Interesting Discussion on Models

In an article about evaluating models, there is an interesting discussion of climate as a modeled entity. Some of the things that cause the Earth to have more stability are discussed, but what caught my attention was how you can close the loop on some open ended things with rules of thumb. (Then the question becomes how those rules of thumb change over time and over changed inputs / conditions…)

So I started here:

then worked my way back up stream into the comments that caused it. Also of particular interest was a graph posted by Willis. one comment above (as I type… but things being nested subject to change over time)

I’m going to talk about the image first, then get back to the first link.

Albedo Corelation With Temperature

Willis points out the strong correlation of albedo with temperature in the tropics as more temperature causes more clouds.

This is because, in the all-important tropics where most of the solar energy enters the system, albedo goes up with the temperature. They are very highly correlated, as you can see below.

But what got my attention was the NEGATIVE correlation at the poles. Being ice, it just doesn’t matter what the temperature is at the poles. They just don’t absorb much no matter what. Yet it isn’t all an ice effect. The corelation is most negative over the LAND of North America, Asia, and Chile / Argentina. One wonders why.

Is it farming and ploughed ground when the weather warms? Is it trees turning from green to brown in fall? Is it ice melting in spring? Is it the freezing out of water vapor and the loss of clouds when it goes to “cold winter nights”?

What are the implications of the equatorial zones being strongly positive corelation and the temperate and polar areas being negative corelation?

My suspicion is that this reflects the water cycle as dominant in our climate and weather patterns. While the equatorial zones demonstrate the “negative feedback” of the thunderstorm thermostat (water as vapor), the temperate and polar zones show the positive feedback that gives us strong seasons (water as ice and snow). Albedo is the reflected sunlight fraction, so a negative albedo corelation would be a positive feedback to heating – as you get hotter you get less albedo and more sunshine absorbed ‘up north’. I note in passing that the American South is about neutral and California is in the positive corelation quasi-tropical mode, despite being dry, but perhaps it is that offshore water and the San Francisco summer fog effect.

In short, is it just showing that summers in the temperate to polar regions have fewer clouds and more nice sunny days, leading to added warming as the winter clouds fade; yet too much heat is not possible as the oceans create lots of cloud and rain as the water approaches the 84 F range and that also causes negative feedback (positive albedo corelation)? Do we have two thermostats at work? One making the tropics stable, the other making the polar regions prone to swings? (Or is it seasonal changes of insolation that drive the swings that cause the albedo shift…)

I can see an endless argument about feedbacks springing from that. When one is positive and the other negative, which wins? Does that change over time?

Back at the first link

CO2IsNotEvil said (bold mine):


“The problem with this analysis is that Psun * (1-a), the amount of solar energy available after albedo reflections, is itself a function of the temperature.”

Not as much as you think. Yes, the albedo in polar regions is larger than equatorial regions owing to ice and snow, but the decreasing albedo from melting ice and snow is quite small. It was larger coming out of the last ice age when there was a lot more of the surface covered in ice, but today, the average fraction of the planet covered by ice is pretty close to the minimum possible. Average polar temps are far below freezing and no amount of GHG action will ever be enough to melt it all and prevent it from returning in the winter. About the only thing that will cause this is when the Sun enters its red giant phase.

Considering that 2/3 of the planet is covered by clouds, which have about the same reflectivity as ice, 2/3 of all future melted ice has no affect on the net albedo.
Polar regions receive less insolation to begin with and when you calculate the increase in the incident power from melting all ice and snow on the planet and distribute that power across the entire planet, its only a few W/m^2 and less than what’s required to achieve the global emissions increase (temperature increase) they claim arises by doubling CO2.

The sensitivity expressed as a change in temperature per change in input power, dTs(t)/dPi(t) is already a function of temperature and that function of temperature is independent of the albedo. None the less, since (1-a) is linear to e, whatever effect albedo has can be rolled into an equivalent value of e, both of which can be expressed as functions of the fraction of the planet covered by clouds. Note that the sensitivity expressed as a change in surface emissions per change in input power is constant, where

dPs(t)/dPi(t) = 1/e

Yes, e is a higher order function of temperature, but when we measure it over the last couple of decades, it’s remarkably constant coming in at about 0.6, where dPs(t)/dPi(t) is about 1.6 W/m^2 of Ps per W/m^2 of Pi. It’s even relatively constant from the poles to the equator where e increases only slightly as the average temperature transitions through freezing.

It is an interesting approach. Looking at the way one place loses albedo but it causes a net gain in other places so the planet as a whole has far less change of energy budget.

But what this points out to me is just how critical it is to think in terms of clouds and ice. Those two are the ones driving this albedo show. If we can’t determine what actually changes them over time, we can’t explain the balance of negative and positive feedbacks from them and how those change over time.

Any model that claims to represent the Earth simply MUST start with solar input, how it varies, and how that varies the ice and clouds of the world. Otherwise they have the wrong albedo numbers, the wrong positive vs negative albedo feedbacks, and the wrong results.

I need to revisit the GCM source codes I’ve got to confirm (or deny) but I think they treated the sun as a constant (only variation being at grid cells via tilt changes) and treated albedo as a derived result of temperature changes caused by gas “forcing”. I don’t remember them using different albedo correlations by latitude or grid cell. (Perhaps I just missed it… some of that code is painfully obscure and dense). Basically, I think it was mostly a case of “making snow” and then adjusting albedo if that happened. (But I need to check it…)

It seems to me that there is plenty of room to “close the loop” on some of those processes; like accounting for 2/3 of ice melt as cloud increases) and reduce some of the positive feedback built into the present models.

Then, a muse per Ice Age Glacials:

IFF the negative corelation (positive feedback) is driven by the ice cycle, the implication is that a modest expansion of the area covered by ice, or having more of it persist seasonally, could tip our climate balance from the present negative feedback stable warm phase into a positive feedback cold phase. Enough of that, and if the positive feedback is to the cold direction, and we would have the oscillator that causes Ice Age Glacials to lock up for 100,000 years waiting for enough solar heating at the polar regions to get us back into a stable warm phase.

Basically, if axial tilt change shifts those positive feedback zones further toward the equator, at some point they win the positive / negative feedback balance wars and we tip into unstable positive feedback net. If due to ice, we lock up in ice. That would be bad…

Perhaps we need more monitoring of the temperature / albedo corelation values… especially as the snows are coming earlier to Europe and Australia and extending further south in North America… Ice Age Now has a lot of such reports listed at the moment:

Subscribe to feed

About E.M.Smith

A technical managerial sort interested in things from Stonehenge to computer science. My present "hot buttons' are the mythology of Climate Change and ancient metrology; but things change...
This entry was posted in GCM and tagged , , , . Bookmark the permalink.

30 Responses to Interesting Discussion on Models

  1. oldbrew says:

    ‘albedo goes up with the temperature. They are very highly correlated’

    Albedo can’t cause itself.

  2. Ron Clutz says:

    Thanks for this post and insights. There is interesting related research and data at the Institute for Atmospheric and Climate Science of ETH Zurich, led by Martin Wild, senior scientist specializing in global dimming and brightening. In contrast to the satellite imagery, ETH is compiling changes in sunlight incident on surface sensors. An overview with links is here:
    Dr. Wild’s money graph is this one:

  3. Ron Clutz says:

    Sorry, forgot to add the legend for the figure:
    Figure 2. Changes in surface solar radiation observed in regions with good station coverage during three periods.(left column) The 1950s–1980s show predominant declines (“dimming”), (middle column) the 1980s–2000 indicate partial recoveries (“brightening”) at many locations, except India, and (right column) recent developments after 2000 show mixed tendencies. Numbers denote typical literature estimates for the specified region and period in W m–2 per decade. Based on various sources as referenced in Wild (2009).

  4. oldbrew says:

    Ron C’s graphic looks more to do with industrial pollution than nature.

  5. cdquarles says:

    Correct, EM. That said, non-solar input is not zero and variable. Extra-solar input is also not zero and variable. So, we must also consider the huge loss of information when using gross statistics, such as the average of averages that is global mean surface temperature.

    I ask this question: “Are we asking the wrong question(s) about the wrong system(s)?”, since it sure seems so to me.

  6. cdquarles says:

    @oldbrew, that’s not necessarily so. “Nature” “pollutes” too, given specific definitions of “nature” and “pollution”.

  7. p.g.sharrow says:

    When I consider the life of the great Ice Mountains birth and death, Dust comes to mind as a controlling factor. During the build, heavy snows caused by warm larger oceans and cool atmosphere conspire to cause build up of snow fields. New fields are white, reflective and as they grow oceans cool and decrease in size. The increased snow fields create their own weather sinks that suck in more storm weather as well as dust from exposed former sea beds as well as from drought afflicted lands caused by cooling reduced ocean evaporation. As the snow fields settle into ice fields the dust in the snow becomes surface dirt that accumulates to become muskage that becomes vast grasslands atop the the old ice fields that are slowly melting. This change of snow cover to dirt cover would greatly change the albedo of the Ice Mountains in a fairly rapid period of time and shift the “local” weather from cooling to warming.
    At present we have warm expanded oceans and cooling atmosphere. So more clouds, more cooling of the atmosphere. The oceans are vast, so cool slowly as they evaporate their water and energy. So how long would it take to evaporate 400feet of ocean? Water to be piled up into Ice Mountains and cool the oceans enough that evaporation is reduced enough to dry the atmosphere and land? As the land and oceans dry, Dust storms increase to shift the balance of snow/ice accumulation. Cool oceans and warm atmosphere would result in reduced clouds and more heating.
    I keep seeing well watered dirt/soil grass lands covering melting “permafrost” lands as the end of an Iceage as the filling cool oceans begin to warm to restart the cycle…pg

  8. Ron Clutz says:

    oldbrew, how much dimming due to aerosols and how much to cloud cover is yet to be determined, especially since aerosols can induce clouds. Some researchers have found the clouds to be dominant at least in the regions studied:
    “Analysis of the Angstrom-Prescott relationship between normalized values of global radiation and sunshine duration measured during the last 50 years made at five sites with a wide range of climate and aerosol emissions showed few significant differences in atmospheric transmissivity under clear or cloud-covered skies between years when global dimming occurred and years when global brightening was measured, nor in most cases were there any significant changes in the parameters or in their relationships to annual rates of fossil fuel combustion in the surrounding 1° cells. It is concluded that at the sites studied changes in cloud cover rather than anthropogenic aerosols emissions played the major role in determining solar dimming and brightening during the last half century and that there are reasons to suppose that these findings may have wider relevance.”
    The cause of solar dimming and brightening at the Earth’s surface during the last half century: Evidence from measurements of sunshine duration by Gerald Stanhill et al.

  9. Pingback: Interesting Discussion on Models – Climate Collections

  10. Larry Ledwick says:

    So how long would it take to evaporate 400feet of ocean?

    Look to pan evaporation rates for a ball park guess.

    Click to access TR34.pdf

    i suspect you could use the coastal pan evaporation rates as a proxy for the nearby ocean evaporation rate (not counting accelerated evaporation by wave action, spray and droplet evaporation)

    Looks like California pan evaporation rates run in the 70-100 inches per year range.
    Hawaii has numbers in the 90 inch per year range (central Pacific proxy)
    Mississippi has numbers in the low 60 inch range, Texas gulf coast (Point Comfort Tx) high 70 inch range.
    North Carolina 50-60 inches (southern Atlantic Coast)
    New Jersey and New York 30+ inches per year.

    Looks like an over all average of 50 inches a year would not be too far out of bounds in the temperate zone.

    That would put 400 ft sea level elevation loss in the one century range and that is not counting recharge cut off as fresh water rivers cut off due to freezing.

    As a ball park guess the oceans could lose hundreds of feet in sea level in just a century or two once the major rivers like St Lawrence Sea way, and the Mississippi cut off as the central northern US freezes up.

  11. omanuel says:

    CERN nuclear physicists recently admitted that everything they thought they knew is probably wrong:

  12. E.M.Smith says:


    You already posted that on the Gravity Light thread. Please don’t start behaving like a spammer, posting the same off topic link on multiple threads…


  13. Lionell Griffith says:

    “CERN nuclear physicists recently admitted that everything they thought they knew is probably wrong”

    I have long thought the theoretical physicists act as if their equations create reality rather than their equations are merely an attempt to describe a portion of reality. They are in fact nothing but models of the real thing and not the thing itself. In that sense, all models are wrong. Some are good enough to use as engineering principles to make things that work. Many are not.

    As for explaining why there is something rather than nothing. If there were nothing, the question wouldn’t even arise. it is an attempt to derive existence from non-existence. A totally empty exercise. Things that exist must necessarily exist. Why is not a valid question to ask. What is, is! What it is and what can we do with it to make our lives better and fuller, is the right question to ask and answer. Anything else is meaningless noise no matter how important it is made to sound.

  14. Larry Ledwick says:

    Some are good enough to use as engineering principles to make things that work. Many are not.

    Which is why most engineering considers a minimum factor of 4 safety factor in calculations (admitting implicitly that they cannot absolutely predict the loading, nor can they absolutely guarantee the quality, strength and fastening methods of the final construction.

    I life safety applications like ropes used by fire departments the NFPA insists on a 10x safety factor between expected working load and ultimate breaking strength.

    Too bad weather models do not use similar safety factors (error estimations) but falsely assert their output is precise and a valid estimate of reality.

  15. pouncer says:

    Larry Ledwick says: “most engineering considers a …safety factor”

    True. In my limited experience. Though with very interesting exceptions.

    One doesn’t engineer the load of a nuclear power reactor with a fudge factor making it, let’s say, 4 times as “prompt super-critical”, as it needs to be in order to get up to desired temperature, nor 2 times smaller than theoretical minimum criticality just to be sure it doesn’t melt down. (One might, however, design and install two identical graphite SCRAM rods, just in case the first one failed to drop.)

    Similarly one doesn’t design a rocket to boost the payload, plus 4 times as much reaction mass (and a redundant engine) as we MIGHT, theoretically, need to reach orbit, all built into the third stage, just in case the main engine failed and we had to fuel the reserve engine for much longer than expected.

    The machines treating cancer tumors with radiation don’t flood an area AROUND the tumor several times bigger than the tumor itself with some factor of higher-intensity radiation than strictly necessary, just in case.

    On the other hand the “mythical man month” principle in software projects suggests we allow 80% of the project duration for the first 25% of the project, and ANOTHER 80% of the time for the next 50% of the project, and …

  16. Ron Clutz says:

    The short version of pouncer’s principle: A poorly planned development project can take three times the estimated time and expense. While a well planned project only costs twice as much.

  17. Lionell Griffith says:

    Ron: The short version of pouncer’s principle

    Typical cause:
    1. Marketing types determine when the product must be delivered justified by “I promised”.
    2. Software engineering designs and works out a careful schedule based upon current resources and provides management with a date of delivery that is twice the time to the marketing promised date.
    3. Management cuts resources in half and sets the delivery date to what was promised by marketing.
    4. When it is clear that the due date won’t be met, management insists that testing be reduced to nearly nothing from the already thin1/3 of the original careful plan.
    5. Customer is unhappy with the product delivered on time because it repeatedly crashed and lacked critical functionality that marketing never told the software engineering department about.
    6. It now takes twice the time already spent to include the undisclosed functionality because it required a complete reorganization of the software and a complete test to make sure it works. The programmers had to do the testing. The test group was laid off in step 4 because they were no longer needed.
    7. The customer is still unhappy because of the extreme delay and low quality final delivery and refused to pay full price.
    8. The software engineer was blamed for the delay and was “allowed” to resign his position. That was OK because he had found another job at higher pay.
    9. Both the marketing person who promised and the software manager were promoted to higher paying positions. Their combined increase was the same as the software engineer’s salary.

  18. Another Ian says:

    Re Lionell Griffith says:
    15 December 2017 at 7:14 pm

    Lionell Sounds pretty much like this performance on our health department payroll

    Click to access 338.pdf

  19. gallopingcamel says:

    “Basically, if axial tilt change shifts those positive feedback zones further toward the equator, at some point they win the positive / negative feedback balance wars and we tip into unstable positive feedback net. If due to ice, we lock up in ice. That would be bad…”

    The last seven glacial cycles lasted about 850,000 years and counting. The temperature plot reminds me of a relaxation oscillation that you see when there is excess positive feedback causing the temperature to remain in a cold state for ~110,000 years and then rapidly transition into a warm period lasting ~15,000 years oscillate and then back again.

    In an electrical relaxation oscillator the time spent in each stable state is determined by components that store energy. In the climate case, water and ice can store massive amounts of energy so I suspect thermal energy storage determines the time constants. This idea explains why the inter-glacials are much shorter than the glacial periods. During the inter-glacials water is the main energy storage medium but during glacial periods continental ice stores far more energy than water can.

    Technically we have been in an Ice Age for the last one million years. Will that continue or will we bounce back to conditions that existed 60 million years ago when polar oceans were 15 degrees Centigrade warmer than today? I suspect that will be determined by Milankovitch cycles, Earth’s magnetic field and cosmic radiation.

  20. beththeserf says:

    ‘…the inter-glacials are much shorter than the glacial periods.’

    Serf musing : Say, why would a little punsey warming, circa 1998,
    be a worry when the Ice-Man-cometh and Cli-Sci knoweth – and
    must also know that efficient energy is the road to survival for
    a population of 7 billion humans – living in high rise…?

  21. richard verney says:

    There is a NASA or NOAA study in its preliminary stage, that suggests that as ice melts in the Arctic, there are more clouds, which means that melting ice has less impact on albedo than initially thought.

    There is a link on one of the NASA or NOAA pages to a summary of the research being conducted by one of their scientists with some preliminary comments.

  22. richard verney says:

    I am not sure that this is the article that I was thinking of, or perhaps it has undergone some change, but a similar point is made.

    Although sea ice and snow cover had noticeably declined in the Arctic from 2000 to 2004, there had been no detectable change in the albedo measured at the top of the atmosphere: the proportion of light the Arctic reflected hadn’t changed. In other words, the ice albedo feedback that most climate models predict will ultimately amplify global warming apparently hadn’t yet kicked in.


  23. richard verney says:

    Whoops, forgot to include the money point:

    Kato quickly understood why: not only is the Arctic’s average cloud fraction on summer days large enough—on average 0.8, or 80 percent—to mask sea ice changes, but an increase in cloudiness between 2000 and 2004 further hid any impact that sea ice and snow losses might have had on the Arctic’s ability to reflect incoming light.

  24. Larry Ledwick says:

    Which makes sense if the ice melting is actually due to either warm water excursions into the arctic basin (polar undersea volcanic activity or warm water intrusions from the northern gulf stream) or wind driven expulsion of ice into the north Atlantic, both exposing more open water (more open water equals more water fog/clouds)

  25. poitsplace says:

    You (and p.g.sharrow) touched on some points that should be blatantly obvious in the proxies. The earth has some specific temperature ranges that are very stable, in its current configuration at least. With the oceans separated by the tiny sliver of land between the americas, the warm end is the interglacials. Above a threshold and the cloud albedo feedback expands so much the earth simply won’t warm any more. Below a certain threshold (world covered in ice) and the dust albedo feedback dirties the snow, preventing it from going farther. And anything between is a hellish see-saw that in all honestly is the apocalyptic hell pushed by climate nutters…but one CO2 would make us slightly safer from.

    An interesting side note is that the ice core data…both snowfall rates AND CO2 work out to be temperature proxies. This is (in a round about way) explicitly stated in the “CO2 amplificaiton” theory used to justify high sensitivity claims, that as temperature increases, CO2 degassing from the oceans (hence the lag) increases warming and causes even more warming. The problem with using this as justification for CO2 sensitivity is that it cannot be falsified. Since CO2 is essentially driven exclusively by temperature in those proxies. Assume CO2 sensitivity is high, and the temperature changed the way it did and sure, you get a match. But it would also work if you assume CO2 sensitivity is NEGATIVE and temperatures fluctuated the way observed in the proxies, because again, CO2 is just a lagged proxy. The only thing proxies seem to show us is that sensitivity of the earth’s climate to whatever things are driving it…gets weak at interglacial temperatures and at glacial maximum temperatures, because the temperature fluctuations are a fraction of the size in those ranges.

    Getting back to the CO2, it is outright insanity to claim that the lagged CO2 changes are “driving” such rapid changes. Moisture driven albedo feedbacks seem to be a MUCH more likely candidates since it can in fact respond over a period of years. And it’s not JUST plain and simple snow.

    Tundra to forest albedo changes would be weakly positive
    Desert to grassland would be weakly positive
    Grassland to forest would be weakly positive
    Exposed continental shelf to ocean would be positive
    Ice/snow on the continental shelf and ocean inundation feedback would be strongly positive in multiple ways (and self reinforcing)
    And BTW it is thought that sea ice extended to within 50 degrees of the equator during the glacial maximum, meaning that sea ice albedo at that time was far stronger even in winter than it is now in the summer.

    For now most of these albedo feedbacks are only possible/meaningful over a small portion of the planet and where there isn’t enough light to matter. One of the only positive ones left in a place it can have any impact would be for the great deserts of the earth to shift to grassland/forest, which happens during interglacial optimums. But I somehow doubt anyone would consider it a tragic loss for the great deserts of Africa, Asia, and Australia to vanish (BTW, the American southwest is one of the only ones that expands somewhat during the optimum)

    Sorry to ramble but thought it was relevant.

  26. A C Osborn says:

    EM, I have seen and think that Northern Hemisphere Albedo actually has little effect due to the inclination of the Solar Radiation during period when Ice is at it’s lowest.
    The sun striking open Ocean at that angle has little more chance of being absorbed than if it was Ice.
    Surely Solar Energy striking the Surface and being reflected, ie going back through a much thicker Atmosphere would leave more energy behind than if it strikes the tops of clouds and is reflected from them?

  27. p.g.sharrow says:

    The great Ice Mountains were not caused by colder temperatures. They were created by heaver wet snows in normal snow areas. It takes heavy wet warm snow to build up into glacial ice. Colder temperatures result in dry light snows. Warm oceans evaporate more water. Warm air carries more moisture. The amount of frozen water deposited is the thing needed to exceed the later summer melt.
    i once lived in Alaska’s Glacier area of the Chugach Mountains and Prince William Sound. “Warm” Japanese current pumps huge amounts of cooled wet air against the cold Alaskan interior and dumps huge amounts of wet snow on these mountains, 100ft and more in areas that rarely see the cold temperatures of the interior. The glaciers advance and retreat due to snow fall amounts, not temperature changes. Cold causes reduced snow falls because cold air carries less moisture to freeze out into snow. The permanent ice cover results in cooler summer air temperatures and later colder winter temperatures. An effect, not a cause of the Ice Mountain’s creation. Warm oceans cause heaver snows in normal snow areas cause glacial ice build up. Dry cold from cold ocean water results in reduction of the snow falls and reduction in the extent of glacial cover….pg .

  28. p.g.sharrow says:

    It is the sunlight penetration into tropical oceans that add energy into the climate system.
    The water intakes of our ship would read up to 80F. 30 feet down south of the Philippines, while 50F at the same depth off of central Japan. I would doubt there is any positive input into the oceans over 60 degrees of latitude. Most of the input takes place below 30 degrees.
    The oceans collect and contain far more energy then is possible in the atmosphere above and largely escapes as water vapor is created. The Arctic/Antarctic ocean areas are energy net loss all the time.. Ice forms from the top down and melts from the bottom up. Ice is an insulator and greatly reduces evaporation as well…pg.

  29. Macha says:

    In terms of heat capacity, the air would need to be about 1000C to raise ocean temperature by 1C. Or thereabouts….

Comments are closed.