Momentary Lapse of Reason

Read the following with caution but this is potentially the most important story the Air Vent has carried.  A paper for open review by Makarieva A.M., Gorshkov V.G., Sheil D., Nobre A.D., Li B.-L. in open discussion here, has revealed what appears to be a gaping hole in climate models.  In previous discussions here it has been stated that the pressure loss by condensation simply didn’t exist in models, but I’ve never checked it myself — until this morning.  Nick Stokes pointed us to the relevant page of CCSM3.  The model is a parametrized version of the physics which leaves out the key factor of pressure drop caused when water vapor condenses.

Now I’m not an expert in models, but I can read math.  The model page linked above indicates the thought process that went into the parametrization and from that, pressure loss from the energy of various components of condensing gas is simply not included.  During the parametrization process, basic physical properties are identified such that you get the right answers without extra calculation.  Skip a step in the simplification and you get the wrong parameter.  What it means is that you can get your mass and energy balance without recalculating pressure.

This is not a SMALL change.  But it is more than one model, what I understand according to Dr.Makarieva it apparently is accepted consensus science!  If true, this physically represents the need for a major change to climate models in general.

As an engineer, you imagine using the basic flow equations, the basic energy content equations, the basic mass equations to calculate pressure and flow which is then linearized for finite element analysis.  The models do all that but are not that detailed in the end.  They apparently skip a  step, a very big one – air pressure change due to condensation.

I know it seems preposterous, but in Meteorology, condensation is still considered by some to increase pressure.  From an anonymous reviewer of the hurricane paper by Anastassia Makarieva:

The conventional wisdom is that in the real atmosphere, the condensation heat causes
expansion of the air, hence a lowering of the specific density, hence a lesser weight of
the air column, and this (together with surface heat flux) causes the pressure drop in
the lower layers. From the viewpoint of dynamic meteorology, condensation is not an
“anti-explosion” or “black hole” as claimed in section 4, but rather a kind of explosion.

Nope, they did write it….. and multiple times. Looks like the cow scoop is on the wrong end of the steam engine.  Of course the physics of the situation are more intellectually subtle than described by a steam engine, but NOT much.  The reason that it wasn’t stated this way before was because I’ve never read the math in a climate model so how would I know what is in them.

Until this morning.

For a moment, lets consider the basic effects we might see from the massive condensation based pressure drop in meteorological and climate models.  Higher winds and stronger updrafts would lead to more lower troposphere cooling, hurricanes suddenly make thermodynamic sense, overshoot of model temperatures on average should be affected, negative feedback could potentially be explained.

So I would like this to be considered a formal invite for anyone of any background to show us just where in climate model above, this loss of volume from condensation is considered. As a non-expert in models, it seems more than surprising that the volume loss isn’t included so maybe Nick Stokes said it best:

It’s complicated in Chap 4 of the CAM3 text, because they use potential temperature (a good idea), and it’s mixed up with the discretisation. They have also done a spectral transformation. Chap 3 may be easier.
In short, I haven’t located yet where the condensation volume deficit is passed into the N-S mass equation. But I’m sure it is there.

If this turns out to be correct, as it appears to be, climate science has some splaining to do.  You want to see some backtracking on uncertainty, uh oh!

187 thoughts on “Momentary Lapse of Reason

  1. One of the best things about science is that it’s flexible. Most frequently, there’s no “absolute truth” because of the complexity of the things science attempts to explain. The scientific process demands scrutiny, and allows for flaws to be detected and corrected. Religion, on the other hand, or politics do not. These alternative forms of reason demand adherence to a “party line” or “god’s will” – thus, not allowing for problems to be identified, or corrections to be made.

    Anyway, nice post, and very very good eye. There’s definitely some ‘splaining to do here, as there seems to be a big, open hole in the middle of the model!

  2. My pea sized brain won’t handle all of the math.

    But will it matter?

    The Team will dismissively wave it off and the MSM will ignore the issue completely.

  3. Wow, evisceration of GCM’s , right before our very eyes.

    Should there be a ‘viewer discretion warning’ ?

    Is ‘parameterization’ just a seven syllable word for ‘crib sheet’?
    (and a crib sheet for the wrong exam at that…)

    Simply Stunning…
    I happened to check the AirVent within minutes of the Climategate posting, and snagged the zip file from the .ru address.
    This seems like an equivalent moment in time.

    Another fine piece of work, hosted by JeffID.

    Cheifio is onto another fine stinker also, regarding the ASOS algorithmic ’round-up-to-whole-degree’ specification.

    Explosion indeed!
    RR

  4. Sorry, but honestly, yawn, yawn, old news. Didn’t you know !!!!

    You guys really have some catching up to do, don’t you.

    The real question is, are clouds a natural shut off valve to convection.?
    Does the latent heat released by condensation reduce the temp gradient back to the earth’s surface sufficiently to shut off convection,
    or does the pressure decrease from condensation and heating (by latent heat release) more than counter act this. ?

    How much of the released latent heat of condensation is reabsorbed by clouds evapourating. ?

    It ain’t all radiated you know, definately not back down, but up some yes, actually more than down, by chance of collision alone.
    The rest, where does it go, around and around…

  5. #8 Derek,

    Can you point to a source which demonstrated the condensation pressure was not included in models? I doubt very much that you can, Anastassia has been quite explicit about her findings on this and has had quite a bit of pushback from the mainstream in meteorology.

    Your first and second question are answered in the paper linked above.

  6. You gotta look at those equations, which are just a small fraction of the code and you wonder about the V&V. The scope for error is enormous and this isn’t engineering code. And then you realise that, as suspected, all climate models plagiarise (is that the buzz word?) the same bit of code. So if one model has an error, most likely they’ll all have the error.

  7. I always considered “the models” meaningless, because no one could ever explain what their value was. All I’ve ever heard is the “it doesn’t mean they are worthless” jive, whenever they are questioned.

    Slow News Day? 😉

    Andrew

  8. If this is really missing we’re looking at a recall of results across the board. All the models will need rework. Unless of course someone can point out where it has been considered.

  9. No way that this is a minor factor. This is what drives the main heat pump from surface to sky, the paper shows that the pump is underestimated pretty dramatically.

  10. There is an enormous time-lag between:

    a.) Discovering a flaw in mainstream views, and
    b.) Getting it considered by mainstream scientists.

    Continuously accumulated flaws will eventually sink even the best-funded ship, as has started to happen to the “Ship of Global Warming.” Other ships in the fleet are being dragged down too.

    See, for example, public comments on the Physics World news story about support that the American Physical Society (APS) tried to give to the sinking “Ship of Global Warming”: http://physicsworld.com/cws/article/news/44024

    Three basic flaws discovered in mainstream views over the past 40-years are being publicized today by Physics World:

    1976: All primordial He was associated with “strange” Xe at the birth of the Solar System.
    1983: Mass fractionation enriches lightweight isotopes at the solar surface and in the solar wind.
    2000: Neutron repulsion generates excess mass in every nucleus with 2 or more neutrons.

    The sinking of this fleet of ships confirms the spiritual foundation of science:

    “Truth is victorious, never untruth” [Mundaka Upanishad 3.1.6; Qur’an 17.85]

  11. It is my understanding that the models are at too coarse a scale to handle actual convection and the heat pump of thunderstorms (which are grossly underestimated in terms of heat transfer). Thus the effect of condensation etc. are folded into the “parameterization” which is a pretty untestable and unphysical relationship. Thus I challenge Nick Stokes to find where it is in the models, because they are too coarse scale to include them.

  12. Craig,

    Perhaps you have found it, this is what it says in the changes to v3.

    Treatment of cloud condensed water using a prognostic treatment (section 4.5): The original formulation is introduced in Rasch and Kristjánsson [144]. Revisions to the parameterization to deal more realistically with the treatment of the condensation and evaporation under forcing by large scale processes and changing cloud fraction are described in Zhang et al. [200].The parameterization has two components: 1) a macroscale component that describes the exchange of water substance between the condensate and the vapor phase and the associated temperature change arising from that phase change [200]; and 2) a bulk microphysical component that controls the conversion from condensate to precipitate [144].

  13. #17 Craig,
    It’s true that models use a grid scale which is too coarse for thunderstorms etc (IIRC, about 100km) and a bit too coarse to do a good job with hurricanes. The part where Jeff has started out (at my suggestion) is what they call a cloud-scale model. I think the idea is that you can model a generic cloud, and then treat an assemblage of clouds at the normal model scale.

    The Navier-Stokes equation are solved at the 100-km scale. So we’d have to figure out how volume change information is transferred from one scale to the other.

  14. People might like to look at the background to this GCM. Wiki is a good place to start, and their project page is here. It’s intended to be a fairly open environment for embedding various experiments in. That’s why it has readily available and extensive documentation.

  15. JeffID Don’t know if you read http://climateaudit.org/2007/02/11/exponential-growth-in-physical-systems/ and other similar, but the quote I like is from Dr. Browning:

    These analytical and numerical results raise a number of troubling issues. If current global atmospheric models continue to use the hydrostatic equations and increase their resolution while reducing their dissipation accordingly, the unbounded growth will start to appear. On the other hand, if non-hydrostatic models are used at these resolutions, the growth will be bounded, but extremely fast with the solution deviating very quickly from reality due to any errors in the initial data or numerical method.

    References:

    Browning, G. and H.-O. Kreiss: Numerical problems connected with weather prediction. Progress and Supercomputing in Computational Fluid Dynamics, Birkhauser.

    Lu, C., W. Hall, and S. Koch: High-resolution numerical simulation of gravity wave-induced turbulence in association with an upper-level jet system. American Meteorological Society 2006 Annual Meeting, 12th Conference on Aviation Range and Aerospace Meteorology.

  16. I think the logic of the CAM3 model doc goes like this. As I mentioned above, they work on a cloud scale in, say, sec 4.3 which we have been looking at, and they come up with a moisture ratio q_bar which is the effect on the “large-scale” – ie the scale of the 100 km or so grid on which the N-S equations are solved. Eq 4.75 is the “budget” equation on this scale, where they add everything together.

    They refer to a script-R as the “convective-scale” liquid sink in the budget equation. This all gets aid again in 4.79, and then they go into discretisation, which adds messiness. q_c, which is the cloud-level moisture ratio, is derived. It’s a perturbation of q_bar as in 4.87. It is derived from humidity etc in eq 4.90.

    This all then gets passed up to the large scale level (N-S equations), via the conservation equation 3.365. This is coupled with another equation for π (total air mass) in 3.365. These are analogous to AM’s 32 and 33, and 3.364 will be coupled to the momentum equation, which relates it to pressure. In fact here they are talking about a hydrostatic atmosphere, so I might be in the wrong place, but they are the right kind of equations, and will actually apply dynamically as well.

    The momentum equations are written in the style these people like in 3.367-8. The link goes:
    q variation is coupled to π in 3.364/5
    π variation is coupled to Φ in 3.363
    Φ gradient then goes into the momentum equation at 3.367-8. In fact, if you think of density ρ as constant (reasonably true on a cloud scale) then the momentum eq has
    ∂(Φ+P/ρ)/∂λ
    In other words the volume variation due to condensation, which went into Φ, is implicitly added in to P at that stage.

  17. Nick,

    If you read the paper you linked they specifically use only updrafts from temperature as the limiting energy for the height of moisture. They ignore the additional power of condensing water in the parametrization. While they have a parameter as you pointed out, they don’t have the right one.

  18. Nick,

    3.364 and 3.365 are set equal to zero. That’s pretty telling.

    Look at how it’s written:

    “Similarly, the mass conservation law for tracer species (or water vapor) can be written as”

    There is no mass conservation ‘law for water vapor’ it can condense!!

  19. Jeff,
    You’re a bit over keen to find a smoking gun here.

    Each term has a time derivative – the rate of change of mass. That includes condensation, or anything else that is removing mass from the system.

  20. Nick,

    Correct me if I’m wrong. In 3.365 there are two equal terms the first term represents the change in mass in a volume over time, the second term describes a constant mass at a velocity vector moving in and out of the volume. del = d/dx,d/dy,d/dz – d’s being partial derivatives. Therefore pi and q are constant values with velocity integrated to determine the change in the volume’s mass correct?

    Because the second term is of static mass density, there is no room for evaporation right?

  21. Interesting article Jeff.
    Another problem with the models is that CO2 is not distributed evenly throughout the atmosphere.
    There are clear pools of concentrations and when you have to combine them with the temperature data…or feed the models….well, I wish the modelers good luck.

    And there is yet another aspect.
    No model takes into account the existence of inversions in the atmosphere.

    Our atmosphere is not warmed by the sun.

    The sun warms the earth’s surface and the earth surface heats up the air. Depending on the color and the properties of the underground solar power is reflected and absorbed. Rock, deserts, urbanized area’s, industrial area’s, scrublands, heat up very quickly and warm the air much more efficiently than wetlands, forests, lakes, or oceans.

    Depending on a large numbers of factors that change every our of the day and depend on wind direction, landscapes, cities, the presence of cooler area’s like lakes. the source of the airflow (maritime, polar or land mass and the vertical temperature development of the atmosphere, vertical air currents, updrafts develop (or not).

    The best conditions for updrafts exist when the air is unstable and there is no inversion. Under those circumstances only small differences in temperature cause the (warmed) air to rise.

    Sometimes these updrafts are so powerful and the amount of air leaving the surface so enormous that a sudden drop in pressure occurs.
    Everybody has experienced these situations, for example you are sitting on a terrace enjoying the sun and suddenly a strong wind blows the entire terrace up side down and
    sun screens get air born. Some thermals are so strong that when you fly into them with a glider your make upward speeds of 15 to 25 ft per second.

    Up drafts or thermals are also the reason you never see hot air balloons during the day time. They need quiet air so they fly early in the morning or late in the evening.
    If they would fly during the day time a thermal could deform the balloon and press the hot air out of it causing it to crash.

    The hotter the sun, the hotter the surface, the more air is heated and transported up in the atmosphere.

    This is a continuous process but it is very dynamic although we can say there is a structure.

    Since gliders carry a GPS recorder (for competitions) thermal databases have been developed that depending on the wind strength and direction, the date and hour of the day provide information of the location and strength (upwards speed) of the thermal.

    The rising warm air is cooled down by about 2 degree Celsius for every 300 ft of altitude rise.

    Because the air cools the amount of water vapor the air is able to contain drops and at condensation level a cloud forms.
    These clouds mark the updrafts (cumulus) and the hight of these clus is determined by an inversion.

    If the inversion is burned out of the atmosphere (is warmed up by the rising air that accumulates under the inversion the cumulus continues it’s vertical development and turns into a cumulonimbus cloud. This only happens when on days when we have an unstable atmosphere.
    When the inversion perssits, the cloud spreads out like a pan cake blocking the sun and the updraft collapses.
    Only when the cloud is dissolved or transported by the wind, the process starts all over again.

    I’ve always wondered how modelers would approach these dynamics and translate them into their models.
    I bet that there are modelers who not even know this process is happening.

    The pressure drop caused by a phase change of water vapor is only another little aspect that modelers have missed.

    There are many more.

    Glider pilot’s know everything about these processes in our atmosphere. They can read the landscape and the clouds and fly their planes over distances of hundreds of miles on a single day without the use of an engine.

  22. Jeff,
    Yes, indeed that is true, and my earlier statement was a little hasty, altho’ in fact what they have written is an absolutely standard conservation equation.

    I think the key is sec 3.3.6. What they do is solve during a time step is if there were no condensation, and then adjust at the end (20 mins) for the change in mass that would have occurred due to condensation and other processes. This is standard for equations that are not tightly coupled – sometimes called operator splitting.

    I have to say that it’s still not clear to me where condensation comes in in 3.3.6. However, I remain sure that they haven’t just forgotten about it. This stuff has been around for thirty years, reviewed by thousands.

  23. Sorry for the mistake but the cooling down of rising air currents is about 1 degree Celsius for every 300 ft (100 meters) of altitude.
    (It’s tim to standardize altitude to the metrical system and temperature to degree Celsius)

    Because the sports of gliding was developed in Germany, all gliders in Europe have their indicators like speed, altitude, sink or climb rate according to the metrical system.
    Speed km/h altitude in meters, climb or sink rate in meters per second.
    Powered flight however, thanks to the Wright Brothers speed is in knots, altitude in ft and sink/climb speed in ft per second/minute

    Always switching when you fly both gliders and powered planes. It’s simply stupid.

  24. Jeff,
    Here is a key sentence in 3.3.6:
    “We are concerned about 3 time levels: qn is input to physics, qn is output from physics, qn+1 is
    the adjusted value for dynamics.”

  25. 34 i>”No model takes into account the existence of inversions in the atmosphere.”
    Not really true. There’s no need for special treatment of it, but it can certainly arise in the solutions.

    Here is an abstract from your old chum Wm Connolley using GCM modelling of Antarctic inversions.

  26. Nick,

    I’ve been reading your links for the entire time. From 3.3.6 there is this wording, it’s kinda gorgeous.

    “However, the surface pressure remains fixed throughout the physics updates, and since there is an explicit relationship between the surface pressure and the air mass within each layer, the total air mass must remain fixed as well. ”

    All they update is the q in 3.417 with no adjustment for the pressure created from the condensation. In fact they hold it to zero change during the update to maintain conservation of air mass (total gas mass and pressure) while changing q. By doing that in a condensation region, they’ve lost their conservation of mass in the form of condensate and increased the total energy per volume.

    from M10

    The value of ps (Eq. 19), air pressure at the surface, appears as the constant of
    integration after Eq. (19) is integrated over z. It is equal to the weight of air molecules
    in the atmospheric column. It is important to bear in mind that ps does not depend on
    temperature, but only on the amount of gas molecules in the column. It follows from this
    observation that any reduction of gas content in the column reduces surface pressure.

    It’s all complicated but the difference between M10 eq 34 and 3.365 is only the S term and you can see what happens to water vapor in that case.

  27. Jef, Nick,

    Allow me to ask a simple (may too simple) question. If water condenses out of air as a cloud (fog/very fine droplets) the total mass of the air parcel (air plus tiny droplets) would not change at all, since the suspended droplets are being “supported” by the air they are suspended in, even as they slowly fall due to gravity (according to Stokes’ law… sorry, I had to throw that in Nick). The surface pressure therefore would not change just due to condensation to form a cloud, since the total mass in the air column does not change. If larger droplets form, then they could fall out of the initial air parcel as rain droplets and so reduce it’s mass, but even that ought not change the surface pressure until the rain reaches the ground, since even falling rain quickly reaches terminal velocity and is still being supported by the air column, and so continues to apply pressure to the surface below. Am I missing something here?

  28. Another paywall, but may be relevant to the discussion.
    Evaluation of tropical cloud and precipitation statistics of Community Atmosphere Model version 3 using CloudSat and CALIPSO data

    The combined CloudSat and CALIPSO satellite observations provide the first simultaneous measurements of cloud and precipitation vertical structure and are used to examine the representation of tropical clouds and precipitation in the Community Atmosphere Model version 3 (CAM3). A simulator package utilizing a model-to-satellite approach facilitates comparison of model simulations to observations, and a revised clustering method is used to sort the subgrid-scale patterns of clouds and precipitation into principal cloud regimes. Results from weather forecasts performed with CAM3 suggest that the model underestimates the horizontal extent of low-level and midlevel clouds in subsidence regions but overestimates that of high clouds in ascending regions. CAM3 strongly overestimates the frequency of occurrence of the deep convection with heavy precipitation regime but underestimates the horizontal extent of clouds and precipitation at low and middle levels when this regime occurs. This suggests that the model overestimates convective precipitation and underestimates stratiform precipitation consistent with a previous study that used only precipitation observations. Tropical cloud regimes are also evaluated in a different version of the model, CAM3.5, which uses a highly entraining plume in the parameterization of deep convection. While the frequency of occurrence of the deep convection with heavy precipitation regime from CAM3.5 forecasts decreases, the incidence of the low clouds with precipitation and congestus regimes increases. As a result, the parameterization change does not reduce the frequency of precipitating convection, which is far too high relative to observations. For both versions of CAM, clouds and precipitation are overly reflective at the frequency of the CloudSat radar and thin clouds that could be detected by the lidar only are underestimated.

  29. Steve,

    I think that is correct and it contradicts Anstassia’s assertion that surface pressure would lower during condensation. It’s also a valid critique of the one sentence in their paper IMO.

    Consider that if you have this pressure drop in the horizontal replaced by a dynamic pressure balance netting to zero change in the vertical direction. Of course air mass being lower density would flow up and around the droplets possibly even lifting them upward.

  30. @27

    This text from that handy ‘Tuning’ link seems to be germanium to our porpoises.
    Although I guess this is not necessarily the same GCM as Nick is referencing…

    http://www.cgd.ucar.edu/cms/pjr/ccm/tuning/tuning.html#csm:special

    My attempted ‘strong’ text emphasis within this quoted section;

    “2.1 Conservation issues

    It is important that all parameterization satisfy column conservation constraints. One example for water substances treated by a convection parameterization is

    * The water vapor removed from the column by convection must appear as either: 1) detraining cloud water (dlf); or 2) precipitation at the surface (precc). That is,

    {sorrry, garbled equations here. Not central to my point}

    where qnew is the water vapor mixing ratio after the convective parameterizations, and qold is the mixing ratio before convection.
    * The water vapor converted to condensate must release the appropriate amount of latent heat which appears as a temperature change within the column. The convection should conserve a thermodynamic invariant (like moist static energy) within a column.
    * Any other trace species transported by the convection scheme should conserve mass.

    There are currently very small conservation errors in the prototype model. These are caused by:

    * Extremely rare corrections to negative mixing ratios generated by the convection and vertical diffusion parameterizations. Error messages are issued by subroutines when these corrections take place.
    * ignoring the latent heat of fusion in the conversion of water vapor to condensate or phase change of condensate at temperatures below freezing.

    These conservation errors result in small imbalances ( << 1 W/m2) in the CCM. We note that there are also small inconsistencies present in conservation that are associated with the use of a moist mixing ratio, and moist surface pressure in the model. In principle, as any process removes water vapor from a cell, the surface pressure (PS), and the mass of air (dp) should change in a grid volume. This ought to also imply a change to any mass specific quantity affected by the parameterization. These changes are ignored in CCM parameterizations from one process to the next. We typically insist that processes conserve assuming a fixed mass of air (and hence a fixed surface pressure) within a parameterization. "

    RR

  31. Steve,

    I think I might be to hasty in 43. I took the discussion like you did that condensation would cause an instant pressure drop, but if the air rushes upward it will mushroom out sideways as more mass from below follows. The total column mass will drop. If the water vapor were carried with the water, as is guaranteed before it get’s big enough to rain, the whole column would expand reducing ground pressure gradually. This might be different from M10 but that’s what it seems like to me. The dynamic pressure of water drops would work against it though.

  32. Steve Fitzpatrick:

    If water condenses out of air as a cloud (fog/very fine droplets) the total mass of the air parcel (air plus tiny droplets) would not change at all, since the suspended droplets are being “supported” by the air they are suspended in, even as they slowly fall due to gravity (according to Stokes’ law… sorry, I had to throw that in Nick). The surface pressure therefore would not change just due to condensation to form a cloud, since the total mass in the air column does not change.

    No, this statement is only true in the stationary limit, and even then only refers to the weight of the air (condensed moisture, which is probably falling and not “supported”, does not count).

    A specific counter example is what happens when a plane flies overhead? Does the pressure increase while the plane is above you? (Short answer is you don’t see a change in pressure, unless the plane has a very low altitude… this is based on real data).

  33. For those who would like to get a primer on how the models deal with this on a mathematical basis, they are all essentially based on the following treatments by Manabe et al.

    Manabe, and Strickler “Thermal equilibrium of the Atmosphere with Convective Adjustment” journal of Atmospheric Sciences. Vol 21, 1964

    This first one deals with the fundamental physical chemistry of the H2O/air equations

    and

    Manabe, S. and R.T. Wetherall 1967. “Thermal Equilibrium of the Atmosphere with a Given Distribution of Relative Humidity.” J. Atmos. Sci. 24, 241-259.

    Sorry I don’t have links to them but you can pick them up easily on Google

    Rgds
    Terry

  34. Steve 41,
    You’ve touched on an important issue re conservation laws. As Carrick says (and as you did), condensation doesn’t immediately change the apparent weight of the air. Although the laws are often called mass conservation, it’s volume change that matters; the eqn of state is the gas law, and a reduction of gas mass causes a volumr change which eventually translates into a pressure change. How that change actually occurs depends on the momentum equation.

    A comment on planes overhead. I’m sure it is true that no change is observed unless the plane is close. But that’s because you could think of the plane’s weight as being borne by a large area, so it’s a small pressure differential. A plane weighs about the same as the column of air over about 10m^2 of ground (or maybe 20,30 etc). If the pressure differential is spread over 1 km^2, that’s only about 1 Pa.

  35. I have just read the discussion paper Condensation Induced
    Atmospheric Dynamics and it seems quite reasonable given that the forces the authors indicate driving atmospheric circulation are the same as those driving the Newcomen steam engine. Given the publication date of the paper, I see it as unlikely that this information is incorporated in present GCMs. It looks probable that the GCMs will have to be updated.

    However updating the models should not be seen as a disaster. It would give the modellers a chance to further explain the physical mechanism for positive water vapour feedback. I think a clearer indication is required of how water vapour (a gas lighter than air) is prevented from rising through the atmosphere, or if it rises, how it is prevented from condensing to a liquid or solid state and increasing the albedo of the planet.

  36. Strangely Thomas Newcomen when he invented his steam condensation engine in 1712, he called it an “atmospheric engine” 🙂

  37. I think the problem is bigger than rather to incorporate something to the existing GCMs. I cannot claim an exhaustive knowledge of the internal structure of GCMs, although I am learning all my way. I am just a physicist who wants to understand what is going on. I am most concerned about the following.

    First, GCMs are claimed to be based on fundamental equations that can be found in every hydrodynamics section of a course in theoretical physics. Second, GCMs are claimed to satisfactorily replicate the existing climate. This means that the models reproduce the wind velocity field of global circulation. The primary driving force behind air circulation in GCMs is differential heating of the planet surface by the Sun. Switch the sun off, and GCMs will yield a zero field. This forcing is specified in the energy conservation equation in the set of equations that are used to solve the problem. Very good, we feel we have described the circulation in models. Using models we predict what will happen in 100 years.

    Then let us turn to theory. Building a theory constitutes an attempt to reproduce the major quantitative parameters of the circulation from basic physical laws (possibly discovering new ones). Here we find (my emphasis):

    Our understanding of the general circulation, even of the idealized dry atmospheres considered here, is still incomplete. But we have the theoretical concepts (such as available potential energy, Rossby radius, etc.) and computational resources that make it seem possible that a general circulation theory of dry atmospheres, at least on the level of scaling laws, may be within reach.

    Incorporating moist processes in a general circulation theory may be considerably
    more difficult. There are dynamic and kinematic difficulties. Dynamically, the distribution
    of water vapor in the atmosphere impacts the effective static stability, for
    example, in the Hadley circulation and in baroclinic eddies. We do not have all necessary
    theoretical concepts to deal with such moist effects. For example, it is unclear
    what forms an eddy available potential energy or the scale of the linearly most unstable
    baroclinic waves take in a moist atmosphere. Kinematically, it is unclear how the distribution of water vapor in the atmosphere is determined in a balance of advection of water vapor and condensation and moistening processes.

    One should guess when it was written. Perhaps it was written in the 1940s, when no one could afford such a luxury like a numerical computer model? Then GCMs came, people read the studies of Manabe and Wetherald to which Nick referred above, and we’ve got the circulation described? But no, this was written in 2006, in a comprehensive and deep review of the status of the atmospheric circulation theory. (On a grand scale, the problem with the current theory is, expectedly, that the realistic differential heating does not produce realistic wind velocities (the velocities are too low).)

    How can this be? How can a theory be still within reach (a considerate and modest statement), while models already describe the world and make far-reaching predictions? The answer to this question will explain why putting experimental rainfall into a GCM one will never get any hint on the condensation-induced dynamics.

  38. RR 44
    The UCAR text that you cited is from 2000 and talks about a prototype code. It could well have been an early version of CAM3, which I think was released in 2004.

  39. Jeff 40,
    “It’s all complicated but the difference between M10 eq 34 and 3.365 is only the S term and you can see what happens to water vapor in that case.”
    No, 3.365 is Like M10 33. In fact, it’s exactly 33 if you add the steady-state condition
    ∂(N_v)/∂t=-S. (N_v=πq)
    That was the point of my alternative derivation of 34. You can’t drop the horizontal components.

    When you’re following what they say about physics updates etc, you need to watch the alternations. You update pressures etc during the N-S phase of the timestep calc, keeping other things constant, then you keep constant those updates, and change the others. The first stage is marked with a *, the second with a n+1.

  40. I’ve been talking here about Navier-Stokes equations in GCM’s. I’ve now put a post on my blog trying to explain how they handle the basic pressure-velocity interaction, and why they are central to GCM’s.

  41. Re: Anastassia Makarieva (Oct 20 02:07),

    Eloquent and important comment.

    Parameterization.

    Either GCMs are built up from theory, or they are built up from theory and then post-hoc adjusted via parameterization to give results that match the available data.

    I recall reading (but did not tag and cannot cite) multiple instances of strong defenses of GCMs, to the effect that they aren’t parameterized, or that they are “robust” — meaning that most/all physically-plausible parameter values yield more-or-less the same output.

    This seems to be an implausible line of argument.

  42. Carrick,
    “No, this statement is only true in the stationary limit, and even then only refers to the weight of the air (condensed moisture, which is probably falling and not “supported”, does not count).”

    What? Consider a suspension of fine particles with density 2.0 g/ml in water (~1.0 g/ml). If the weight fraction of the particles is 25%, the volume for 100 grams of suspension is 25/2 + 75 =87.5 ml, and the net density of the suspension is 100/87.5 = 1.143 g/ml. You can measure the pressure on the side of the container at any depth below the surface and confirm that the pressure is exactly what you would expect based on the height of the column, the net density of the suspension (1.143) and Earth’s gravity; the pressure rises with depth exactly 14.3% faster than it would for water. It does not matter how big the particles are; they could be 75 nm and never settle due to Brownian motion, or they could be 10 microns and settle over 2 hours, or 100 microns and settle over 60 seconds… you still measure the same initial pressure at a specified depth below the surface, which gradually falls as the particles sediment out of the suspension and are no longer interacting with the fluid. So particles (or droplets) which are sedimenting through a viscous medium (like air) most certainly are supported by the fluid. Otherwise they would continue to accelerate as if in vacuum… but as Nick’s great, great, great, great uncle G.G. pointed out, they quickly reach a fixed velocity. 😉

    The airplane argument (no measured pressure increase from an overflying plane) just tells you that the weight of the plane is supported over such a vast area that you can’t measure it. Near the ground, the momentum of downward directed air flow (from the wings) shows up as a pressure you can easily measure because it is localized. The force of gravity acts on everything that is in the atmosphere, and this generates net surface pressure. If you fly 10^10 planes at the same time, the total mass of the atmosphere would increase significantly, and so would the surface pressure. You can’t measure the pressure effect of a single 10 micron particle sedimenting through a big container of liquid either… it is just too small an effect to see with instruments, but that doesn’t mean it is not there.

  43. 60 Amac
    You should read that quote again and see if you can find the word “model”. Or “GCM”. It is a review of the current state of theories of global circulation. Models of global circulation are different. They rely on fluid dynamical principles and local models of clouds etc. They have their problems too, but are not what Schneider is writing about there.

    Indeed, the first sentence, omitted from the quote, makes the contrast:
    “The simulations presented in this review demonstrate that we can simulate broad continua of possible planetary circulations and can infer unambiguous macroscopic scaling laws governing atmospheric circulations. Our understanding of the general circulation, even of the idealized dry atmospheres considered here, is still incomplete.”

    A free copy of the article is here.

  44. I see the problem like this:

    We can infer macroscopic scaling laws governing atmospheric circulations and simulate a broad range of circulations. But when we put the basic parameters of our own circulation into those unambiguous laws that we have inferred, what we obtain from those laws does not match the observations. This is a typical problem of a theory suggesting that an essential process is missing. So far it has not been possible to prove that differential heating is enough to produce the observed winds.

    GCMs, in my understanding, manipulate more freely with scaling parameters and tune them, using a large number of empirical parameterizations, in such a manner that the resulting circulation matches the observations as close as possible — but they do not know why it is so. This is the only logical way I see to resolve the contradiction between theory and models. So whether a pressure gradient is generated by condensation or by differential heating, a GCM will maintain it close to observations.

  45. Steve, as my example with the airplane demonstrated, suspended particles (whether it be an airplane or a cloud of particles) doesn’t satisfy the statement “the pressure is equal to the weight of the column of air divided by it’s area.”

    Beyond, that it’s elementary to see that suspended particles don’t follow that usual statement. Again Stoke’s Law:

    rho Du/Dt = -grad p – rho g

    If you set Du/Dt = 0 and assume horizontal stratification in p, the result is trivially satisfied. (Otherwise it’s not trivially satisfied, and in general not satisfied at all.)

    But notice this is the pressure of air molecules we are summing, and the result requires horizontal stratification.

    Now, what happens if we have a parcel of air condensate? We no longer have a stratified medium (so there are horizontal and vertical components to grad p) and all bets are off, just like the example of the air plane. Actually what happens in this case is we end up with motion of the air, but it’s well known that condensation (or evaporation) of a cloud generates local air motion (what exactly happens depends on the lapse rate of the air).

    It’s true if you average over one spot for long enough time, the “static pressure” you measure will be exactly equal to the weight of the column above it. But it’s not true that the pressure you measure at any instance is equal to the “static pressure”, even forgetting about high frequency propagating sound. (Boundary layer physics for one thing generates violations of this, as does any sort of convective motion above it.)

    I’ll see if I can dig up some more text book examples where that column of air problem doesn’t apply, when I get to my office this morning (running late).

  46. So whether a pressure gradient is generated by condensation or by differential heating, a GCM will maintain it close to observations.

    WRT model parameterization, this reminds me of the Steig deconstruction a bit. Because Steig’s model did not include higher order PC’s, it was not possible to produce a reconstruction with constrained regional warming of the peninsula. When fitting the flawed model to observation then comparing (vis a vis Jeff’s videos) we could see all kinds of spurious warming and cooling over time – an inevitability when the model could not express higher order regional warming.

  47. Carrick,

    “Steve, as my example with the airplane demonstrated, suspended particles (whether it be an airplane or a cloud of particles) doesn’t satisfy the statement “the pressure is equal to the weight of the column of air divided by it’s area.”

    I never said anything about a specified column of air. The area is the entire Earth surface area. The “column” is the entire atmosphere.

    Local pressure variations due to atmospheric motion (weather) don’t change the surface pressure (on average), since this must be the total weight of the atmosphere divided by the area of the earth’s surface. Condensed water in the form of droplets for sure can cause motion of the air, but that does not mean the condensed droplets (in clouds or as falling rain droplets) do not add to the total weight of the atmosphere, and so to the average surface pressure. My objection was to your statement ‘only refers to the weight of the air (condensed moisture, which is probably falling and not “supported”, does not count)’. The weight of droplets, falling fast or slow, most certainly does add to the total weight of the atmosphere, and so contribute to the average global surface pressure.

    Do you believe that flying 10^10 airplanes (about 5 * 10^11 tons, compared to ~5.3 * 10^15 tons atmosphere) would not increase average atmospheric pressure by about 1 part in 10^4 (0.1 millibar)?

  48. This is probably unwise, but here goes.

    Most GCMs use the hydrostatic approximation in which the pressure at any point is exactly equal to the weight of the air column above it. Very high resolution weather models sometimes use the proper non-hydrostatic equations, but this isn’t very important at coarse scale (I’m sure there is a paper that has demonstrated this somewhere).

    Given the hydrostatic approximation, condensation *per se* has no effect on air pressure (exactly as stated in Steve Fitzpatrick’s point above). The paper being discussed appears to conflate condensation and precipitation out of the column – whether that is just a looseness of language or a conceptual issue is unclear to me. (The existence of clouds should provide some indication that these two processes are different). The removal of water mass (via precip) or it’s addition (via evaporation from the surface) does of course affect the pressure since it changes the total mass in the column. The magnitude of the effect is however relatively small – water vapour in the tropics (around 60 mm liquid water equivalent) contributes about 6 hPa out of ~1000 hPa at the surface. Gradients in pressure due to water vapour effects are of course much smaller.

    Some GCMs do not take the mass of water of water vapour into when calculating pressure, though they all do use it in density calculation when calculating buoyancy forcing or deciding whether convection will occur (usually via the virtual potential temperature). Some GCMs (for instance GISS AOM (subroutine MSTCNV) ) do include the mass of water vapour in pressure calculations, (and this will be included in the next version of ModelE). However, the difference this makes in the circulation is relatively small. It certainly is not the difference between winds and no winds. It should be included because it is a real effect, but that is a long way off from saying that it is dominant, or even important.

    Analogies with steam engines are misplaced – in that case you are adding mass to the gas, and so of course the pressure rises. A better analogy is that of an ice cube melting in fresh water. This does not change the level of water in the glass and so does not change the pressure at the bottom either. Condensation of water vapour to form a cloud does not affect surface pressure, though it does affect the density of the air mass via the change in both the mass of the gas and the latent heat and that obviously affects the flow subsequently).

  49. We measure pressures when flying.

    We have the total pressure measured by the petot tube .
    This pressure is compensated by the static pressure.

    This provides the actual speed of the aircraft relative to the air.

    Above the wing there is under pressure responsible for about 70% of the Lift
    Under the wing there is over pressure producing the remaining 30 % of the lift.

    The highest pressures can be found at the front side of the aircraft, the nose cone and the leading edge of the wings, the elevator and the stabilizer.
    The lowest pressures in the center of the vortex which develops from the Left and Right Wing tip.

    The difference in pressures depend on the aerodynamic design and the air speed of the aircraft.

    Because air flows from points with high pressure to points with low pressure the airflow over a wing is bend off in the direction of the wing tip where it causes a vortex
    (this drag is called Induced resistance) To lower the induced drag and to suppress the effects and strength of the vortex (fuel consumption an air safety) the Winglet was developed and now applied in almost all commuter aircraft.

    Almost all modern aerodynamic features and the application of smooth and light weight composites originate from innovations in Gliders.

    Local pressure differences are equaled out very quickly but there is so much energy in a vortex that aircraft that fly behind a starting are landing aircraft need to separate.
    Air traffic controllers raise the separation time up to three minutes after the take off of a Boeing 747 or Airbus to 1 minute for the smaller commuter.

    During soaring competitions the wings of gliders carry water ballast (because it makes them faster when speeding from one thermal to another like a heavy skier down a hill is down more quickly compared to a light person although both have the same angle of descent)

    It has been a proven competition strategy for the pilot flying the plane in the highest position in an updraft or thermal to release his water ballast while circling thermal thus reducing the ascending speed of the air column as it cools down due to the evaporation of the water released.

  50. Anastassia,

    “GCMs, in my understanding, manipulate more freely with scaling parameters and tune them, using a large number of empirical parameterizations, in such a manner that the resulting circulation matches the observations as close as possible — but they do not know why it is so.”

    They actually are specifically not recognizing the effects of condensation based volume change from what I can tell. I’ve had a nice discussion with Nick on the matter but am convinced that they are not including the effect at all. What happens is that they hold ground pressure and mass of the total volume constant as the phase change occurs, then they adjust the specific humidity. After that is done physics calcs perform other operations which don’t correct the first problem. The volume and pressure are explicitly constant during phase change.

    What has been noticed from this assumption is a global imbalance in energy which is fairly minor and is then corrected for (brute force style). Because the models assume that the energy available for convection is basically the energy of the atmosphere not including the phase change they only consider (realize) that a small energy imbalance must create only a small effect on outcome — small effects can’t change winds much.

    They completely miss the pressure gradient created by the condensation and I am more convinced than ever that it is a huge problem.

    BTW, there is evidence that models don’t match observation all over the place. It’s not a battle you probably want to fight IMO, you have enough on your hands, but I sure do. In my opinion this is the biggest thing to hit climate science I’ve ever seen. I could be wrong still, models are complex but I don’t see anywhere where these problems you have identified are corrected for. I have searched 100% of the correct code sections and not seen it. I’ve read very carefully Nick’s comments, I agree with his attitude that it ‘must be in there’.

    But it isn’t and actually it’s explicitly stated not to be. Perhaps we should be taking their word for it.

  51. Steve Fitzpatrick:

    The area is the entire Earth surface area. The “column” is the entire atmosphere

    Actually, it’s not true even that the average pressure over the entire Earth is equal to its weight divided by its surface area. Because the Earth is curved, that produces a geometric correction.

    For simplicity, let’s assume we have a thin atmosphere of height H above a sphere of radius R, and further can neglect variations in g.

    From Stoke’s law, we have,

    p0 = rho g H

    which is the standard result.

    The volume of the atmosphere is

    Delta V = V_top – V_bottom = 4/3 * pi * [(R+H)^3 – R^3]

    Writing out the terms gives:

    Delta V = 4 pi (R^2 H + R H^2 + H^3/3)

    so the weight is given by

    W = rho g Delta V = 4 pi rho (R^2 H + R H^2 + H^3/3)

    The area at the surface is

    S = 4 pi R^2

    So

    W/S = rho g H * (1 + H/R + H^2/3 R^2)

    In the limit that R->infinity, W/S = p0.

    Thus when you are discussing the approximation ‘the pressure of a column of air is equal to it’s weight divided by its area” you are always implicitly assuming “flat earth” = regional scale physics.

    I’ll answer your other question separately…

  52. Steve:

    The weight of droplets, falling fast or slow, most certainly does add to the total weight of the atmosphere, and so contribute to the average global surface pressure

    Well they probably do, but it isn’t something you’re going to find in Stoke’s equation, and you’re going to find in general that the weight on the ground isn’t equal to the weight of the air + the water droplets above you.

    In short, it’s not governed just by Stoke’s equation (you have to include equations of states for gases and liquids, mass conservation equation + thermodynamic equations). And even for Stoke’s equation you can’t ignore the terms on the left-hand side of the equation: Du/Dt = du/dt + grad . u,.

    Just what pressure you observe on the ground when the water vapor condensates depends on a number of factors, including the instantaneous lapse rate of the air (unstable, neutrally stable, conditionally stable, etc… plays a role).

  53. One further point: the change in density related to condensation is dominated by the latent heat release, not the reduction in q (which is why moist convection is so important for climate).

    Take a super-saturated air mass at (say) 288 K, and 0.02 kg/kg water vapour at near sea level (1000 hPa) (saturation would be 0.01687 kg/kg). The density is rho = p/RT * (1+q)/(1+1.609q) = 100000/(286.9*288) *1.02/1.032 = 1.196 kg/m3.

    Now if the excess water vapour condensed spontaneously (which is why I started with super-saturated vapour), you would change 0.00313 kg/kg of vapour to liquid condensate. The latent heating is 334,000 J/kg, and specific heat of air is 1005 J/kg and so the temperature will rise from 288 to 288+0.00313*334,000/1005 = 289.0 (ie. increases by about 1 deg C). The new density is then
    100000/(286.9*289) *1.01687/1.027 = 1.194 kg/m3 i.e. less dense than before. The change in density due to the change in q would have been +0.002 kg/m3, and the change from the change in temperature is -0.004 kg/m3 (twice as big and in the other direction). All of these density affects are included in GCMs.

  54. According to RPS,

    First, there are always tunable parameters within each parameterization, and there are always quite a few more than one or two.

    In my class on modeling, the students have documented the number of tunable parameter for a range of parameterizations, and 10 and more are common for each individual parameterization (e.g. see the class powerpoint presentations at ATOC 7500 for my most recent class).

    Second, the only basic physics in the models are the pressure gradient force, advection and the acceleration due to gravity. These are the only physics in which there are no tunable coefficients. Climate models are engineering codes and not fundamental physics.

  55. Gavin:

    Most GCMs use the hydrostatic approximation in which the pressure at any point is exactly equal to the weight of the air column above it. Very high resolution weather models sometimes use the proper non-hydrostatic equations, but this isn’t very important at coarse scale (I’m sure there is a paper that has demonstrated this somewhere).

    This does illustrate the differences in approach for different fields.

    I’m specifically interested in using pressure sensors in remote sensing applications, and in general when I field a sensor, we don’t have instantaneous hydrostatic equilibrium. When you have a cloud forming or evaporating, that is certainly a violation of hydrostatic equilibrium, and the physics I’m interested in, I need that violation to see any effect on pressure.

  56. My bad, I flubbed the material derivative equation….

    Du/Dt = du/dt + u.grad u

    I was in too big a hurry to get to my sammich.

  57. #69 Gavin,

    The removal of water mass (via precip) or it’s addition (via evaporation from the surface) does of course affect the pressure since it changes the total mass in the column. The magnitude of the effect is however relatively small – water vapour in the tropics (around 60 mm liquid water equivalent) contributes about 6 hPa out of ~1000 hPa at the surface. Gradients in pressure due to water vapour effects are of course much smaller.

    This is an exact and helpful description of how the effect is treated in the conventional paradigm. In reality, pressure gradients associated with condensation cannot be estimated from the value 6 hPa that corresponds to the weight of vapor. Just have nothing do to with it.

    Indeed, suppose that there is a process that removes the gas from the atmosphere over an area of a linear size L. This process (let it be precipitation) will create some pressure shortage in the area where it operates compared to the ambient environment. The air will flow towards the precipitation area tending to relax this pressure difference. The stationary value of the gradient will, obviously, depend on the relative rates of the process that creates the pressure shortage (precipitation) and the process that erases it (air flow caused by that pressure shortage). A simple scale analysis would show that for large-scale problems the resulting gradient will be described by Eq. 37 in M10.

    The small value (6 hPa) corresponding to the weight of vapor in the column is due to the fact that vapor has already condensed in the upper atmosphere and has a “compressed” distribution compared to other gases. The value of relevance for calculating the condensation-induced pressure gradient is the adiabatic liquid water content (corresponding to approx. 30 hPa, surface vapor pressure) — namely this amount determines the precipitation rate.

    But my main point here is that in the meteorological literature, to my knowledge, there is not a single paper where this effect would be treated from the basic physical principles or estimated in any other way. It was ignored without being evaluated.

    #76 Gavin,

    The density effects are not relevant altogether to the effects of mass removal that are discussed. Pressure at the surface depends on the amount of gas in the column and is not dependent on density. For this reason, whether condensation is accompanied by release or absorption of energy or if L = 0, provided matter is removed from the column, surface pressure will fall. Again, the magnitude of this fall will depend, in a sigmoid fashion, on the rate at which the matter is removed.

    I have a message from my colleagues, and do it with pleasure myself, to thank all the critical commentators (Nick, Carrick, Daniel, Steve, Gavin), as we are learning and trying to better shape our stuff. Please, criticize as harshly as you can, you are welcome!

  58. “…that of an ice cube melting in fresh water. This does not change the level of water in the glass and so does not change the pressure at the bottom either. ”

    Ummm. I only point this out becuase I have seen this statement made (or worse) in discussions related to Artic Sea ICE melt and rising sea levels.

    An ice cube floats in water because the density of the ice cube is less than the density of the water it displaces (by volume).

    This is due to the consistent spacing of the molecules in the crystal lattice of the ice cube. When the ice cube melts, the mass does not change, but the density goes up (water has a nonuniform spacing of the molecules) – meaning the volume of the water from the ice cube will be less than the volume of an ice cube. Thus the displacement of the original water is also reduced.

    Thus the overall level of the water in the glass after total ice melt goes DOWN.

  59. Carrick #79,

    I do not know what Stokes’ law you are talking about, but I am talking about this one:

    Terminal velocity of sedimentation = (D^2) * g *(RHOp – RHOf)/(18*visc)

    Where RHOp is the particle density, RHOf is the fluid density, D is the particle diameter, g is gravitational acceleration, and visc is fluid viscosity

    Once the particle reaches this terminal velocity (very quickly for small particles), the drag force on the particle equals the gravitational force…. the speed is constant because the particle is ‘supported’ by the fluid drag.

  60. #81 This is off-topic, but the issue with Arctic sea ice melting into salty water (which does slightly raise sea level) is due to the different salinties in the ice (~5 psu), compared to the ocean (~35 psu). If the salinity is the same (as with a normal ice cube and freshwater), the melt has exactly the same density as the original water and so the mass displaced by the ice volume is exactly the same as the mass of the melt – thus no change in sea level. In the ocean, fresh water is less dense than salty, and so the density of the melt is less than the density of the displaced water and so the volume of the melt is a little greater than the volume of the displaced water, and so sea level rises slightly.

  61. Gavin,
    “This is probably unwise, but here goes.”

    I hope the responses to your comments here will dissuade you of that suspicion. Most here treat people with some measure of respect, even when they disagree.

  62. I have seen a number of references to the temperature increasing as water vapor condenses and releases the latent heat. I thought the process was a rising packet of moist air would be adiabatically cooling until the water started to condense and then the temperature would not drop as fast as the packet continued to rise. The air continues to cool at the moist lapse rate that is slower than the dry lapse rate. There should never be an actual temperature increase. Its latent heat, not sensible heat. Even an apparently static cumulus cloud has an updraft feeding more moisture and and the up flow keeps the cloud droplets suspended at the same altitude even though they have a downward air speed.

  63. #80

    This is an exact and helpful description of how the effect is treated in the conventional paradigm. In reality, pressure gradients associated with condensation cannot be estimated from the value 6 hPa that corresponds to the weight of vapor. Just have nothing do to with it.

    Nothing in my statement above is associated with a ‘paradigm’ of any sort. It is simply statics. Anything that creates pressure gradients will cause flow, and removal of mass from a column clearly does. In that everyone seems to be in agreement.

    The small value (6 hPa) corresponding to the weight of vapor in the column is due to the fact that vapor has already condensed in the upper atmosphere and has a “compressed” distribution compared to other gases. The value of relevance for calculating the condensation-induced pressure gradient is the adiabatic liquid water content (corresponding to approx. 30 hPa, surface vapor pressure) — namely this amount determines the precipitation rate.

    Obviously the amount of water in the atmosphere is determined by where it has condensed and precipitated out. But this can be easily determined from a column calculation of the moist adiabat (start with a saturated parcel and lift it through the column). This is independent of any horizontal pressure gradients or flow. The amount of water that there would have been without that condensation is irrelevant for determining the impact of condensation on the environment.

    But my main point here is that in the meteorological literature, to my knowledge, there is not a single paper where this effect would be treated from the basic physical principles or estimated in any other way. It was ignored without being evaluated.

    Perhaps I’m not clear what ‘effect’ you are discussing. The fact that removal of mass causes a reduction of surface pressure is trivial and I presume it was known to Euler, Bernouilli and everyone since. Are you talking about this, or about a change in pressure due simply to condensation (i.e. cloud formation?). In that case there is no change in surface pressure, but there are changes in density (dominated by latent heat release) that will effect the environment and cause pressure differences at any specific height in the column. GCMs include these effects, but don’t in general use water vapour in the pressure calculation directly (but as I said, this is a small error – around 0.6% at the surface and increasingly smaller at altitude).

  64. Gavin, #81,

    I’m not so sure about that. When ice freezes out of the ocean, the density of the non-frozen part of the ocean rises (on average) due to higher salt concentration, but the total mass and overall average density ought not change (lower density ice plus higher density saltier water). The amount of salt trapped in the ice ought not make any difference. Trapped salt just make the ice ride a little lower in the water. The total volume with or without ice (at constant temperature) ought not change at all. Can you explain why the above logic is incorrect?

  65. Gavin, thanks for your comments.

    One related question, if you know the answer: Is Anastassia’s statement that “On a grand scale, the problem with the current theory is, expectedly, that the realistic differential heating does not produce realistic wind velocities (the velocities are too low).” a fair statement (i.e., suggesting tha something is missing), or do you feel the velocities are realistic?

  66. Gsvin, Anastassia’s point is that the problem we’re addressing isn’t with density change of the volume. It’s the gas pressure change that we’re discussing and it’s not fully explained by density because as vapor condenses it no longer contributes to the pressure but still contributes to density.

    I really don’t have time right now but if you read Anastassia #80 you can see that is also what she is saying.

  67. #88 This is all described in this paper,

    Shepherd, A., D. Wingham, D. Wallis, K. Giles, S. Laxon, and A. V. Sundal (2010), Recent loss of floating ice and the consequent sea level contribution, Geophys. Res. Lett., 37, L13503, doi:10.1029/2010GL042496.
    http://www.agu.org/pubs/crossref/2010/2010GL042496.shtml

    and I’m pretty sure I recall Bob Grumbine working it out way back in the sci.environment days… oh yes:
    http://moregrumbinescience.blogspot.com/2009/04/ice-and-sea-level.html

  68. #87 Gavin,

    Perhaps I’m not clear what ‘effect’ you are discussing. The fact that removal of mass causes a reduction of surface pressure is trivial and I presume it was known to Euler, Bernouilli and everyone since.

    We are discussing the effect precipitation produces on surface air pressure and what types of gradients are generated. In particular, we show that precipitation in the tropics produces pressure gradients of sufficient magnitude to drive the Hadley cell. We assert that this effect of precipitation on pressure gradients has never been discussed in the meteorological theory. Actually we speak a lot of this in the paper.

    In reply to your comment, I pointed out that pressure gradients associated with mass removal cannot be estimated from the standing value of vapor weight (6 hPa) and explained why. Your statement that “gradients in pressure due to water vapour effects are of course much smaller” does not specify smaller than what.

    Sorry I’ll be offline for the following 8 hours.

  69. #89 I don’t see any evidence for this. Wind velocities in GCMs are close to observed, and even better in higher resolution weather forecast models. Possibly there is a noticeable difference in moving to non-hydrostatic equations, but I don’t know offhand of a paper demonstrating this.

    #90 This is not clear to me: pure condensation (without removal of mass) (i.e. to form a cloud) does not affect the pressure directly. It does affect the temperature and the density via latent heat and equation of state effects. Your statement “because as vapor condenses it no longer contributes to the pressure” is wrong. Only if you have precipitation out of the air mass will the pressure change. With removal, the pressure will obviously decrease, and the density will decrease also (but if L=0, the density would increase).

    Consider a thought experiment. Take a closed container filled with super-saturated air and place it on a scale. Will you be able to detect the moment of condensation by monitoring it’s weight (i.e. the pressure on the scale)?

  70. Gavin #92,

    OK. 0.05 mm per year contribution to ocean rise from the portion of the ice that sticks up in the air is believable. But 5 mm level rise per century at current melt rates is not a lot.

  71. Gavin,

    I don’t have time but your thought experiment doesn’t cover the problem I’m addressing. I thing some math is in order for this so it may take until tonight.

    This is the sort of thing I mean.

    http://www.chem.uiuc.edu/clcwebsite/can.html — of course the can’s mass didn’t change at all.

    Condensing vapor no longer contributes to the pressure in the local region. Remaining gasses are less dense causing significant horizontal and vertical flows. Ground pressure is something else. As I understand it, CAM3.0 is holding dry air mass constant and ground pressure constant while making this transition. I don’t see where the pressure drop is computed in the layer from the condensation parametrization. If the gas mass is held constant, the system actually gain’s a little mass during condensation which is later corrected by energy rebalance. I may have that wrong because model details are new to me, but the condensation volume/pressure loss is real.

    Anastassia’s made no real claims about models, that is my own doing. She’s of the opinion that her calculations aren’t standard fare, Nick and I are simply trying to figure out where these effects are accounted for. Nick keeps writing that they have to be, I agree but we can’t find em. So far, it looks to me like she’s right.

  72. I’m much more in my element when we’re discussing proxy processing methods, so I’ll just sit back and listen and try to figure out what I have to learn in order to understand the GCMs. I just wanted to thank all the contributors for a facinating read.

  73. Jeff Id:

    BTW, there is evidence that models don’t match observation all over the place. It’s not a battle you probably want to fight IMO, you have enough on your hands, but I sure do.

    You mean like large regional discrepancies of over 400% in specific humidity or 4K in temperature?

    Temperature and humidity biases in global climate models and their impact on climate feedbacks
    V. O. John and B. J. Soden
    GEOPHYSICAL RESEARCH LETTERS, VOL. 34, L18704, doi:10.1029/2007GL030429, 2007

    Click to access 2007GL030429.pdf

    Although they claim that it doesn’t matter because the biases are robust.

  74. #93 Thank you. It is now clear to me that we are discussing the impact of the removal of water vapour from the column on the pressure gradients, not the process of condensation itself. I would certainly recommend clarifying that further in your paper and in discussions.

    As I stated above, this effect is not included in many GCMs since the water vapour mass is not included in the pressure field/3d air mass calculation. This is of course an approximation and is something that will eventually be fixed. However, it is already included in some GCMs, GISS AOM in particular, and perhaps others. The difference it makes to the circulation is small, but I am not aware of any paper that has specifically calculated the impact. The implication that this is an unappreciated approximation would be an exaggeration. I mention it in Schmidt et al (2006) for instance:

    In common with most other models, we make some basic assumptions at
    the outset, which though minor, have consequences throughout the
    model, namely: water vapor does not add to atmospheric mass
    (i.e. globally integrated surface pressure is constant)
    , the latent
    heat of atmospheric water vapor does not depend on temperature
    (i.e. all atmosphere-surface freshwater fluxes are assumed to be at
    0\deg C), the potential energy of water vapor/condensate is neglected,
    condensate is not advected, and the pressure gradient calculation does
    not include humidity effects. We hope to be able to relax these
    constraints in future versions. The principal prognostic variables in
    the atmosphere are the potential temperature, water vapor mixing ratio
    and the horizontal velocity components. Virtual potential temperature
    is used for all density/buoyancy related calculations.

    I don’t know if Gary Russell specifically tested in GISS AOM whether including the water vapour mass in the pressure field on its own made any significant impact to the circulation, but I can ask. Of course, someone else could take the code and do the experiment themselves. It might be a little tricky (trying to track down the water-related mass field change ‘MG’ in the cloud and surface flux routines), but it should be doable.

  75. Jeff Id said
    October 19, 2010 at 3:23 pm

    #8 Derek,

    ” Can you point to a source which demonstrated the condensation pressure was not included in models? I doubt very much that you can, ”

    #80 Anastassia Makarieva said
    October 20, 2010 at 12:52 pm

    ” But my main point here is that in the meteorological literature, to my knowledge,
    there is not a single paper where this effect would be treated from the basic physical principles or estimated in any other way.
    It was ignored without being evaluated.

    I hope you see the answer therein Jeff id. Although your question was phrased (unintentionally I genuinely assume) as a “negative”.

    I realized almost immediately I first thought about this issue, that the models scaling would not allow it’s reasonable inclusion.
    [clouds are not represented individually in models just as a % of cloudiness per “cube”.
    Most “cubes” getting larger with altitude, from memory there were 11 (not equal height) layers of “cubes from earth to “space”,
    maybe more or less now I do not know, or frankly care.
    These “cubes” are quite massive at least partially to reduce the computing power required to crunch the models.]
    Scaling and the none inclusion of this effect explain why models can not model hurricanes, let alone clouds.
    It is often admitted to indirectly by “climate modelers” in that they do not understand clouds yet..

    It all goes back to my basic point I have made so often, question the principles, do not waste your time quibbling the figures.
    Quibbling the figures is what your supposed to do, you’ll miss the trees for the wood. This is a case in point.
    As the simple explanation above hopefully shows.

    It’s a deep, deep rabbit hole AGW, in all it’s forms, they’ve had a long time, and a lot of (our) money to protect it with.

  76. #94, Gavin,

    “Consider a thought experiment. Take a closed container filled with super-saturated air and place it on a scale. Will you be able to detect the moment of condensation by monitoring it’s weight (i.e. the pressure on the scale)?”

    No, but you won’t be able to take the lid off because of the large pressure differential. This is why pots have “spherical” lids.

  77. Derek, for clarification of #8, I guess you are saying that you knew all about this specific issue even prior to this article’s release. It was kind of hard to decipher in the middle of all that commentary.

  78. #101, Paul, alternatively perform Gavin’s experiment with a sealed plastic bottle, and watch it collapse as the weight remains constant.

  79. Re: Paul linsay (Oct 20 15:45),

    He did say supersaturated, so the temperature will go up too.

    I think if you run the numbers that the pressure on the walls of the container will be less than the pressure on the floor when water droplets are present. My gut feeling is that the pressure on the floor will be the same whether the water droplets are suspended or in a thin layer on the floor. But the pressure on the walls only depends on the gas pressure, which will also be the same whether the droplets are suspended or in a thin layer on the floor. If the container is not in a gravitational field, then the pressure on all the walls will be the same and determined by the number of moles of gas, the temperature and the volume of the can less the volume of the condensed water.

  80. #94, Gavin,
    “Consider a thought experiment. Take a closed container filled with super-saturated air and place it on a scale. Will you be able to detect the moment of condensation by monitoring it’s weight (i.e. the pressure on the scale)?”

    No, the weight will not change as no atoms have been removed from the container. The pressure in the container will however drop and this is what is being discussed in the paper. Gavin has indicated that steam engines are not comparable, but this is not entirely correct. The Newcomen engine was a steam condensation engine, and due to its primitive design there was both air and water vapour in the working cylinder. It was the condensation of the water vapour that pulled the piston down into the cylinder.

    The paper under discussion indicates that the condensation of water vapour within the atmosphere creates a similar force, which due to the geometry of the atmosphere ( thin and wide) results in largely horizontal forces as the gas in the atmosphere tries to correct the pressure imbalance. Discussion of vertical movements and pressures seem largely a distraction from the central issue in the paper, which is that the horizontal movement of air in the atmosphere is driven by the condensation of water vapour rather than temperature gradients as previously supposed. In modeling wind, present GCMs must be modeling using temperature gradients with an added fudge factor.

  81. Gavin,

    “we make some basic assumptions at the outset, which though minor, have consequences throughout the model”

    Can you show us a calculation of pressure loss you would typically see from condensation at altitude so we can understand that it is truly a minor effect. Some of us really are under the impression that it is not. Not surface pressure loss as you may think but pressure loss in the region of the condensation. If you can show something simple (and germane to the point) it might be very convincing.

    We’re almost entirely engineers and scientists so simply calling the effect small is easy but won’t convince us. Many of us are under the impression that the condensation effect can even drive the flow in a hurricane. Numbers don’t lie though so if you can do it well, you will convince me.

  82. 94 Gavin, 104 DeWitt –

    Please can I check my understanding of this thought experiment?

    Surely if it is a closed container its total mass is unchanged by the physical state of its contents hence the reading of the scale would remain unchanged?

    If the question is “does it’s bouyancy change and hence effect the reading on the scale” I would again say no – if it is a fixed volume container and the mass of its contents remains constant surely its bouyancy also remains constant?

    And I’d expect the average pressure on the floor of the container to remain unchanged regardless of the state and/or arrangement of the contents. Pressure is force/area so (for a fixed area floor) a change in pressure on the floor would indicate a change in the downward force experienced by the floor. This downward force should surely be equivalent to the total mass of material being held in vertical equilibrium against gravity and hence should be unchanged?

    Sorry if I’m missing something obvious – I have thought about this without running numbers and I’d appreciate knowing where I’ve gone worng. I’ve tried to see how supersaturaion would change this but can’t. Thanks.

    (whilst typing this up I see Konrad has posted a similar view)

  83. 106 – Jeff, I recall this comment number 52 from Nick Stokes last time we discussed Anna’s work:

    “I think the effectiveness of condensation as a power source is overrated in this discussion. Saturated air at 30C contains 4.2% wv. So after complete condensation, the volume drops by that amount. But according to Emanuel (1991), air rising in a hurricane drops by 33% in degrees K. So WV is a small factor in contraction.”

    I haven’t reread the entire thread – I just went back to find this reference. From memory I don’t recall it being refuted. I should say I haven’t followed this closely or read Anna’s current paper.

    https://noconsensus.wordpress.com/2010/02/02/the-power-behind-hurricanes-and-tornadoes/

  84. Steven Fitzpatrick:

    I do not know what Stokes’ law you are talking about, but I am talking about this one:

    LOL. That explains some of the confusion.

    I’m talking about Stokes equation… the one for fluid mechanics (as in Navier-Stokes).

    Another topic we disagree to disagree on …. if for no other reason that I’m railed out at work right now..

    Carrick

  85. DeWitt #104,
    “think if you run the numbers that the pressure on the walls of the container will be less than the pressure on the floor when water droplets are present. My gut feeling is that the pressure on the floor will be the same whether the water droplets are suspended or in a thin layer on the floor. But the pressure on the walls only depends on the gas pressure, which will also be the same whether the droplets are suspended or in a thin layer on the floor.”

    The “pressure on the floor” part is right, the “pressure on the walls” part is not right. See my comments starting at #41 above, especially the discussion of a suspension of dense particles in water… I have done the measurement, and the pressure on the side of the container when dense particles are suspended really does rise more the the pressure you would expect from the fluid alone. Suspended water droplets are supported by “drag” as they settle through the air. The pressure (walls or floor) only depends on the mass of material above the altitude where you are measuring and the gravitational force applied.

  86. Carrick,

    “I’m talking about Stokes equation… the one for fluid mechanics (as in Navier-Stokes).”

    Well, I am glad that is cleared up, I was beginning to think one of us was crazy, and I was hoping it was you. 😉

    “Another topic we disagree to disagree on”

    Say what? Never heard that one before.

  87. 41, 110 Steve (with apologies for joining the discussion late)

    110: “The “pressure on the floor” part is right, the “pressure on the walls” part is not right. See my comments starting at #41 above, especially the discussion of a suspension of dense particles in water…”

    Steve – In the the thought experiment proposed in 94 – are you saying the pressure on the walls is increased to support the mass of the falling particles via drag whilst the pressure on the floor is unchanged? Surely the resultant pressure on vertical walls will only have a horizontal component? If it were to have a vertical component to supply the reaction to the drag force wouldn’t this be transmitted vertically downward to the point where the walls join the floor? If that were to happen and the pressure on the floor is unchanged the sum force on the floor would increase and this wouldn’t this change the reading on the scale?

    94 Gavin – I read back up the thread and saw your 69 analogy with melting ice so I don’t think we are in disagreement over the thought experment you proposed. I’d welcome any comment on the significance of supersaturation – was this simply to provide the “thought conditions” for precipitation?

  88. Curious, #112

    I am saying that there is no distinction between particles of different sizes, they all have the same effect on the pressure versus height function of a fluid column.

    A salt solution in water with a density of 1.05 g/ml applies the same pressure to the side of it’s container as does a suspension of 150 nm (or 1,500 nm, or 15,000 nm) particles in water that produce a net suspension density of 1.05 g/ml. Put a pressure gauge on the side and the reading at the same depth will be the same… the size of the “particles” does not matter… even if that are ions (and please trust me on this, I have done the measurement). The net pressure at every point depends only on altitude, the mass of what is above that altitude, and gravitational acceleration.

    Same thing with clouds. The net density of the cloud (water droplets plus air) is what matters in the behavior of the “combined fluid” is not the density of the individual phases.

  89. #106 I’ll think about it some more….

    However, I think you can demonstrate to yourself that cloud formation (which is the condensation without removal that you are talking about), does not cause some huge direct effect on the circulation. (Note, however, that Anastasia appears to be talking about something else – the impact of the removal of water by precipitation).

    We have often seen clouds form and dissipate on otherwise uneventful afternoon, right? Now think about what would happen as a cloud started to form – according to your suggestion this would cause instant horizontal convergence and the cloud would quickly be reduced in size. Similarly, if a cloud was dissipating, you would see instant divergence, pulling the cloud apart. Neither of these things accord with my experience.

  90. #112 super-saturation was posited so that the condensation can happen without external drivers, which would complicate the situation.

  91. Curious, #94,

    I do not think Gavin is coming back. So, the increase in sea level with melting of floating sea ice has to do with the difference in density between sea water and pure water, which turns out to be about 2.6%. When ice floats in sea water, the volume that is above the surface of the sea is higher than if that same ice were floating in pure water. When the ice melts, that “extra volume” (about 2.6% of what melts) returns to the sea, and so raises the level. If salt were incorporated in the ice at the same concentration as the salt in the ocean, there would be no change form melting.

    Here is the interesting thing: Increases in sea ice must therefore influence the net sea level. If you combine the floating sea ice in the arctic and the antarctic, the annual freeze/melt cycle (north minus south) ought to show up as a seasonal fluctuation in global sea level unless the opposite sinusoidal accumulations are exactly balanced. Worth looking at? Don’t know, but I sure don’t have time to look!

  92. 113 – thanks Steve. I think, in a static system, we are in agreement over the pressure at a given altitude being a function of the weight of material above that altitude. Where I remain confused is how you see drag operating on the walls in a moving system (your 110 comment)? Rereading your comments I see they could be interpreted to mean that the “drag” component is resolved via the floor rather that the walls – in which case we are in agreement again. Have I understood you correctly? Thanks

  93. 115 – Gavin, thanks that clarifies. In your thought example what strikes me is that we seem to agree on the vertical component as experienced on the floor being unchanged by precipitation. However in the horizontal case I think there will be a pressure drop and the only thing preventing movement of air into this lower pressure zone will be the plate stiffness of the walls of the container. Do you agree with this?

    Also, I echo comments from upthread, thank you for participating.

  94. Gavin,

    I’m just reading other things right now pertinent to the discussion.

    You wrote:

    “We have often seen clouds form and dissipate on otherwise uneventful afternoon, right? Now think about what would happen as a cloud started to form – according to your suggestion this would cause instant horizontal convergence and the cloud would quickly be reduced in size”

    This isn’t a simple enough situation for me. For instance, if the cloud was narrow enough to see its sides, wouldn’t the air pressing from the top mushroom out reducing column pressure and allowing the updraft to continue from the bottom?

    I’ve seen formation of cumulonimbus cloud where only the top billowed visibly, did the sides contract somewhat? It’s hard to say but mostly it was a powerful updraft.

    The vapor pressure of 30 C saturated air is o.6 psi. This appears to be enough for condensation to launch an air column a half kilometer without considering the thermal effects that work to increase the flow as you point out in 76. The problem is that it s a continuous process driven by an imbalance in humidity. As the air expands, continued condensation keeps the temperature up and the pressure low driving ever upward with more energy feeding in from below.

    The condensation itself is a wind driver and a powerful heat pump.

  95. Jeff Id,

    Aren’t you talking about the pressure change in the air based on the loss of kinetic heat?

    The cooling of the water vapor heats some of the air through conduction but cools the parcel through radiation. Thus we get a large pressure loss and high gradients?

  96. 116 Steve – I’ll have to leave sea ice for another day but I agree: there is lots of interesting stuff to discover! 🙂

  97. Re: Steve Fitzpatrick (Oct 20 18:34),

    I have done the measurement, and the pressure on the side of the container when dense particles are suspended really does rise more the the pressure you would expect from the fluid alone.

    But your system is at constant pressure, not constant volume. If you replace water with denser particles and keep the same total mass, the volume will drop and the average density will increase, increasing the pressure on the walls where the liquid makes contact. Conversely, if you keep the volume constant by adding the same volume of denser particles as the water you remove, the total mass as well as the density will go up, so of course the pressure increases. A sealed can is constant volume and constant mass. Condense some of the water and the pressure inside the can must drop once the can returns to ambient temperature whether the water is suspended or not because the average density doesn’t change. The transient behavior may be more complex at low mixing ratios. There may be an initial increase in temperature until the temperature of the gas reaches the level where condensation stops. That may cause the pressure to increase depending on the initial mixing ratio of water vapor to dry gas. Then further condensation will depend on the rate of heat transfer from the can to the surroundings. At very high mixing ratios and large delta T, the soda can full of steam placed in ice water for example, there will be no temperature or pressure increase, just a rapid decrease in pressure.

  98. Re: curious (Oct 20 18:07),Re: curious (Oct 20 18:07),

    “…air rising in a hurricane drops by 33% in degrees K. So WV is a small factor in contraction.”

    I haven’t reread the entire thread – I just went back to find this reference. From memory I don’t recall it being refuted.

    See the reply starting at comment #62. In order to lift a kilogram of air from the surface to 15.5 km, you have to do work, specifically 53.9 kJ. That work will come from the kinetic energy of the gas. So by the time you get to 15.5 km, the temperature difference will be a lot less than 100 K. There’s also the fact that the heat sink at 15.5 km isn’t infinitely fast, which also decreases the temperature difference.

  99. #119 Jeff Id
    “I’ve seen formation of cumulonimbus cloud where only the top billowed visibly, did the sides contract somewhat? It’s hard to say but mostly it was a powerful updraft.”

    Even with the depressurization effect described in the paper, cumulus clouds should still appear to grow horizontally. The apparent expansion of the visible cloud can occur within a contracting air mass. As surrounding air is drawn horizontally towards the site of initial condensation, this air too will be reduced in pressure causing water vapour within it to condense and become visible. Clouds should appear to grow in clumps.

    In the case of cumulonimbus the speed of air moving toward the site of initial condensation should be greater due to the greater amount of water vapour condensing. The air moving inward will acquire momentum, and even after equalizing pressure near the site of initial condensation, this momentum will persist. Trapped from below by rising air this momentum should translate into vertical motion. This may account for the rate of vertical development of cumulonimbus exceeding the rate of rise of the moist air mass from which it formed.

  100. 123, 124 – thanks DeWitt, I’ll have a read and also, following a more thorough read upthread, apologies to all for repetition. Reading back up I think the points I was interested in had been made – it was the thought experiment that caught my eye.

  101. As Vaclav Havel has just pointed out, this study showing the perfect match of the “Random Walk Model” to recent temperature and climate changes obviates the need to appeal to any other causality.
    “Fitting of Global Temperature Change from 1850 to 2009 Using Random Walk Model,” Guangxi Sciences, Vol. 17, No. 2, May 2010, pp. 148-150.

  102. If you adiabatically expand 1 m3 of dry air at 1 atm and 303.2 K to 0.5 atm, the volume won’t double because the temperature drops. The final temperature is ~249 K and the final volume ~1.64 m3. If the initial 1 m3 of air contains the saturation pressure of water vapor, the final volume and temperature will be higher because the of the latent heat released by the condensing water vapor and possibly the heat of fusion of water to ice if the temperature drops low enough. But it will still be at a lower temperature than 303.2 K and the volume will be less than 2 m3. Unfortunately, there are no handy little apps on the internet to do the moist air calculation like there are for dry air, or at least I haven’t found one.

  103. Curious #117,

    “Rereading your comments I see they could be interpreted to mean that the “drag” component is resolved via the floor rather that the walls”

    I think you are ready to understand the issue at hand. “resolved via the floor” has nothing to do with it. You need only stop trying to separate in your mind the macroscopic world from the microscopic world; it is (and must be) a continuum.

    The “drag” component of settling particles just means that settling particles are always supported by the fluid medium that surrounds them (Stokes’ Law)… via drag between the particles than the fluid. That drag MUST translate into lateral pressure as well as vertical pressure, even if the net movement of the particles is only in the vertical direction.

    Suspended particles always change the net density of the fluid they are suspended in no matter the size of the particles. Everybody understands that NaCl solutions in water are more dense than pure water, and that such solutions behave (as viewed by we macroscopic monkeys… as Jeff likes to say) as a continuous fluid, even though we know from having studied the kinetic theory of matter that there is no such thing as a continuous fluid.

    So the real question is: how are the dense <<0.5 nm diameter ions of sodium chloride suspended in water (AKA a salt solution) in any way different in how they influence the density of the "solution" than an equal weight of equally dense particles of 50 nm diameter suspended in water? Or 500nm particles? or 5,000 nm particles? If you measure the density of the salt solution and the particle dispersion (using a hydrometer, for example), the density measurements say the influence is exactly the same. If you put a pressure gauge on the side of a tank holding either the solution or the particle suspension, the pressure gauge says the effect is the same for both: pressure depends only on the net density of the entire fluid (liquid plus particles) independent of the size of what produces that net density.

    There is no difference in large scale behavior between a solution and a suspension…. we only impose a difference based on our macroscopic experience and prejudices; we think too much about the individual particles and not enough about how individual particles converge into the behavior that we see as a continuum. I hope this helps you.

  104. Gavin says : “Now think about what would happen as a cloud started to form – according to your suggestion this would cause instant horizontal convergence and the cloud would quickly be reduced in size.”

    Or alternatively the (moist) air that converges on the forming cloud helps create the cloud.

  105. DeWitt, #132,

    I am a chemist by training. I am familiar with potentially strong interactions between different molecules and ions, of course. The density of ionic solutions is not perfectly linear with concentration (for example). But that is not at all the concept Curious is struggling with.

  106. Re: Brian H (Oct 20 21:55),

    We’re back to unit root time series again. A random walk model is the same as a unit root or an AR(1) model with a coefficient of 1. It’s non-stationary. If the climate were truly a random walk then there would be no limits to temperature change. But it isn’t. You can’t have a unit root in a system with dissipation like the Sun/Earth/space. So finding an apparent unit root just means that the time series isn’t long enough to distinguish between a unit root series and a near unit root or other fractionally integrated series, which are stationary. That also means confidence limits calculated using an assumption of a unit root are too large. As usual, you are suffering from confirmation bias.

  107. 133-I don’t think that the Strong Interaction has anything to do with what you are talking about.

    Just a little physics humor to lighten the mood! 🙂

  108. I guess its time for a physical experiment. Fill a vacuum chamber with warm moist air. Depressurise the container until the adiabatic cooling causes the water vapour to start to condense. Lock off the vacuum line and observe the gauge. If pressure continues to fall after the vacuum line is closed then the effect described in the discussion paper would have a physical basis. To observe the effect the chamber may need to be quite large.

  109. Re: Konrad (Oct 20 22:53),

    You’re pumping stuff out. That isn’t adiabatic expansion. You need a cylinder with a piston. Then you move the piston to expand the volume. But there are problems. Surfaces are a big problem. They conduct heat and offer nucleation sites. If you expand rapidly, the pressure and temperature will follow the dry adiabat. That temperature will be too low and the eventual condensation will raise the temperature. What you will eventually get to is the moist adiabat conditions of temperature and pressure for the given amount of gas and water vapor and volume increase. The real question is how much work is done by a given adiabatic expansion of moist air and do the GCM’s calculate this correctly. If the volume change from condensation isn’t properly taken into consideration, the work done will be too high because the volume change will be too large. That would mean the final temperature would be too low. I think. Maybe.

  110. I would like to clarify the condensation/precipitation issue as related to our work. In the stationary pattern we are talking about, there is a balance between the ascending flux of vaporous moisture that condenses (can be for the present purpose approximated as w \rho_v, where w is vertical velocity and \rho_v is vapor density) and the descending flux of precipitating liquid moisture \rho_l w_l, where \rho_l is the density of liquid water in the atmosphere (kg/m^3) and w_l is the mean terminal velocity of droplets.

    The relation w \rho_v \sim \rho_l w_l simply says us that there is no liquid moisture accumulating in the atmosphere or, conversely, there is no outflux of liquid moisture from the atmosphere. From this relation it is easy to estimate how much liquid vapor actually remains in the atmosphere. Global mean vertical velocity is around 1 mm/s, while terminal velocity of rain droplets is of the order of 1 m/s. Therefore, as the air ascends, only a tiny fraction of condensing vapor, \rho_l/\rho_v \sim 10^{-3} remains in the atmosphere (on average). The rest is almost immediately precipitated. To my knowledge, there are many models where precipitation is assumed to be instantaneous, and this is, in the above sense, a safe assumption.

    This estimate is confirmed by direct observations of the liquid water content in the atmospheric column [called the liquid water path]. It is around 50 g/m^2 in the tropics (can vary broadly, of course, but the characteristic magnitude is that, see, e.g., Wood et al. (2002)). In the meantime, the precipitable water content (total amount of water in the column, vapor included) can be in the range of 150 KG/m2 (e.g., Duan et al. 1996). Note that the adiabatic liquid water content (i.e., the amount of liquid that should have remained in the column with ascending moist saturated air absent precipitation) is several times higher.

    Therefore, while the retention of liquid water in the atmosphere is certainly relevant for describing cloud microphysics, for the large-scale circulation problems that we discuss in the paper is physically reasonable and possible to use the notions of condensation and precipitation interchangeably.

  111. Re: DeWitt Payne (Oct 20 22:36),
    Oh, but there is no limit to the size of the excursions. It’s just that the more extreme ones are proportionately far less likely. In any case, the size and direction of the next few are predictable only at pure chance levels.

    And saying that the “series isn’t long enough” is just to say that there is no basis for rejecting the true null hypothesis: that there is no deviation from random walk behavior.

  112. I dont know whether the following is of any use but I thought it might give a feel to the magnitude of the effect.

    According to the Wiki, 505,000 cubic kms of water (rain) falls annually.
    This means that on average about 1384 cubic kms of water condenses each day.

    Using the simplistic figure of 18g of condensing water leaving a 22.4l “hole” in the atmosphere, we get about 1.7 million cubic kms of “hole” that the atmosphere has to fill by shifting around.

    To put that into some sort of perspective its air movement filling a “hole” maybe 1/3 the size of Australia and 1km deep every day and that sure feels like an effect that cant be ignored or simplistically parameterised to me.

    Of course I could be completely wrong with my calculations or logic.

  113. Steve, the pressure in a volume of gas depends on temperature at constant gas molecule number.

    When water vapor condenses to liquid and radiates away the heat of vaporization, the gas temperature doesn’t change, but the number of gas molecules decreases because water vapor is removed from the volume. At constant temperature, fewer molecules = lower pressure.

    The density of the volume doesn’t change, because it still has the same total mass — now composed of gas plus water droplets. But the pressure has diminished along with the fewer gas molecules.

  114. #140

    Therefore, while the retention of liquid water in the atmosphere is certainly relevant for describing cloud microphysics, for the large-scale circulation problems that we discuss in the paper is physically reasonable and possible to use the notions of condensation and precipitation interchangeably.

    I doubt very much whether conflating two very different processes in your terminology will be beneficial to a swift review of your paper. It has already lead to much confusion…

  115. #8 Derek

    The real question is, are clouds a natural shut off valve to convection.? Does the latent heat released by condensation reduce the temp gradient back to the earth’s surface sufficiently to shut off convection, or does the pressure decrease from condensation and heating (by latent heat release) more than counter act this. ?

    This is a very important point. The problem is that convection is currently described using the notion of the convectively available potential energy (CAPE). It is based on the following idea: we know how the temperature and density of an adiabatically ascending parcel change with height. If the ambient temperature decreases with height more rapidly, then the ascending parcel will always remain warmer and lighter than the surroundings, such that the ascending motion will continue.

    But, apparently, such motion changes the temperature gradient towards the moist adiabatic profile. So if there is no external forcing to sustain a supra-adiabatic lapse rate, convection will damp itself out. This negative feedback acting against the maintenance of convection is a big theoretical problem to which Derek referred.

    The situation is precisely the opposite with the condensation-induced dynamics (mass removal). Here, as far as the mass removal is proportional to vertical velocity, convective motion enhances itself in a positive feedback loop — as long as the incoming air is saturated with moisture. Mass removal is not affected by whether the lapse rate is exactly adiabatic, slightly lower or slightly higher: the convection will be sustained. It is one of the key differences of the proposed mechanism of convection as compared to those already considered in the meteorological theory.

  116. “Here, as far as the mass removal is proportional to vertical velocity, convective motion enhances itself in a positive feedback loop — as long as the incoming air is saturated with moisture. Mass removal is not affected by whether the lapse rate is exactly adiabatic, slightly lower or slightly higher: the convection will be sustained.”

    And the difference between theories is evident. A constant flow of energy in a thermal environment where energy was needed to continue the process.

  117. #144 Gavin,

    Thank you for your comments. My colleagues and I will certainly undertake efforts to clarify the issue. I plan to inform the handling editor of our paper about all discussions of our work in the web, for their evaluations to be as informed as possible. Rather than aiming at a swift review, we do indeed aim to clarify our message and find interested people among meteorologists capable of pursuing independent parallel research on this topic.

    I cannot agree though — unfortunately — that what has happened so far can be characterized as a confusion. Rather, this reflects the fact that the mass removal associated with condensation has not received any attention in the theory and is missing from textbooks. Accordingly, it is indeed true that many mainstream meteorologists are uncertain about the effects of condensation.

    For example, Dr. Peter Haynes, Co-Director of the University of Cambridge Centre for Atmospheric Science, is explicit in being unsure (my emphasis):

    But then a second shortcoming is that no careful justification of this drop in pressure
    is given. The implications of moisture and change of phase on thermodynamics have
    been considered carefully by physicists (including meteorologists) for two centuries or
    more and you would need to show either that the standard approaches (e.g. set out
    in textbooks such as ’Thermodynamics of Atmospheres and Oceans’ by Curry and
    Webster) imply a drop in pressure associated with condensation, or else where those
    standard approaches – including perhaps approximations that are usually made – are
    wrong.

    This is not a marginal view and it was expressed by Dr. Haynes after having spent months studying our paper. It is difficult to presume that during all this time no thought about precipitation has come to his mind. But precipitation is not the main issue here, as I said above.

    You say in #69 that

    “Given the hydrostatic approximation, condensation *per se* has no effect on air pressure (exactly as stated in Steve Fitzpatrick’s point above).”

    As you certainly agree, condensation turns some part of vapor to liquid. Liquid does not follow the same equation of state as gas (vapor). Therefore, there cannot be any realistic process where condensation would not change local gas pressure. Such processes do not exist.

    Condensation may not immediately change the pressure at the surface, this is true. Even if there were no droplets formed, it would take some (short) time for the air in the column to readjust to the new state of equilibrium. But condensation immediately lowers air pressure where it occurs and initiates this process of adjustment. The liquid water load can additionally impede the rising air flow and cause some delay in reaching the equilibrium. But in many cases (large drops) one would observe that the limiting time scale for hydrostatic adjustment is not the rainfall, but namely the time scale of the upward gaseous motion.

  118. To clarify in #147:

    Even if there were no droplets formed, it would take some (short) time for the air in the column to readjust to the new state of equilibrium.

    That is, even if upon condensation liquid water instantaneously disappeared from the column, it would anyway take some time for the air to re-adjust to the new equilibrium by upward motion.

  119. #89 Eric,

    One related question, if you know the answer: Is Anastassia’s statement that “On a grand scale, the problem with the current theory is, expectedly, that the realistic differential heating does not produce realistic wind velocities (the velocities are too low).” a fair statement (i.e., suggesting tha something is missing), or do you feel the velocities are realistic?

    #94, Gavin

    #89 I don’t see any evidence for this. Wind velocities in GCMs are close to observed, and even better in higher resolution weather forecast models. Possibly there is a noticeable difference in moving to non-hydrostatic equations, but I don’t know offhand of a paper demonstrating this.

    My point in #54 was to draw attention to the glaring mismatch between the level of performance of the theory and models. My reading is that in 1980 Held and Hou offered what could have become THE solution to circulation problem. They satisfactorily reproduced the length scale of the Hadley Cell starting from a few fundamental assumptions and basing, conventionally, on consideration of differential heating. However, the circulation intensity that the theory yielded proved to be one order of magnitude lower than in reality (see page 662 in Schneider 2006. As far as I understand, had the velocities coincided in that paper, one could celebrate solution of the general circulation problem. But they did not, and the solution, to a varying degree, remains elusive until now. To us it is not surprising, as we state that an essential process has been overlooked.

    In the meantime, I feel that already in the 1980s (and much earlier) GCMs were reported to be good enough to reproduce the velocity field. I am not aware of any published concerns that the models showed a one order of magnitude weaker circulation. From these observations one is led to conclude that GCMs apparently did not need the theory, their level of performance was reached by other means. At least it is my reading of the situation.

  120. Jeff Id #119

    I don’t think there is visible contraction on the sides of that cumulonimbus in the picture. The spreading at the top (anvil) is due to the Cb reaching the inversion at the tropopause, with some overshooting at the center.

    As for positive feedback when small clouds appear on a clear afternon sky or divergence when they dissapear (Gavin #114), I suspect the mass of water vapour involved is rather small, not comparable with the effect of condensation of a column of precipitable water several cm high, as in big storms and hurricanes. But even so, it should be worth the effort doing the numbers and see if an experiment to detect the change is feasible.

    After all, whatever the conclusion of this very interesting theoretical discussion, we should verify it experimentally.

  121. The title of this thread reminded me of a great Pink Floyd song that uses the phrase. Yes, I know this is the Phony Corporate Version Era of Pink Floyd, but it’s still a great song and sounds oddly appropos to the big picture. 😉

    One slip, and down the hole we fall
    It seems to take no time at all
    A momentary lapse of reason
    That binds a life for life
    A small regret, you won’t forget,
    There’ll be no sleep in here tonight
    – Pink Floyd, “One Slip”

    Andrew

  122. Gavin #82

    There is an approximation being made with this statement, that should not be forgotten and indeed does change its certitude.

    Water has dissolved gasses in it. The freezing process will retain these gases. During the entire melting process, roughly 9% of the ice will be exposed to the air. As the ice melts at this boundary these gases will escape to the air.

    Thus, the “mass of the ice melt” is going to be less than the mass of the displaced water.

    Yes, the amount of mass loss, is in most cases small and can be ignored. We know this from repeated testing and validation. However, knowing the approximation is being made is necessary and prevents surprises later.

  123. OK – I really messed up the blockqoute. Please delete or ignore #158

    Gavin #82 said

    “If the salinity is the same (as with a normal ice cube and freshwater), the melt has exactly the same density as the original water and so the mass displaced by the ice volume is exactly the same as the mass of the melt – thus no change in sea level. ”

    There is an approximation being made with this statement, that should not be forgotten and indeed does change its certitude.

    Water has dissolved gasses in it. The freezing process will retain these gases. During the entire melting process, roughly 9% of the ice will be exposed to the air. As the ice melts at this boundary these gases will escape to the air.

    Thus, the “mass of the ice melt” is going to be less than the mass of the displaced water.

    Yes, the amount of mass loss, is in most cases small and can be ignored. We know this from repeated testing and validation. However, knowing the approximation is being made is necessary and prevents surprises later.

  124. To the extent that it is relevant, cyclones over sea do not have the same moisture sources to sustain them as cyclones over land. The rain does not usually fall upwards.

    As I noted 2 years ago, cyclones can continue over land that is usually hot and often normally dry desert, for days. Here are two path maps to show this. Land distances can exceed 2,000 km.

    Even if a correction is made for pressure drop by condensation, the moving cell over land moves on to places with a lesser chance to source replacement moisture.

    So why does a cyclone not collapse sooner after precipitation over land?

  125. It seems to me that there is a real pressure drop that could be communicated to the surface provided that the amount of liquid water in the column is increasing.

    While droplets are being generated or growing the momentum of the liquid component will be increasing. This at least gives us mechanism for a pressure drop as we have a rate of increase in momentum per unit area due to the rate of condensation per unit area and the increase in the liquid phase velocity as the droplets grow.

    This is not the same case as the particles in suspension case, as mass and momentum of particles per unit volume in the suspension case is not prone to the spontaneous generation or growth of particles.

    If this be the case then a pressure drop could be communicated to the surface whilst condensation was occuring at least provided that the droplets have an absolute and increasing downward velocity. Conversly a pressure increase would occur whilst failing droplets were evaporating.

    Now it could be that increase in down momentum of the liquid phase would be balanced locally by an equivalent increase in upward momentum of the gas phase. In turn that might be communicated to the surface as a pressure wave of some sort.

    In general whereas the condensation does not affect the mass of the column it could affect its “weight” provided that the rate of condensation and acceleration of the droplets is not totally counterbalanced by an upward acceleration of the gas phase during condensation.

    I am not sure that actual precipitation is required nor am I sure whether this would be a first order effect, probably not as it would be driven by the rate of condensation not the amount of condensation.

    As soon as the rate of condensation returns to zero and the droplets reach terminal velocity the pressure communicated to the surface would be once again be just the column mass/area times gravity.

    Anyway it is just a thought and it does not consider many things that go on in a real atmosphere, but I thought I would say that I do not think that the suspended particle analogy is obviously appropriate in this case, due to the fact that condensation is by a generation and growth of particles (droplets) per unit volume which is not the true in the suspension case.

    Alex

  126. Actually, the article is about how clouds are the thermostat. They form as the day heats up in the tropics and reflect an increasing amount of light as the day wears on then they dissipate at night allowing more heat to escape. If you looked at Earth from the perspective of the Sun, the “afternoon” half of the Earth is cloudier than the “morning” half and the sunlit side is cloudier than the night side.

    The clouds even respond to the seasons following the angle of the sun as the ITCZ moves with the seasons.

    Cloud response acts as a thermostat. And large storms can dump tremendous amounts of heat well above most of the CO2 in the atmosphere (or most of the atmosphere, period, for that matter).

  127. Numerous hang gliders, paragliders and sail plane pilots have experienced ‘up close and personal’ the meteorological environment of which the physics is herein under discussion.

    Hang glider and paraglider pilots in particular are able to give a vivid and precise account of what happens through the transitions from wet thermal to suspended cloud to actual precipitation particularly in the cumulus to cumulus-nimbus sequence. As a former hang glider I ‘went there’ a number of times in my younger days.

    Paraglider pilots in particular whose wings have poor best glide speeds have had amazing experiences, in some cases resulting in prolonged unconsciousness and even death.

    To me there appears to be merit in the views of Makarieva et al. but I am too unfamiliar with the math to comment mathematically.

    However, it would perhaps be useful to hear others with at least a scientific training and extensive gliding experience express their actual hands on-based opinion on this matter. After all, it was only by going to the Moon that we even began to correctly understand the whole daily temperature cycle there.

  128. Hang gliding under over-developed cumulas cloud can be delightful as lift is light, smooth and widespread. It is called ‘cloud suck’.

    However, from onset of rain there is invariably significant loss of lift, a likely but not invariable onset of cross winds and in extreme cases downdrafts manifested as increased turbulence. The rule of thumb is – get outta there (or land if too low).

    Note most hang gliders fly with basic instruments – almost invariably at least a variometer and usually an air speed indicator and altimeter so the above is a lot more than subjective.

    Herer is the only YouTube example I could find – very mild example due to orographic effect location (nice music too) but obvious loss of lift even so.

    http://www.youtube.com/watch?v=uw6pwzPNrDk

  129. hey there and thank you for your information – I’ve certainly picked up something new from right here.
    I did however expertise some technical issues using this web site, since I experienced
    to reload the site a lot of times previous to I could get it to load properly.
    I had been wondering if your web hosting is OK?
    Not that I am complaining, but slow loading instances times will sometimes affect your placement in
    google and can damage your quality score if advertising
    and marketing with Adwords. Anyway I’m adding this RSS to my e-mail
    and can look out for much more of your respective interesting content.
    Make sure you update this again very soon.

  130. Worldwide national academies of sciences were united on 24 Oct 1945 when nations were united in Stalin’s effort to expand the boarders of the old USSR to engulf the globe.

    ResearchGate is now starting to break NAS’s totalitarian grip on science and research journals in the west.

  131. Sweet site! I identified it although browsing on Yahoo Information. Do you have any ideas on how to get listed in Yahoo Information? I’ve been trying for a even though but I never ever seem to be to get there! Thanks

  132. “The conventional wisdom is that in the real atmosphere, the condensation heat causes
    expansion of the air, hence a lowering of the specific density, hence a lesser weight of
    the air column”. My feeble experiment to decide which side of the fence I sit on is the “cloud in the bottle” experiment. Squeeze the bottle and the cloud disappears, release the pressure and the cloud reappears.
    So, my take on that is that heat released within the cloud (causing more pressure) would make the cloud disappear!
    So the conventional wisdom cannot be correct! I believe that people are looking at this all wrong. First of all energy isn’t always released as heat. This is lost on so many people. A cloud is a mixture of air, water vapor and water droplets. The droplets are falling and evaporating as they fall, while the lighter than air water vapor is generally rising and condensing into new droplets as it rises. So we are getting a neat transfer of energy as droplets condense (releasing energy) and droplets evaporate (absorbing energy) all through the cloud. This has the effect of shunting lighter water vapor down past heavier air particles going UP all through the cloud. This shunting is the pump that drives air up through the cloud. This efficient “shunting steam engine” explains why clouds weighing as much as a hundred elephants are suspended in the air. Not only are they suspended but they are actually “sucking air” up through the clouds! They will keep sucking until the “condensation power source” is exhausted. I would love to see answers to this post. Thanks Brian.

Leave a reply to DeWitt Payne Cancel reply