Urban Wet Island?

The June 2010 issue of BAMS (Bulletin of the American Meteorological Society) contains the State of the Climate in 2009 report (Arndt, D. S., M. O. Baringer, and M. R. Johnson, Eds., 2010: State of the Climate in 2009. Bull. Amer. Meteor. Soc., 91 (6), S1-S224). One feature of the report is the display of key climate indicators, each being a time series which is unambiguously expected to be going in some particular direction due to global warming. You can view graphs and download the data here.

One of the interesting graphs is multiple estimates of specific humidity (the amount of water vapor in the atmosphere). In a warming world, specific humidity is expected to increase because a warmer atmophere will naturally hold more water vapor. Because water vapor is also a greenhouse gas (in fact, its the most prominent one in our atmosphere) this leads to one of the expected feedback mechanisms of man-made global warming. As the globe warms due to CO2 (and other anthropogenic greenhouse gases) more water vapor enters the atmosphere, which enhances greenhouse-gas warming due to water vapor, therefore amplifying the CO2-induced warming.

Three data sets of specific humidity (in grams H2O per kg air) are available. The first is from Dai 2006, Recent Climatology, Variability, and Trends in Global Surface Humidity, J. Climate, 19, 3589-3606, the second from Willett et al. 2008, Recent changes in surface humidity: development of the HadCRUH dataset, J. Clim..21, 5364:5383, the third from Berry & Kent, 2009, A New Air-Sea Interaction Gridded Dataset from ICOADS with Uncertainty Estimates, Bulletin of the American Meteorological Society, 90(5), 645-656 (DOI: 10.1175/2008BAMS2639.1), which is data for the marine environment only (which nonetheless is most of the surface of the earth). Plotted together (as specific humidity anomaly) they look like this:

All three data sets show increasing atmospheric humidity. The Dai (2006) data indicate increase at 0.06 g/kg/decade, the Willett et al. (2008) data indicate increase at 0.08 g/kg/decade, and the Berry & Kent data indicate increase at 0.07 g/kg/decade. These results are consistent with what’s expected given the observed global temperature increase over the same time span.

In fact the changes in humidity track the changes in global temperature surprisingly well, as can be seen by superimposing a scaled verstion of global temperature anomaly from GISS, shown here in black:

Not only is global humidity rising with the warming trend as expected, even its year-to-year fluctuations match well with those of global temperature. This is extremely strong evidence that, just as expected both from computer models and from basic physics, the dominant factor in global humidity is global temperature.

It also raises a question for those who doubt the correctness of observed global temperature increase: if (as so many denialists claim) the globe isn’t warming because the global temperature estimates are wrong, then why does the specific humidity track it so well?

Any bets on how long it takes someone to suggest that the increase in global humidity isn’t real — that the data are wrong and the apparent increase is only due to the “urban wet island” effect?

50 responses to “Urban Wet Island?

  1. Watching the Deniers

    Humidity is rising, as we’ve known for some time.

    The deniers are treading the path of the creationists: in order to dismiss climate science they need to dismiss the entire discipline if science and declare it corrupt.

  2. It’s one of those painfully obvious thing (or at least, painfully obvious when someone points it out).

    It’s really unanswerable, which is why the deniers will produce that problem-solving hammer, “correlation does not imply causation”. To which I respond *headdesk* *headdesk* *headdesk*

  3. Can surface humidity measurements be used to infer water vapor changes throughout the whole atmosphere?

    • Gavin's Pussycat

      Something is said about this in Dai (2006):

      The surface observations provide much better spatial and temporal sampling than radiosondes. Because water vapor in the lower troposphere is generally well mixed and the lower troposphere accounts for the majority of atmospheric total water content, variations in surface q, especially on weekly and longer time scales, are found to strongly correlated with atmospheric PW [precipitable water, GP] (Reitan 1963; Bolsenga 1965; Liu 1986; Liu et al. 1991).

      In fact, Smith (1966) derived a power relationship (over both land and oceans) between surface specific humidity qo at pressure level po and atmospheric q at pressure level p: q/qo = ( p/po)^lambda, where parameter lambda varies with latitudes and seasons within a range of about 1.11-3.50 (Smith 1966; Liu et al. 1991). These studies suggest that historical records of surface q, which has near-global sampling, can provide useful information on changes in atmospheric water vapor content, especially over the oceans where upper-air soundings have been unavailable.

      Executive summary: yes, so it would appear. Of course if climatic regimes start going walkabout, the global map of lambda might change. But I would think that is a second-order effect.

  4. This is a great post and I would love to see Watt’s reaction to this post!

  5. Another Note is that this method could also help to verify the results of different temperatures reconstructions. The thing that it looks like to me is that CRU would match the graph better because of its 1998

  6. “Any bets on how long it takes someone to suggest that the increase in global humidity isn’t real — that the data are wrong…”

    Or how long it takes for someone to take that position while simultaneously using exact same data to support the claim the world stopped getting damper in 1998?

  7. Michael Hauber

    Interesting, but the short term changes in humidity seem to be stronger for a given temperature change then the long term changes.

    I suspect this is tied in with ENSO, and that ENSO changes affect humidity more strongly than the long term AGW warming trend.

  8. David B. Benson

    Aha. I went back to reread
    to refresh about all this.

    Thanks, Tamino!

  9. Slightly related to this topic why has the US cooled as the globe has warmed?

    [Response: It hasn’t.]

  10. What is/are the actual physical process(es) by which specific humidity is measured? (Ideally, there’s more than one method and the three data sets were compiled with different methods, or multiple methods each … but one method would be fine if we’re really confident in it.)

  11. Thanks Tamino.

  12. Tamino, some research that I have seen suggests that urban areas, in the summer and during the day time at least, to be drier than the surrounding countryside (as quantified by the mixing ratio). That is they are warm, dry islands….

    So, if anything, long-term increases in specific humidity (q) from surface stations located in urban areas will likely be underestimating the observed long-term increase in q.

  13. Knowing the deniers a bit, there are plenty of excuses they could come up with. Expect things like:
    1. Yeah, so it’s warming, so of course water vapor increases. Doesn’t say anything about the cause though.

    2. Ah! So you admit it’s water vapor that governs global temperatures!

    3. THE EARTH IS COOLING! Just look how water vapor has decreased since 1998!

    Feel free to add more…

  14. “Because water vapor is also a greenhouse gas (in fact, its the most prominent one in our atmosphere) this leads to one of the expected feedback mechanisms of man-made global warming.”

    One question here: Even though I know water is the most prominent greenhouse gas, I thought the radiation bands that it absorbed were pretty much saturated so any increase in humidity would not lead to any big increase in warming? Thanks for any help.

    • Gavin's Pussycat

      It’s complicated… I suggest playing with Dave Archer’s simulator.

      More seriously, your question is based on a misconception: saturation doesn’t matter. What matters is: for which part of the spectrum H2O blocks outgoing radiation from the Earth surface. You see, the greenhouse effect works in such a way, that increased CO2 leads to the “radiating surface” for the wave numbers where CO2 is semi-opaque (the “wings” of the band) , moving up to higher levels in the troposphere, where temperatures are lower. Less radiation escapes to space, which must be compensated by surface temperatures — and the whole atmospheric temperature profile, close to the adiabatic lapse rate — shifting up.

      Now, for those parts of the spectrum where H2O is opaque, this won’t help: the radiation escapes not from the surface, but from the “top of the wet troposphere” where humidity drops low enough for H2O to become transparent, a few km up. As tropospheric temps go up, this level will move up too — precisely to the level where the temperature is the same as on the old level before the change. The amount of outgoing radiation in this part of the spectrum does not change.

      This means that the remaining part of the spectrum where H2O is transparent for surface radiation has to do all the work: temps have to rise even more than without the H2O. This is basically the way in which water vapour “amplifies” CO2: the amplification factor is the inverse of the fraction of the Planck curve where water vapour is transparent to surface IR radiation.

      One can use Tyndall’s “barrier in a stream” metaphor: increasing CO2 is like making the barrier higher, so that the level of the water behind it has to go up in order to re-establish the same run-off. Increasing H2O works then like making the barrier narrower by completely blocking run-off over part of its width. It leads to a further increase of the water level behind it.

  15. Ah, thanks for unearthing this, Tamino. It would be interesting to see this framed in the context of extreme precipitation events too.

    Cheers – John

  16. Gaz, surely the denier response is going to be that relative humidity leads temperature. ( Just don’t ask them for a mechanism.)

  17. Do the changes in specific humidity correlate with

    a) changes in global rainfall
    b) global cloud cover?

    over the short-term and longer-term?

    Do we know?

  18. DreamQuestor

    Just out of curiosity, does the concept of thermal expansion also apply to water vapor in the atmosphere? Is it known approximately or exactly how much the thermal expansion of the ocean is contributing to the increase of water vapor in the atmosphere?

  19. gss_000: The whole “saturated absorption bands” argument is a myth. But it’s not an easy subject to grasp, and the sources I trust don’t discuss it much.

    This is the best answer I could find: http://ca.answers.yahoo.com/question/index?qid=20100202003034AAcq3ji – it’s discussing CO2, but the logic applies equally to water vapour.

    • Thanks! I’m kind of shocked how wrong I was, since I got that explanation in grad school (albeit not from a dedicated climatologist). Still, thanks for the education.

  20. I’ve been wondering about this for a while now, ,so thanks Tamino for doing the legwork. It’s an elegant result.

    Of course, it won’t satisfy the likes of Watts’ and Codling’s minions who promulgate the nonsense that there has been no increase in atmospheric moisture. It will, however, be interesting to see how they now go about defending their claims…

    I suspect that you can expect some trolls landing here sooner or later to do exactly that.

    • On their behalf, how can I resist!

      As shown here (see the 7th bullet under ‘consider’ and no. 7 under ‘references’, the dataset they use to defend their claims is this one. Choose variable = specific humidity, analysis level = 400 mb (try lower ones, too), lat & long for global coverage (presuming you believe the earth isn’t flat).

      Comments different from ours and references different from Tamino’s on a postcard, please.

  21. Egil Heine Strand

    On-Topic, but maybe spin off material:
    I’ve heard/read sceptics arguing for Anthropogenic Water Wapor as the cause for warming – the notion is that Water Wapor being so powerful greenhouse gas and humans producing water wapor by combustion of fossil fuels and agriculture has increased the content of the atmospheric H2O and this being the culprit of increasing temperature, not CO2.
    My logic tells me that this is wrong, but i suffer from the lack of data regarding this, so my question is if anybody can tell me, preferrably with sources:
    – What is the increase in content of water wapor in the atmosphere over the same period we have been pumping CO2 into the air?
    – How big is the direct human contribution to the amount of water wapor in the air?

    (Imply the knowledge of that water wapor will not accumulate as CO2 does – I and my “opponents” know that)

  22. EHS,

    Extra water vapor stays in the air an average of 9 days. Extra CO2, 200 years. We could double water vapor tomorrow, and in a few weeks all the excess would rain out.

  23. Egil Heine Strand


    Yeah – I know that, as stated in the end of my post – but the questions?

  24. Egil, the extant data toward answering your first question is linked in the original post. Shortcutting it a bit for you:


    For your second question, all I have is intuition, I’m afraid–but FWIW, what my intuition says is that considering the total surface area of the planet composed of water, much of which is subjected to frequent strong winds and sunshine, probably the human contribution forms a pretty damn small proportion.

    Of course, the human contribution to CO2 flux is pretty small, too, but there the long residence time compensates.

    • The amount of water produced by combustion is minute compared with the water already in the air, let alone in the oceans, etc. In any case, you can’t just add water to the air and expect it to stay there. You have to increase the temperature first or it just condenses out.

      Does anyone have any figures for water in the air (vapour plus clouds)?

  25. EHS,

    Unless I’m mistaken, Tamino’s first graph in this thread gives a partial answer to your first question. We have been pumping CO2 into the air since the mid-1800’s, but Tamino’s graph starts at about 1970. I don’t know if humidity data for previous years is available.

    I’m not sure just what you want as an answer for your second question. A balanced chemical reaction for the combustion of alkanes (such as natural gas or gasoline) shows that there will be one more water molecule than CO2 molecules produced. (For example, combustion of one molecule of octane produces 8 molecules of CO2 and 9 molecules of H2O.) Of course, fuels such as coal are not alkanes, so the amount of water they release follows a different formula.

    Of course, as Tamino has shown, adding CO2 to the atmosphere causes the earth to warm, which causes water to evaporate, which makes it a human contribution to the water vapor in the atmosphere.

    • Egil Heine Strand

      Thanks for responses.

      Just so I’m shure I’ve got this right:
      Our contribution to atmospheric water wapor will be negligible due to the physics that gives humidity as a function of temperature and that the air will strive for an equillibrium state according to the temperature at any given time so whatever H2O we put into the air will fall out as extra precipitation as to get “rid off” the H2O that is too much and as a consequence we will not directly alter the natural humidity state.

      Only a pulse or step up of extra water wapor will be noteable, but only for the time it takes for the system to reach equillibrium again – IE 9 days.

      That right?

      • Kooiti Masuda

        I think that the number “9 days” is the mean residence time of water vapor, that is the ratio of the total amount of water vapor in air and the rate of exchange at the surface. The time length which a molecule experiences from entering the atmosphere to the exiting is variable, but the average of it can be given by the simple calculation. The corresponding value for CO2 is about 3 years. I had an endless discourse with a person who insisted that anthropogenic CO2 in the atmosphere would be lost at this time scale so that it cannot be the principal cause of the observed increase of concentration.

        Here we need a time scale of decay of anomalous concentrations. The number “200 years” mentioned by BPL is an estimate of this for CO2. (Actually it is complicated, so IPCC AR4 did not show a single number. But I think BPL’s number is right as a rule of thumb.) I feel obliged to show the value for water vapor, but it is not so easy. If any scientist has made a quantitative estimate I would like to learn. I guess that “9 days” is right order-of-magnitude, though.

      • Re water vapour. See Gavin’s Water vapour: feedback or forcing? about half way down the page.

  26. “- How big is the direct human contribution to the amount of water wapor in the air?”

    In the troposphere, none. Water vapour depends on temperature alone. This is why Tamino’s correlation shown above is so unsurprising, and so damning.

    Unless you argue that the increase in water vapour caused by the increase in temperature caused by the increase in CO2 is “anthropogenic” – but it’s easier just to call it a feedback.

    As BPL said, we just can’t put extra water vapour into the atmosphere. It just falls straight out. So the mere concept of Anthropogenic Water vapour is silly. It’s meaningless.

    The situation in the stratosphere is a little more nuanced, but it sounds like you want to stick to the basics with your opponents for now.

  27. I tried to compare the correlation between Willett et al. 2008 and Hadley’s V3 and Gistemps. I came up with a correlation of 0.79 between Gistemps and willett and a correlation of 0.82 between hadley and willett.

    I was wondering if anyone knew where to find all the raw data for this graph so comparisons can be made for each. I think that the satellite measurements and hadley are likely to fit these graphs the best.

  28. I just compared the RSS and UAH data with the humidity data and they both fit very well (0.8474 and 0.8479 respectively) but this is one of the problems with linear models in that sometimes the points line up nice and correlation is high but yet when plotted there can be distinct divergences. What I noticed is that visual inspection of the graphs showed that RSS clearly follows the humidity trend from Berry and Kent (the longest humidity reconstruction) much better than UAH. I dont know if I would call this any sort of validation of RSS but if you plot them up you can see for yourself. The data is easily accessible through http://www.ncdc.noaa.gov/bams-state-of-the-climate/2009-time-series/?ts=humidity

  29. I just read this post, very interesting, and a great summary. One post however commented on evaporation, and I thought I would leave a comment to highlight a very interesting phenomenon closely associated with the humidity story here, which is that global evaporation trends are actually decreasing. This is because although vapour pressure may increase with temperature, it is the vapour pressure deficit (difference between actual vapour pressure, and how much water the atmosphere can hold when saturated) and insolation which have the biggest influence on evaporation. Thus although temperatures may rise, we should not expect an increase in evaporation, nor should we expect a feedback between temperature, vapour pressure, and evaporation.

    • Thanks for this observation, Josh. Yet I’m struggling a bit with the implications.

      So, some questions:

      1) Any “mustn’t miss” information on evaporation trends you can point me to?
      2) Why doesn’t the vapor pressure deficit increase, if evaporation is declining even as temperatures (and hence “how much water the atmosphere can hold when saturated”) increase?
      3) How can the specific humidity increase with warming (as stated in Tamino’s post) if evaporation is not also increasing? The water has to enter the atmosphere from somewhere–ocean, lakes, organisms–and only the (one presumes, relatively small) last case doesn’t rely mostly upon evaporation for atmospheric uptake. So where is that water coming from, and how does it get into the atmosphere?

      Your comments seem to me to create large contradictions/conundrums. Can you clear up some of these issues?

    • Interesting Josh, some time ago (2 years) there was an article that asked the question “Is our atmosphere drying out?”. See http://chriscolose.wordpress.com/2008/06/23/is-the-atmosphere-drying-up/

      From plant and soil observation this year, though we’ve had more rain than normal in July/August, it also evaporated at an absolute astounding rate. Weeds are this year near unstoppable, tonnes cut compared to other years and incredible long views, easily 100-150km from my hillside, bluer than normal skies (central Italy)… yet the hygrometer has been indicating higher than normal humidity… strange days.

  30. The first article I know of about the decrease in evaporation is:

    The first author of this study has also followed this up with much more detail in subsequent papers, mainly dealing with Australian data, which is where I work.
    I’m adapting the penman equation (pan evaporation model) for the LGM in Australia at the moment, so its fresh in my mind, here goes:
    Evaporation, or here, pan evaporation can be divided into the sum of two components, evaporation due to radiative forcing, and that due to aerodynamic forcing (which is the one that includes vapour pressure and wind speed). Importantly, temperature is factored into both components, but only via a derrivative of saturation vapour pressure over temperature, known as the Clausius-Clapeyron curve:
    (note: normally used to describe phase transitions, which are not important here, but the shape of the curve, and hence dP/dT, is)
    If solar irradiation remained constant, and evaporation was only due to this radiative component, because of the relationship between dP/dT and some constants, evaporation would actually decrease with increasing temperature. The opposite is true for the aerodynamic component, if wind speed and vapour pressure deficit are constant, then increasing temperature would lead to an increase in evaporation if this were the only component. Thus, temperature has opposing effects on the two components of evaporation, and what remains is solar irradiance, wind speed, and vapor pressure (saturated and actual). As has been mentioned, temperature does influence the specific humidity because it increases with temperature (I will call it measured vapour pressure, Ea), but temperature rise due to greenhouse gases also has the peculiar effect of increasing the saturated vapour pressure (Es), or the ammount of water the atmosphere can hold when saturated (again the Clausius-Clapeyron curve) . This means Es – Ea (vapour pressure deficit) remains roughly constant when averaged across the globe, although it may obviously increase slightly in some places, particularly in arid areas, but its effect on overall evaporation may not be substantial. Thus we are left with wind speed (which can account for much of the reduced evaporation in australia) and also solar irradiance, which can account for some trends in reduced evaporation through so called ‘global dimming’
    The relationship between precipitation and evaporation is even more complicated. In arid or semi arid areas, there is often a strong relationship between increasing precipitation and decreasing evaporation. This is because the increased rainfall is generally associated with increasing cloud cover, and therefore solar irradiation decreases, and vice versa. More temperate or tropical areas however are not so sensitive to such subtle changes in cloud cover, probably because of the added complexity of increased vegetation cover. In these areas, a fully wet surface equals potential evaporation, no matter how much extra water you add. As everything starts to dry out, evaporation declines and is limited by the amount of water that the soil can supply to the air and plants for evaporation, which in turn depends on many local physical properties. Ok, I think that about does it for now!

  31. For EHS, if he’s still in his discussion, or others interested: http://www.ipcc.ch/publications_and_data/ar4/wg1/en/ch2s2-5-6.html

    Above link discusses small radiative forcing found by Boucher of irrigation H20 which was dominated be evaporative cooling…

    Also the link below is to one calculation showing combustion h20 being even more trivial (as is also stated in the link above). http://atoc.colorado.edu/~englishj/ATOC5600home.htm

    • Uthan: Your first link is to Fotser & Ramaswamy et al in AR4 G1 2007: They are in error when they say that “anthropogenic use of water is less than 1% of natural sources of water vapour and about 70% of the use of water for human activity is from irrigation”, citing Doll (2002) and Boucher et al. (2004), see below. These authors add they exclude radiative forcing from anthropogenic sources of tropospheric water vapour, “since these sources affect surface temperature more significantly through these non-radiative processes, and a strict use of the RF is problematic. The emission of water vapour from fossil fuel combustion is significantly lower than the emission from changes in land use (Boucher et al., 2004)” (Forster and Ramaswamy et al.,2007:185).

      The error is in that final statement. Given that the chemistry indicates a ratio of around 1:2 for H2O:CO2 emissions from combustion of various hydrocarbonssome , using authoritative data (www.globalcarbonproject.org, Le Quere et al., 2008) we know that the H2O component of anthropogenic hydrocarbon combustion emissions was of the order of 18 Gt H2O equivalent in 2006-07, much more than their estimate for land use change emissions of 1.47 GtC (= 5.39 GtCO2). Our estimate is consistent with the Gaffen and Ross (1999) calculation of direct emissions of water vapour by combustion of hydrocarbon fuels at 10 Gt of H2O in 1990.

      Your second link is also misleading. Water vapour emissions from combustion are indeed small relative to general evaporation , but if the c.30Gt of CO2 emissions is dangerous, why is the associated c.18 Gt of H2O irrelevant, given the known higher RF of atmospheric H2O?

      [Response: When extra water vapor is added to the atmosphere, it is quickly removed by precipitation. The time scale for this process is days. When extra CO2 is added, it is slowly removed by natural processes. The time scale for this process is hundreds, even thousands, of years.

      There is absolutely no comparison between the anthropogenic contribution to atmospheric CO2, and to H2O. This is very basic.]

      • Not sure what you mean: “They are in error when they say that “anthropogenic use of water is less than 1% of natural sources of water vapour and about 70% of the use of water for human activity is from irrigation”, citing Doll (2002) and Boucher et al. (2004), see below.”
        Use of water for human activity is different than combustion (water produced from combustion). And as you can see below, neither combustion nor irrigation comes close to 1% of the natural evaporation.

        Where is the error?

        Also, the the Boucher paper calculated a trivial radiative forcing from irrigation (0.03 W/m2) – so even if they’re wrong and combustion adds 10 fold more H20, and assuming any of it matters after addition in the lower troposphere (which Boucher et al admit is uncertain), and that relative humidity can rise from this addition, it’s still a trivial amount of radiative forcing compared to CO2 and the temperature dependent specific humidity.

        Also, not sure why you find the 2nd link misleading, that link and the cited Gaffen and Ross (thanks for pointing that paper out btw) both state that anthropogenic water vapor additions are trivial in relation to solar evaporation (both find a 10,000 times less water added from combustion than from natural evaporation).

        They state:
        “The consumption of fossil fuels produces both carbon dioxide and water vapor as combustion products. This anthropogenic water vapor source could influence background water vapor levels. Based on carbon emissions data (Marland et al. 1994), we estimate global water vapor emission from fossil fuel consumption to be of order 1012 (in 1960) to 1013 (in 1990) kg yr−1. Evaporation contributes of order 1017 kg yr−1 to the atmosphere (van der Leeden et al. 1990). Thus, on a global basis, evaporation from the surface far exceeds anthropogenic water vapor emissions. Given the fast cycling of water vapor in the atmosphere, it seems unlikely that the global anthropogenic source can account for the observed trends of several percent per decade.”

        This makes sense to me, how would increasing the amount of evaporation by 0.01 % lead to a several percent increase in humidity?

    • Egil Heine Strand

      Thanx a heap, Utahn
      Had almost forgotten this thread – nice to be reminded of it and with such useful information embedded – über-brilliant