Little Feedback on Climate Feedbacks in the City by the Bay
by Roy W. Spencer, Ph. D.
The Fall meeting of the American Geophysical Union (AGU) here in San Francisco this week is amazing for it’s sheer size: many thousands of Earth scientists presenting talks and posters on just about every Earth science subject imaginable.
Today was my chance (PDF of presentation) to try to convince other scientists who work on the critical issue of feedbacks in the climate system that some fundamental mistakes have been made that have misled climate researchers into believing that the climate system is quite sensitive to our greenhouse gas emissions. A tough sell in only 14 minutes.
It was standing room only…close to 300 scientists by my estimate. There were only a couple of objections to my presentation…rather weak ones. Afterward I had a number of people comment favorably about the ‘different’ way I was looking at the problem.
And while that should be comforting, it is also disturbing. Since when in science did the issue of ‘causation’ become a foreign concept? When did the direction of causation between two correlated variables (in my case, clouds and temperature) become no longer important?
If temperature and clouds vary together in ‘sort of’ the same way in satellite observations as they do in climate models, then the models are considered to be ‘validated’. But my message, which might not have come across as clearly as it should have due to time constraints, was that such agreement does NOT validate the models when it comes to feedback, and feedbacks are what will determine how much of an impact humans have on the climate system.
Andrew Lacis, who works climate modeling with Jim Hansen, came up and said he agreed with me that, in general, the feedback problem is more difficult than people have been assuming. In a talk after mine, Graeme Stephens gave me a backhanded compliment when he agreed with at least my basic message that the way in which we assume the climate system functions (in my terms, what-causes-what to happen) IS important to how we then deduce how sensitive the climate is to such things as our carbon dioxide emissions.
The three organizers of the session were very gracious to invite me, since they knew my views are controversial. One of the three was Andrew Dessler, who works in water vapor feedback. I had never met Andy before, and he’s a super nice guy. They all agreed that there needs to be more debate on the subject.
But most of the talks presented followed the recipe that has become all too common in recent years: analyze the output of climate models that predict substantial global warming, and simply assume the models are somewhere near correct.
There seems to be great reluctance to consider the possibility that these computerized prophets of doom, which have required so many scientists and so much money and so many years to develop, could be wrong. I come along with an extremely simple climate model that explains the behavior of the satellite data in details that are beyond even what has been done with the complex climate models…and then the more complex models are STILL believed because…well…they’re more complex.
Besides, since my simple model would predict very little manmade global warming, it must be wrong. After all, we know that manmade global warming is a huge problem. All of the experts agree on that. Just ask Al Gore and the mainstream news media.
George E. Smith (10:58:46) :
If it worked the way it has been postulated other places, wouldn’t the whole durn thing become unstable? Warming -> more water vapor -> more GHGs -> more trapped radiation -> more warming -> more water vapor -> and so on?
Why you don’t need to be a publishing climatologist to have a superior view of climate:
“[…] recipe that has become all too common in recent years: analyze the output of climate models that predict substantial global warming, and simply assume the models are somewhere near correct.”
To go further on my own idea, that water vapor has to come down at some point, so wouldn’t precipitation go up? Precip falling as snow has an albedo effect after the clouds move away also, so there is a feedback, probably not huge, but still a feedback.
Rhys Jaggar (09:35:08) :
“……I would expect that there is some sort of impulse response to the system that could account for the phase delays or lags. The components could then be separated from the impulse response based on lag and their magnitudes examined.”
Very good point.
An impulse and the response of a dumped oscillating system.
Central England Temperature Anomaly (H. Lamb).
http://www.vukcevic.talktalk.net/CETAnomaly.jpg
http://upload.wikimedia.org/wikipedia/commons/2/2b/Damped_spring.gif
I share your pain, Dr. Spencer. After I once quit a job in frustration, my colleagues who had never once sided with me in the corporate wars started telling me how they needed me there to “Speak the truth”. Your colleagues’ distracted behavior was likely caused by their hearing their dinner bell chime about that time. The AGW dinner bell actually plays a little song that goes:
Acting on Spencer’s rant
Will cause you to lose your grant
Going with the flow
Is the smart man’s way to go
Why risk the pain of speaking true
When Spencer’s there to do it for you
CH
George E. Smith (10:58:46
‘Well of course we do know that dry desert air cools rapidly at night, but less rapidly if there are clouds. So how come that evil CO2 doesn’t keep the desert air warm at night, when there is no water.’
Back went I started looking into “Co2 drive the climate” thing, I did a comparison between Yuma, Arizona and Shreveport, Louisiana. Both have/had about the same populations number. Both are with in miles of being along the same latitude. The elevation is close to being the same. Both are about the same distant from a large body of water.
The average temperature lows were about the same, but Yuma’s average highs were 12 degrees F warmer than Shreveport. Why? Water vapor, clouds, and precipitation.
Dr. Spencer,
The reaction you experienced is the calm before the storm. AGW theory (CO2 driven catastrphism) is falling apart after incresingly dominating the public square for the past ten years.
The storm that is coming is people moving on from yet another failed apocalyptic vision. Reasonable and transparent presentations such as yours gives people a lot to ponder. They have to think about this, since the drama they have been promised is failing. It is in effect a period of mourning. You have delivered a eulogy to an apocalypse, as it were. Don’t be surprised at muted responses.
“”” Dave F (11:28:09) :
George E. Smith (10:58:46) :
If it worked the way it has been postulated other places, wouldn’t the whole durn thing become unstable? Warming -> more water vapor -> more GHGs -> more trapped radiation -> more warming -> more water vapor -> and so on? “””
Well no it wouldn’t. Positive feedback can increase the gain; but only in certain conditions does it become a runaway situation.
Easier to describe in a simple Voltage amplifier feedback system (not necessarily a typical one)
Say I have a Voltage amplifier with a forward gain (A) =10, so if I apply a one Volt step to the input, the output will step up 10 Volts (which would saturate modern Op amps that may run off 5 Volts or less).
So now I apply some positive feedback by picking off 1% of the output Voltage and adding that to the input in such a way as to increase the input. My loop gain (A.beta) is 10 x 0.01 = 0.1 so my 10 Volt output signal will generate a 0.1Volt feedback signal (beta = 0.01).
So not my total input Voltage is 1 + 0.1 = 1.1 Volts. (I’m doing this in slow time, but my amplifier really has a much shorter propagation delay).
So after the forward propagation delay and the feedback delay, my total input signal is now 1.1 Volts, and my Amplifier gain of 10 dutifully turns this into an output Voltage of 11 Volts, and the feedback facto of 0.01 changes the feedback signal to 0.11 Volts, so now my totla input is 1.11 Volts after two prop delays. This now produces an output Voltage of 11.1 Volts, and a 0.111 Volt feedback signal giving a total input signal of 1.111 Volts.
Well you can see how this is going. After an infinite amount of time, it all quits changing, and the total output swing is now 11.11111111… Volts, or 100/9 Volts. So it is like the tortoise and the hare.
Now this happened because my loop gain (A.beta) is less than 1.000
so the ouput was definitely increased by the positive feedback; but it didn’t run away.
Because of the propagation delay (all physical processes take time to act), the feedback signal is slightly delayed from the original input signal that created it, and you have to vector sum the components, instead of simply straight line add them as I did.
so real feedback systems also involve a time response; and you can’t really define the stability criteria of a feedback system unless you know what that time response is.
Guess who was the last climate scientist to actually put the time (frequency) response into his feedback model. Well don’t ask me because I have never seen anyone do it.
In the case of our GHG system, the water cheats, and before the heating gets too far out of whack, all that water vapor starts to form clouds which blocxk out more and more sunlight (as well as reflect some back into space), so it is impossible for our system to runaway; well not untill we boil away all of the oceans to eliminate clouds.
And as to more water vapor causing more precipitation; it is well known that over time (climate) total global precipitation must equal total global evaporation. That has the advantage of keeping the oceans here on the ground, instead of up in the sky.
sell SCIENCE July-7 2007 Frank Wentz (RSS) et al; “How much more Rain Will Global Warming Bring?”
The answere is (he actually measured it) is a one degree rise in mean global surface temperature causes a 7% increase in evaporation, Precipitation, and total atmospheric water.
Those highly accurate computer climate models (well they call them GCMs) agree completely with the 7% increase in total atmospheric water content; but they say that the evap/precip, is only 1% to 3% per degree C. Well that is only a factor of2 2/3 to 7 times error from what the RSS team measured with actual satellites.
It is left up to the reader to decide whether to believe those high speed computer models, or to go with what was actually measured here on planet earth.
Rhys Jaggar (09:35:08) “[…] interaction between climatologists, engineers and electronics guys might be valuable.”
Administrative structures do more than interfere with such interdisciplinary endeavors. Oh sure, every institution has some “Interdisciplinary Centre for …” with a webpage peppered with flowery language, but the reality is that interdisciplinary research-needs infinitely exceed the limited resources not earmarked for “pure” studies that fit nicely into the traditional slots which academia likes. It’s a turf war – & the catch-all “interdisciplinary” (a HUGE category) has to compete with the mutually autonomous kingdoms-of-old, as if it is merely an equal category. In short: Despite flowery language, it doesn’t work (for more than a lucky few projects at this (unevolved) stage).
You are right to raise the issue for revisiting as often as necessary.
“”” old construction worker (13:35:56) :
George E. Smith (10:58:46
‘Well of course we do know that dry desert air cools rapidly at night, but less rapidly if there are clouds. So how come that evil CO2 doesn’t keep the desert air warm at night, when there is no water.’
Back went I started looking into “Co2 drive the climate” thing, I did a comparison between Yuma, Arizona and Shreveport, Louisiana. Both have/had about the same populations number. Both are with in miles of being along the same latitude. The elevation is close to being the same. Both are about the same distant from a large body of water.
The average temperature lows were about the same, but Yuma’s average highs were 12 degrees F warmer than Shreveport. Why? Water vapor, clouds, and precipitation. “””
Lemme guess, Shreveport LA also has exactly the same humidity as Yuma AZ does. And I am sure photgraphs taken around shreveport show the typical terrain looking exactly like Yuma.
Strange what causes places to be different.
RealClimate really does suck when it comes to answering questions!!
I posted a link to the NASA carbon Dioxide map and a link to U.S. vs Global Temps and asked the question. The U.S. temperatures for the last 100 years look mainly cyclic. Warm in the 1930’s followed by some cooling and then again some warming just now reaching the 1930’s decade of warmth.
My question was how come, if the data is valid, does the Global average temp spike up drastically but the U.S. temp cycles? I stated that if Carbon dioxide were the primary driving force for the Global warming then the U.S. should show a drastic upward spike since the Carbon dioxide concentration is highest in the U.S. (before it globally diffuses). The removed the post after moderation. I actually thought that was a good question that they might consider answering. Maybe someone on this site has an answer, if possible try to explain the data in terms of AGW theory. Thanks!
Dr Spencer,
I am a great fan of your work and believe you have a very rare approach in climate science, that is to say the traditional scientific approach. I was very interested on your last comment of simplicity. I really like the simple model you have developed. It is unfortunate that the advent of high computing power has led scientists to suddenly favour complexity simply because we now can. It is completely against science to favour complexity in my view. Science demands that we go for the simplest model possible. Simple models are easier to disprove or falsify than complex models, being of a higher empirical content, and therefore should be naturally favoured by science. The more complex a model is the easier it is to account for a wider range of observations and the harder it is to falsify. F = ma is easy to falsify and therefore a very favourable scientific model. F = ma + ma^2 + ma^3 + ma^4 +va + va^2 + qa is not easy to falsify and should be a last resort for any scientist. Thankyou for your stimulating posts (here and on your blog) and for your refreshingly conventional scientific approach.
Regards,
Alex Buddery
Stephen Wilde (09:35:32) :
“You don’t seem to realise that solar radiation penetrates sea surfaces to over 100 metres ………….”
Do you have evidence for that statement ?
George E. Smith (13:48:45) :
‘Lemme guess, Shreveport LA also has exactly the same humidity as Yuma AZ does. And I am sure photgraphs taken around shreveport show the typical terrain looking exactly like Yuma.
Strange what causes places to be different.’
Yep. If it was all based on CO2, both places would look the same
“”” George E. Smith (13:48:45) :
“”” old construction worker (13:35:56) :
George E. Smith (10:58:46
‘Well of course we do know that dry desert air cools rapidly at night, but less rapidly if there are clouds. So how come that evil CO2 doesn’t keep the desert air warm at night, when there is no water.’ “””
According to my Handy dandy Black Body Radiation Charts (Warren Smith Modern Optical Engineering), almost exactly 25% of BB radiation is emitted at wavelengths shorter than the peak wavelength, and about 47% of it is emitted at wavelength longer than 1.5 times the peak.
For the solar spectrum peaking at about 500 nm, that means that 25% is at less than 500nm, and only 1% is at below 250 nm (1/2 peak); 47%a bov 750 nm, and only 1% beyond 4 microns (8xpk) In that 750 nm to 4 micron range, water vapor has many absorption bands, while CO2 has a few starting at about 1.5 microns, above which only 12% exists. The water vapor takes out a little bit less than half of that spectrum, probably accounting for 20% of the total solar radiation.
That 20% of soalr radiation removed by water vapor over Shreveport LA, warms the atmosphere, whcih allows it to hold more water, but it cools the ground sicne that 20% never makes it to the ground (as solar spectrum radiation).
In Yuma on the other hand, which is as dry as a T-Rex fossil, the ground gets the full benefit of that extra 20% of solar radiation; whcih is why the high temps are so much higher than in Shreveport.
Same goes for any dry desert; not only does the temperature crash at night but the ground (surface) temperatures are much higher during the days, because of the higher solar radiation at ground level.
Bear in mind, in the Weather is not climate vein; that when you talk about more clouds bringing more cooling; you are talking about an increase in average cloud coverage, over climate time scales; last night’s weather doesn’t matter much.
George, as I have said before the only thing worst than being cold and hungry is being WET, cold and hungry.
The initial question should be; where does the Earth get its heat? From the inside molten core, or from something outside the Earth? Perhaps both?
Another question might be; when there is an ice age, what happens to the molten core? Has it cooled down? Is it affected at all? If it cools, what mechanism heats it up again?
The molten core seems to have an effect on climate when it erupts through volcanoes, which affect the climate by cooling it with the clouds formed on the sulphur particles.
With regard to the outside heat, daily the earth is heated by the sun, our closest star.
What cools the earth? Well, shadow does, as the parts of the earth experience night when they are on the other side of the earth from the sun.
The next thing that cools the earth is clouds.
How do clouds form?
Lower altitude clouds and higher clouds seem to form in different ways.
The lower clouds are being shown to form around electrons set free by cosmic rays. More cosmic rays, more clouds.
What affects the number of cosmic rays reaching the earth?
The solar wind deflects a lot. There is more solar wind when there are sunspots.
The sun, as it travels on its orbit through the Milky Way meets up with areas where there is more cosmic rays, and areas where there is not.
Ni Shaviv’s work on the course of the sun through the Milky way, and Svensmark’s work on cloud formation by cosmic rays at the Danish National Space Center, and the SKY and CLOUD experiments, to me explain the major and minor ice ages and the current interglacial warm period.
I cannot find any theoretical work done on the molten core contribution to our climate, although I am sure someone will be able to point me to some.
The computer climate models using CO2 as the driver of climate, do not appear to be able to explain the ice and warming periods of the past, whereas increasingly research is showing that the time frame of cooling and warming has a correlation with the earth’s position in the solar system, and the cosmic rays hitting the earth.
Norman;
“I(Norman) stated that if Carbon dioxide were the primary driving force for the Global warming then the U.S. should show a drastic upward spike since the Carbon dioxide concentration is highest in the U.S. (before it globally diffuses).”
Is it ? In terms of humans per square mile the USA is at about 1/10th of the UK and about 1/15th of Belgium.
So I might suggest that since Europe is running at about 10 times the number of humans per square mile the CO2 concentration should be higher in Europe than the USA (before it globally diffuses).
Adding that CO2 production is caused not just directly by humans but human activity – industry, intensive agriculture, etc.
Roughly;
Canada 8 persons per square mile
USA 82
UK 651
Germany 594
France 305
To old construction worker.
To add to George E. Smith’s explanation. Weather may not be the same as climate if you define climate in terms of long term averages, but the processes that govern both are the same. Radiation heat transfer in space is for practical purposes “line of site and speed of light”. On a clear, calm, and dry night the surface of the earth cools just as fast as heat can be conducted from the ground below to the surface. That rate is directly proportional to temperature differences while radiation to space is essentially proportional to the difference between the surface temperature and outer space temperature to the fourth power. It is possible that any green house gases may reduce the coefficient (Stephan-Boltzman constant), but it is not likely that it will ever be to the extent that conduction is not the rate controlling factor (think resistances in series).
Norman (14:13:55) :
My question was how come, if the data is valid, does the Global average temp spike up drastically but the U.S. temp cycles?
Because there are mny parts of the world we don’t hear from, so GISS and CRU can get away with being selective about the station records they use. In the U.S. they can’t, so you see a more realistic picture.
Keith Minto (14:24:48) :
Stephen Wilde (09:35:32) :
“You don’t seem to realise that solar radiation penetrates sea surfaces to over 100 metres ………….”
Do you have evidence for that statement ?
I’d say several tens of metres rather than over 100. It varies depending on the amount of turbidity caused by biota and tidally stirred silt. 70 metres seems to be about the limit for measurable opacity.
Dr. Spencer,
Really good models, like theories, should predict an observation not yet made. What predictions does your model make that others can experimentally verify?
BTW, I am a big fan. Keep up the good (and difficult but much appreciated) work.
Mike Ramsey
Stephen Wilde (06:36:23) :
pochas (06:45:26) :
Jim (08:37:39) :
There are some major effects at play here which will tend to deeply confuse the attributions of cause and effect.
On the oceans the blooming of cyanobacteria (phytoplankton) at/near the surface is a complex function of SST, PAR (photosynthesis band SW IR flux) and nutrient availability. When phytoplankton bloom they emit dimethyl sulfide which transfers into the air and decomposes to form sulfates i.e. cloud condensation nuclei (CCN). All other atmospheric factors being equal (lapse rate etc) this (at a minimum) increases the rate of low cloud formation.
In addition, the bloom is subsequently often predated by zooplankton and cyanobacteriophages (lierally viruses) so the sea surface is soon covered with mono- and multilayers of lysed cell membrances and their contents – lipids, isoprenes etc – some of which are volatile and also transfer to the air to be (biogenic) CCNs.
The organic layer itself increases surface albedo lowering SST. Some majoe blooms are of coccoliths (carbonate-containing cyanobacteria). This also increases surface albedo.
These blooms are commonplace and easily visualised with satellites – especially in the mature phase (= aformentioned organic layer). Their extents are often massive.
Several years ago there was a major review of about 9 mainstream GCMs. I can’t remember the leading author’s name (Japanese though). They were all shown to be very poor at low level oceanic cloud prediction, particularly in the equatorial zones and those other parts of the ocean where cyanobacterial primary productivity is known to be (albeit sporadically) high.
One might also note that roughly analogous effects occur over large forested areas. Rises in SST and PAR lead to spurts of leaf or needle growth, increased ET and increased emissions directly into the air of volatile isoprenes etc i.e. biogenic CCNs. This is why forest create clouds and rain. This particularly occurs where there is a high forest biomass and also essentially contiguous forest cover right up oceanic coasts for reason I don’t have space to elaborate here (e.g. good Russian paper last year mentioned in New Scientist).
In a nutshell I am implying that the global climate system is not purely physical. It incorporates very significant biogenic physical and physicochemical effects which have evolved over Gyr and Myr timecales.
GCMs are particularly poor at incorporating these complex biogenic effects – even e.g. where they have had ‘bolted on’ oceanic biogeochemical sub-models (the latter are still at a relatively low level of development climate -wise).
We should therefore not be at all surprised, with Roy Spencer, that the chain(s) of causation between clouds and temperature are not well understood.
This is why Roy’s ‘back to basics’ approach (of comparing satellite data very closely with GCM performance) is so sensible.
In a world culture and science milieu awash with late 20th century post-modernist arrogance we are still, as yet babes, both in the woods, and on the painted ocean.
Nic (15:14:26) :
“So (Nic) might suggest that since Europe is running at about 10 times the number of humans per square mile the CO2 concentration should be higher in Europe than the USA (before it globally diffuses).”
Thank you for your reply to my question. I was looking at the NASA map (on a previous article on this web site) but here is a link that shows how much each Nation emits of Carbon Dioxide. China is now #1, United States is #2 and the Entire European Union is #3.
The link: http://en.wikipedia.org/wiki/List_of_countries_by_carbon_dioxide_emissions
My basic question, if carbon dioxide is the primary driver for Global Warming then the United States temp for the last 100 years should be the one spiking, not the Global temp. Currently United States and China should have by far the biggest spikes of temp on any other spots on Earth if carbon dioxide is a primary warming driver.
Smokey (06:39:00) :
GISS always shows more rapid and greater warming than the actual recorded temperatures; their adjustments never show more cooling.
And if you get your info from realclimate, you will be misinformed.
These plots of randomly selected long term stations I made some time ago show otherwise:
http://img11.imageshack.us/img11/7440/gissrawtemps.jpg