochastic noise in one of the very few parameters in the model — perhaps albedo or a_lw, the longwavelength absorptivity (or just to T_s itself) to make the system a set of Langevin equations. From that it is straightforward to compute the autocorrelation function for radiation only relaxation and see how feedback in a_lw affects things.
rgb
Don’t listen to Roy, Willis. Discovering stuff on your own is what makes you get up in the morning. It also makes you a much more interesting fellow than Joyless Roy.
What is also interesting is what happens at the 30C limit.
The x-axis is not the same as your previous Clouds regulate temperature article but I think the two combined show …
At the 30C sea surface temperature limit, clouds form and increase the reflection of solar radiance from about 25 W/m2 to 175 W/m2 in most of the scenarios. Solar irradiance at the surface falls by 150 W/m2 (an average of day and night – during the height of the day at 4:00 pm, this would be a very significant reduction in daytime solar radiation approaching 700 W/m2).
The outgoing long-wave also is impacted by this cloud increase at the 30C limit. It appears to me, the reduction in long-wave is about 125 W/m2 in most of the scenarios.
So, there you have it, the Cloud Feedback at the 30C limit (based on the eyeball method versus the actual numbers). The cloud increase at the 30C limit is truly a Negative influence. Reflecting 150 W/m2 of solar irradiance, holding in 125 W/m2 of long-wave.
Not a linear + 0.7 W/m2/C at this temperature but a very strong negative -25/W/m2/0-1C.
I can’t see what happens at the other sea surface temperature ranges, but we clearly have a complex non-linear function for the cloud feedback.
First, @daniel kaplan
“a 28*C span is about 10% of the Kelvin temperature of 300K. Outgoing LW radiation goes as the 4th power of the Kelvin temperature, so it should vary by about 40%. A variation from 200 to 300 W/m2 is about right.”
I appreciate that demonstration. Yes, it is correct not to think of the variation in linear terms.
@Willis
I want to comment on the 10 Deg tropics only. I see in all cases that the temperature you noted earlier on in your cloud formation time-of-day articles there is a swing in the trend of the blue dots. As the sea temperature rises there is a significant drop in LW and it is consistent through all seasons. The slope of the blue dots is definitely down-right with increasing temperature.
The only thing which can create that at ±10 latitude is a layer of clouds. Whether they are locally broken by thunderstorms punching through is a separate matter. But the blocking of LW has to be accompanied by a concurrent rise in SW reflecting off the top of the clouds, not so? Thus the total will remain about the same as is required. I say that because the alternative (heat is trapped and funneled in massive quantities under the clouds towards the poles) is untenable for such a thin atmosphere. I wouldn’t get far.
I am not sure if the data you are working from can detect the time of day or not. If it can’t, then maybe there is a data set elsewhere for some small region of the tropics.
If you had the same analysis by the hour, you should be able to see that the trend line of the blue dots changes during the day, with LW dropping rapidly as the visible clouds start to form. More specifically, they should be ‘IR-visible’ which could be sooner, and that would affect OLW. As the clouds are visible, that reflects SW insolation. Maybe there is an overlap or underlap of importance. Can’t tell.
A second parse is the sea temperature near land compared with the open ocean. I suspect there will be a land-effect in there. Pick one or there other and look for a clear difference. The ocean off west coast of Africa provides a very dry atmosphere in which water has to be gained. Very different west of India after the monsoons. I expect to see a later formation of clouds in the dryer air. The combined view would have a vertically thick line of dots from left to right which is exactly what you have produced.
Prediction: the LW will change with the time of day in those tropical areas with clear mornings, in the same manner as you have already observed for ‘cloudiness’. Hardly a surprise, but you have put some hard numbers on it. If you repeat the procedure with SW the blue dots will inflect the other way at 30°. Also not a surprise (unless there are surprises) because of unknown unknowns.
A couple of other charts to explain why the outgoing long-wave appears to be so similar.
The long-wave emitted by the tropics versus the poles is not actually that much different in W/m2 because the energy is related to the fourth power of the temperature.
The surface temps at the poles are still emitting around 200 W/m2 while the tropics should get up to 450 W/m2 (but obviously don’t given the OLR charts from Wills). This is because we are measuring the OLR from the troposphere where temps approach -18C or 240 W/m2.
http://s14.postimg.org/ck1mcrbwx/Surface_Temp_Longwave_by_Lat_W_m2.png
And then, if we are measuring from the sea surface only, well it does not vary that much across the globe. We get close to 32C is some places and sea water freezes at -1.9C. The sea ice can get colder than that but we have a much more constrained temperature range and more constrained emission of energy. The sea surface temperature is actually much higher than the land surface temperature (partly because Antarctica weights the land lower and there is less ocean at the poles because of this due to Antarctica). The Implied long-wave emitted by sea surface temps ranging from -1.5C to 34.5C (note this is not actually a linear line although it appears so in this range).
http://s24.postimg.org/kud839mxh/Sea_Surface_Temp_Longwave_W_m2.png
I think some of what Willis is observing is explained in http://www.kidswincom.net/CO2OLR.pdf. Radiation from clouds is the big rate controller. Think about how fast the temperature drops on a clear night with low humidity.
dp says:
October 8, 2013 at 10:38 pm
Upwelling and downwelling isn’t a 50/50 ratio. Most of the atmosphere is above the horizon so even some “downwelling” radiation is still aimed at the stars.
No you need to consider the geometry of the atmosphere it’s incredibly thin.
The earth’s radius is ~6370 km the scale height of the atmosphere is 8.5 km. Pick a point at 8.5km above the earth and calculate the angle subtended by a tangent to the horizon, you’ll find that the ratio of the upwelling/downwelling doesn’t vary much from 1 (i.e. 50/50).