Penetrating Radiation

I notice that there has been interest recently in the question of the difference between (shortwave) solar radiation and (longwave) infrared radiation as it affects ocean heating. Water is essentially opaque to infrared radiation, while shortwave radiation (especially in the blue wavelengths) can penetrate to substantial depths. realclimate takes the position that this doesn’t matter. I get the impression that GCMs, used in IPCC climate modeling, treat both shortwave and longwave radiation identically. I don’t propose to comment on these matters, but I happen to have read some interesting specialist articles on the topic, which are web available. It doesn’t look to me like the realclimate position on this matter is a gimme. There are some interesting connections between biological (phytoplankton) activity and penetration. Anyway here are a few articles that I’ve seen – no representations as to completeness – and which are available on the Internet. First, a really nice survey: Bissett et al. (2001), Resolving the Impacts and Feedback of Ocean Optics on Upper Ocean Ecology, Oceanography, 14(4) URL Manizza et al. (2004). Modelling the impact of phytoplankton on upper ocean physics on global scale, GRL. URL1 URL2

Marine phytoplankton partly regulate Earth’s temperature and climate through feedbacks whose full understanding is still under investigation [Gildor and Follows, 2002; Boyd and Doney, 2003]. They alter the cycle of elements such carbon and sulphur [Watson and Liss, 1998] which in turn modifies the radiative forcing of the atmosphere.

Sweeney et al., Impacts of shortwave penetration depth on large-scale ocean circulation and heat transport URL

While many one-dimensional studies of the mixed layer have considered the importance of heating due to VIS at depth (i.e. Denman, 1973; Simpson and Dickey, 1981; Dickey and Simpson, 1983; Lewis et al., 1983; Woods and Onken, 1982; Woods et al., 1984; Martin, 1985; Siegel et al., 1999; Siegel et al., 1995; Ohlmann et al., 1996; Ohlmann et al., 1998), the climate modeling community has been slow to implement these parameterizations in OGCMs. As a first step, most models assume that all of the solar irradiance is absorbed at the surface in the same way that latent and sensible heat are passed across the air-sea interface.

Thus, some of the systematic deficiencies in the present-day climate models, such as the colder than observed cold tongue in the equatorial Pacific may simply be related to inaccurate representation of the penetrative radiation and can be improved by the formulation presented here.

There are many references to articles by Murtugudde. Here is one reference: Should Climate Models Account for Biological Feedbacks? URL A little googling or following the references turns up more. The topic is rather a large one and the specialist literature doesn’t seem to support the realclimate view that it doesn’t matter, but that’s just an impression.


  1. Posted May 23, 2005 at 6:25 AM | Permalink

    The statement that the wavelength dependence of absorption does not matter is laughable. The oceans cover about 3/4 of the planetary surface. Roughly 8 percent of the radiation is reflected by the oceans (albedo), and the percentage is again wavelength dependent. Phytoplankton – tiny form of life in the ocean – is an important contributor to the absorption.

    Also, phytoplankton is killed by the ultraviolet rays: A reduced amount of phytoplankton, for example, is believed to reduce absorption and lead to reduced warming – see Frouin and Iacobellis at (phytoplankton plays big role for Earth’s temperature) – although the sign seems to be a controversial topic, too.

    The purely physical effects described by Sweeney are another example. The heat transport by the oceans is definitely affected by the temperature dependence on the depth, and these things are affected by the spectrum of incoming radiation, too. And they show examples where one gets wrong signs – the dependence of warming trend on the depth of penetration comes out with opposite sign than if the wavelength dependence is included. They complain that although these important effects have been known for a long time, the climate modellers fail to include them.

    Any of these contributions to the absorption – and their trends (which may be caused by hundreds of known and unknown factors) – is a priori larger than the absorption of the IR rays by the atmospheric carbon dioxide. The statement that “it does not matter” really means “we’re cheating everyone, and please shut your mouth otherwise you help the anticommunist heretics”.

  2. Posted May 23, 2005 at 7:19 AM | Permalink

    One more comment: it is apparently a prevailing opinion that the “cold tongue problem” – the fact that virtually all currently existing models predict colder Western Pacific than observed – is due to insufficient vertical resolution of all these models. The inability to include a realistic depth of the water column that absorbs the IR light and – separately – the UV light is an example of this kind of problem. Let me quote the page I linked:

    “Common model problems arise from the inadequate vertical resolution especially in the upper 300 m, e.g., problems of poor meridional heat transport, poor east-west slope of the equatorial thermocline, poor equatorial thermocline sharpness, poor structure of thermostad below the equatorial thermocline in east Pacific, and poor equatorial SST cold tongue all arise from inadequate vertical resolution.”

    In the models, one obtains the cold tongue problem because of strong winds – however, these winds agree with observations and therefore the incorrect final prediction for the temperature must be a real defficiency of the model. It fun to look how the people try to adjust the models and nothing works: the usual suspects are the clouds whose effect is understood extremely poorly in all these models as even their most faithful fans know very well. So they try to play with the “stratocumulus” clouds, replace their description by a different framework, and the results are still very poor. In a paper that shows evidence that the 11-year solar cycle has a significant influence on stratospheric temperatures, other discrepancies of similar kind are mentioned.

  3. Posted May 23, 2005 at 10:18 AM | Permalink

    Of interest is the diurnal cycle of skin temperature vs. sub-surface temperature. During a sunny day, the skin (the upper less than 1 mm) is much warmer than just below the surface. Which points to solar IR absorbed at the surface. This has a huge impact on water vapour and (sea) surface air temperature. Wind speed is another factor. See for some background:

    • Willis Eschenbach
      Posted Apr 18, 2009 at 4:35 PM | Permalink

      Re: Ferdinand Engelbeen (#3), you point out the diurnal changes, viz:

      Of interest is the diurnal cycle of skin temperature vs. sub-surface temperature. During a sunny day, the skin (the upper less than 1 mm) is much warmer than just below the surface. Which points to solar[sic] IR absorbed at the surface. This has a huge impact on water vapour and (sea) surface air temperature.

      You are correct, and there is more to the phenomenon. This is because (in contrast to the atmosphere) the ocean is stable during the day and unstable at night.

      During the day, warm water (whether warmed by visible or IR) rises to the top and stays there. At night, the surface water radiates and evaporates, cools, and sinks. This is the main (although not the only) driver of vertical circulation in the upper mixed layer of the ocean (usually ~ 100m).

      During the day, then, IR strengthens the vertical thermal stratification in the upper ocean. It leads to higher daytime ocean skin temperatures, particularly in the critical free surface temperature. This leads to increased losses through radiation (radiation increases by T^4). It leads to increased losses through convection (through greater ocean/atmosphere temperature differences and greater instability). It leads to greater evaporative losses (through Clausius-Clapeyron [exponential with T] plus greater wind-driven evaporation [linear with wind speed]).

      This means that during the day, the bulk temperature of the mixed layer is immaterial to the rates of radiation, convection, and evaporation. All that matters is the skin temperature. I live, surf, and dive in the deep tropical Pacific. At certain times when the wind and sea are calm, I can be swimming in toasty warm water. But with each stroke of my arms as I paddle, my hands plunge through the surface warmth into much cooler water a half metre down. And during those times, I can feel the extra warmth in the top centimetre. (I imagine I can feel the yet higher free surface temperature, too … but probably that’s just me.)

      This means that during the day much more of the absorbed IR is immediately lost, compared to the the visible radiation. Visible radiation is absorbed at depth. It heats the bulk of the mixed layer.

      By contrast, IR is absorbed at the very surface. In the tropics much of that IR energy just breaks water molecules loose from the surface. In a funny way, this absorbed radiation should never even be counted in the oceanic energy budget. It is absorbed by a water molecule and breaks it free from the ocean. The energy never heats the ocean in the slightest.

      The net daytime result is to ensure that the majority of absorbed IR is immediately returned to the atmosphere.

      At night in the ocean, a curious thing happens. It’s kind of an inverse of the way that cumulus clouds form during the day. As daylight fades, at some point the ocean stops being a net absorber of energy. It begins to lose more energy than it is receiving. The surface starts to cool. Due to local inequalities, sinking cells form at intervals. These are akin to the thunderstorms, in that they set up cellular circulation. Cool surface water moves laterally and sinks in vertical columns of cooler water. This is similar to how during the day warm surface air moves laterally and rises in vertical columns of warmer air to begin the day’s mixing.

      As occurs during the day, this leads to larger areas of gradually rising water at nighttime, with smaller areas of more rapidly sinking water scattered among them.

      During all of this, of course, the free surface of the night-time tropical ocean is both gaining and losing heat through radiation, conduction/convection, and evaporation/condensation. Stronger nighttime IR slows the onset of the night-time circulation. This leaves warmer water on the surface longer, allowing more losses of all types. Stronger nighttime IR also slows the speed of the circulation. This reduces the rate at which cooler water arrives at the surface.

      Finally, the tropical cumulus and cumulonimbus (thunderstorm) clouds are much more prevalent during the daytime than the nighttime. By dawn, it is usually clear. Since clouds are the most effective “greenhouse gas” (~ 100% IR absorptivity/emissivity), this leads to much lower downwelling IR levels at night than during the day. Using averages in this regard is very misleading.

      What all of that means to me is that the effect of e.g. a 1 w/m2 change in visible energy will be different from the same change in IR.

      Given the complexity of all of that, I was very interested in Gavin saying above that:

      In a response to one comment, I pointed out that the difference between LW and SW heating of the mixed layer was not a big factor in trying to explain the recent changes in ocean heat content.

      Gavin (or anyone), a citation to the comment in question would be great. My take on it is that the situation (ocean/atmosphere interface as affected by day/night/clouds/IR/visible) is too complex and difficult to model for it to be represented in anything like the necessary complexity. I suspect that Gavin might be right, but I’m curious what he’s basing his opinion on.

  4. Dave Dardinger
    Posted May 23, 2005 at 12:54 PM | Permalink

    Re #3 Thank you for the link. I’d wanted to see what the structure of SSTs with depths looked like. But are you sure it’s solar IR which results in the warming rather than an increase in back-radiation from a warmed atmosphere? I didn’t think the sun had much IR. Or does it depend on just what frequencies we’re talking about? It looks to me from the drawing of daylight temperatures that the heating occurs pretty largely down to a meter or so, which would probably include some radiation from the red end of the visible spectrum as well as near IR.

  5. Roger Bell
    Posted May 23, 2005 at 7:50 PM | Permalink

    The previous discussion prompts me to ask a question that has puzzled me for a while, particularly concerning the recent paper by Hansen et al. The ellipticity of the earth’s orbit leads to it being about 3% closer to the earth in the northern winter than in the southern winter. So the energy received at some spot on the earth , say at 40 degrees North in the northern winter, is 6% greater than that received at the corresponding spot at 40 degrees South in the southern winter. If this is combined with the fact that much more of the Southern hemisphere is covered with water than is the case in the Northern hemisphere, is it necessary to carry out a much more detailed calculation than Hansen did?

  6. Posted May 24, 2005 at 2:46 AM | Permalink

    In a response to one comment, I pointed out that the difference between LW and SW heating of the mixed layer was not a big factor in trying to explain the recent changes in ocean heat content. How this can be understood to imply that it doesn’t matter in any respect is beyond me. Differences in the physics of SW and LW heating in the ocean are important and are treated consistently in all current climate models.

  7. Posted May 24, 2005 at 5:22 AM | Permalink

    Dear Roger, the numbers you mention are correct, indeed. The eccentricity of the Earth’s orbit is about 0.0165 which means a 3.3% difference between the minimal and maximal distances. Because the amount of radiation decreases as the squared distance, it amounts to a 6.6% difference between the maximal and minimal amount of solar radiation. This is a very significant difference and the typical data are always averaged over the year period.

    You cannot neglect 6% errors if you want to make decent calculations. The average inflow of energy to the Earth is about 342 Watts and 6 percent of this amount is almost 20 Watts. Hansen claims that he can determine the “imbalance” (0.8 Watts per meter squared in his case) with accuracy of 0.15 Watts which is 200 times smaller. So be sure that he should rather be careful about the difference between the average distance from the Sun and the maximal distance, for example. There are other reasons why his advertised error margin of 0.15 Watts per meter squared is unrealistic.

    One more thing. Imagine that the Earth is a blackbody and you increase the solar radiation by 6 percent. What would it mean for the temperature of the Earth? The radiation emitted by a blackbody – which should agree with the absorbed solar energy – goes like “T to the fourth” where “T” is the temperature in Kelvins. This means that a 6% increase of the radiation means a 1.5% (because of the 4th power) increase of the temperature in Kelvins. Because the room temperature is about 300 Kelvins, 1.5% of this amount is roughly 5 degrees of Kelvin. I suppose that this is too large an error to be neglected.

    This was just a very rough global counting. If you want to calculate how various changes – and eccentricity – affects local phenomena such as the Gulf stream, these variations become even more important. If you focus on long-term questions, you are averaging over the year and the eccentricity becomes less important.

  8. Steve McIntyre
    Posted May 24, 2005 at 7:41 AM | Permalink

    Re #6 from Gavin (presumably Schmidt of realclimate): In the CCM documentation at section 6.2, there is an option discussing penetrative radiation. Gavin said that penetrative radiation is handled consistently in "all" models. This claim seems hard to reconcile with Sweeney’s seemingly opposite comments. It would be nice to know whether Sweeney agreed with Gavin’s comments. It would also be interesting for someone to look through the CCM documentation on this topic and see how it works. I would like to parse through some of the climate models, but am obviously otherwise occupied right now. Steve

  9. Posted May 24, 2005 at 4:40 PM | Permalink

    In response to Dave #4:
    Near halve of the sun’s radiation energy is in the infrared (think of a burning glass!), see:
    The absorption coefficients for (pure) water are at:
    Seawater might be different in the visible spectrum, due to the presence of plankton, but see for seawater (no data):

  10. Dave Dardinger
    Posted May 25, 2005 at 9:02 AM | Permalink

    Re #9 Yes I’ve been doing some reading since posting that. I like the Oceanography textbook in your middle link. I’ve been wanting to read a book now that it’s becoming clearer that there’s a lot of very complex things happening. (Certainly more than are being considered in global climate models, as can be seen by looking at Steve’s link to the ccsm3 model manual.)

  11. Roger Bell
    Posted May 26, 2005 at 7:19 AM | Permalink

    There’s a useful, if small, diagram of the solar flux given as Fig 1 in the paper by Bissett et al referred to above. The absorption bands that can be seen at wavelengths of 0.77, 0.9,1.1,1.4 and 1.8 microns are caused by absorption in the earth’s atmosphere, not the Sun’s. Perhaps climate modeller’s have used the data we have with much better spectral resolution to check their calculations of telluric absorption – they jolly well should have done. The diagram is a bit misleading, since we can observe the solar spectrum from the ground down to below 0.3 microns. The overall smoothness of the solar flux is a bit misleading, since there are an enormous number of absorption lines in the spectrum, particularly at shorter wavelengths.

  12. TCO
    Posted Sep 18, 2005 at 9:16 PM | Permalink

    Gavin came here, but didn’t finish the discussion.

  13. Posted Nov 8, 2005 at 9:28 PM | Permalink

    The emissivity question is one that is not easily modeled. Radiation can be absorbed at one frequency and emitted at a second – and an absorbing or radiating molecule can also exchange this same energy via conduction. Coupling this with precipitation of water vapor and changes due to any number of other factors gives us a quagmire from which one might believe most anything.

    I think the climate models are meaningless: I have worked with circuit simulators in electronic engineering. These simulations of a simple closed system often dramatically fail to represent reality – and I’ve fooled my self by fudging the component models to make things work as expected. Why do we expect meaning from such a simulation?

    To jump from such a simple model to one of an open system – using estimates with unknown error bands expands the old saying of Garbage in – garbage out – to garbage in – mixed in garbage yields garbage out. I suppose skeptics must endeavor to point out the folly of the pursuit – but is it that difficult to see that climate change is unknowable?

    There is a famous saying in physics:
    “Give me four parameters and I can fit an elephant. Give me five and I can wag its tail”

    (The source of the above quote?? Variants of the statement have been attributed to C.F. Gauss, Niels Bohr, Lord Kelvin, Enrico Fermi.)

    The next line is from a biology paper unrelated to climate:

    When one considers that these models may have parameters that number in the tens to hundreds and are only growing in size, the possibility of generating meaningful computer models is a fantasy.

  14. John Olson
    Posted Apr 17, 2009 at 2:04 PM | Permalink

    In response to Roger Bell, not only is there an observable difference in hemispheric solar radiation, but the nature of elliptical orbits is such that transit velocity is greater near perihelion than near aphelion. This results in one hemisphere currently receiving about 5 days less winter and the other hemisphere receiving about 5 days less summer.

%d bloggers like this: