Hot … and Wet

The future will be hot, at least, hotter than the present. But temperature isn’t the only thing that will change with global warming — so will the water cycle. Some places will become drier, some wetter. In fact, some have already become wetter.

One of the “places” that’s already wetter is the atmosphere. It contains about 4% more water vapor worldwide than it did just a few decades ago, which has a profound effect on climate:

It’s a direct result of global warming: warmer air tends to hold more water vapor. It’s also one of the main feedbacks in global warming, since water vapor is also a greenhouse gas, so increased water vapor due to global warming will amplify global warming.

The global atmosphere is one thing — but most of us live on land, and in a single location. The GHCN doesn’t just store historical temperature data, it also offers historical precipitation data. There are about 2500 observing locations covering the continental U.S., and they’re located here:

I took the monthly precipitation and computed anomaly (its departure from the mean for the same month). Then I combined the anomaly data since 1900, for all stations in each very large grid, 5 deg. latitude tall and 10 deg. longitude wide, covering most of the continental U.S. Then I smoothed the combined precipitation anomaly data, to get an idea how precipitation may have trended over the last century-and-a-decade-and-a-year. This is certainly an imperfect procedure, since precipitation may not show the same degree of long-distance spatial correlation that temperature does (and for other reasons too). Nonetheless, it’s at least a first glance at what the data may be telling us — an exploratory analysis, but hardly a definitive one.

And here’s the result, graphing the smoothed time evolution for each grid on top of the map of station locations (click the graph for a larger, clearer view):

For each grid box, the bottom is an anomaly of -10 mm/month average, the top is an anomaly of +20 mm/month.

It surprised me that so many of the grid boxes show a recent upward trend. This is particular true for New England, and much of the Midwest.

It’s also worth noting that precipitation isn’t the only factor is how wet or dry a region is. As temperatures rise evaporation will increase, so some areas may become desertified even if their precipitation doesn’t decline. But precipitation is certainly a major factor! Some parts of the U.S. already show signs of change, and I regard that as trouble. I like the way weather has been distributed throughout my lifetime — we’re adapted to it. I expect that, unfortunately, both the places that get wetter and those that get drier will be ill prepared for the change.

And of course, changes in precipitation in one region can have a profound impact on other regions. Just ask those who live near the Mississippi River.


40 responses to “Hot … and Wet

  1. Also, even in a location with no annual precip trend, there may be different trends in winter vs. summer, and, more importantly, a trend towards more precip occurring in single events. See pg. 30 in the following pdf (warning, 3 MB size):

    (now I’d like to see the one-day precip figure in your 4×6 lat/long grid box format)

  2. “This is certainly an imperfect procedure, since precipitation may not show the same degree of long-distance spatial correlation that temperature does.”
    Definitely true – the temperature anomaly may be correlated on two sides of a mountain, but the precipitation anomaly probably won’t be. If someone wanted to take this to a publishable level of analysis, I’d suggest the geographical breakdown would be best done by catchment areas – that’s mostly what will matter to people/agriculture anyway.

    • A bigger problem would probably be the uneven distribution of stations, which typically biases your averages tirades conditions in more populated areas.

      This is most important in mountainous areas, especially the Rockies, where stations are sparse but precipitation is high and a larger proportion of precipitation becomes river flow downstream.

      But this quickly gets beyond what you can ever address in a blog post.

  3. Daniel Bailey

    Thanks, Tamino. Timely post on the heels of the Texas drought and the record flooding along the Mississippi.

    It may be somewhat paradoxical, but a trend towards wetter and drier is what some recent studies show. I recently profiled Dai et al 2010, which discussed that very thing, here:

    • Ironically much of Louisiana itself is in severe dought:

      When the Morganza spillway was opened the ground was so dry it absorbed a considerable amount of the initial surge ans so the level in that area did not rise as quickly as expected. A small, incidentally fortuitous occurrence given the amount of water that is expected to arrive there.

      As we watch this tragedy slowly, inexorably unfold before us on a day-to-day basis we can only feel a sorrowful compassion for those designated as expendable. At the same time we need to be mindful. The analysis done here and that we do ourselves with the data we have indicates that without the required mitigating actions we are all in one sense or another deemed expendable.

    • I was thinking about that myself when talking with someone in Memphis about the flooding she’s got. I really want to grab some policymakers by the lapels and shake sense into them. Both the Texas droughts and the midwestern flooding are a sign of things to come, according to our own experts. This might be a freak year for now, but perhaps it won’t be so freaky before the century’s out. We desperately need policy to get a grip on the impacts of climate change, even if just domestically.

  4. Over at WUWT, several people have latched on to the paper by Wentz et al. ( ) that precipitation seems to be increasing faster than climate models predict and used it to try to propose a picture whereby the water cycle speeds up and the evaporation/condensation transports the extra heat as latent heat up into the atmosphere where it gets above the greenhouse gases and then safely escapes to space.

    I see a few problems with this picture, but would welcome comments on these or further thoughts that people have here:

    (1) If it were correct, it would seem to make a definite prediction that the upper tropospheric temperatures should be increasing faster relative to the surface than the models predict…which directly opposes their mantra that the “hot spot” in the tropics that the models predict isn’t there, which seems to imply the opposite. Or, to put it another way, to the extent that the increase in the water cycle does slow the warming at the surface due to increased latent heat transport from the surface to upper troposphere, it is already included in the models as the lapse rate feedback…And, the evidence from the satellite and radiosonde data regarding the “hot spot” suggest, if anything, that this feedback is being a bit overestimated by the models.

    (2) Another point is that changes in heat transport between the surface and atmosphere tend not to lead to much change in surface temperature because the heat transport processes already in place basically respond (mainly because the temperature profile tends to stay pretty close to the adiabatic lapse rates). That is why top-of-the-atmosphere radiative balance plays so much more important a role than surface radiative balance in determining the surface temperature change. There is a nice calculation illustrating this sort of thing in L.D. Danny Harvey’s book “Global Warming: The Hard Science” and the issue is also alluded to more indirectly in the Wentz paper itself, at least in relation to clouds when they say, “For example, variations in modeling cloud radiative forcing at the surface can have a relatively large effect on the precipitation response (4), whereas the temperature response is more driven by how clouds affect the radiation at the top of the troposphere.”

    • Assuming that higher-than-expected evaporation/precipitation would lead to higher-than-expected heating of the upper troposphere by latent heat transport, boosting the lapse rate feedback, wouldn’t that have to go along with a higher-than-expected moistening of the upper troposphere, boosting the water vapor feedback? And since the amplifying water vapor feedback dominates over the dampening lapse rate feedback, shouldn’t one expect the net effect to be further warming? (My reasoning here is probably way too simple, but from your description it sounds like the WUWT folks are being simpler still.)

    • Joel,

      I need to think about this some more, but the WUWT crowd are clearly spinning this paper without thinking things through… hypotheticals, some of which may even contradict each other in the real world. But as we know misinformation and obfuscation is their game.

      This is probably just yet another of many, many occasions when the Wattoids think that they have stumbled upon some great “truth” only to have is refuted, or for them to ‘stumble’ on a ‘truth’ that has already been refuted. But that is how they manufacture doubts and debate….They should follow Chris Cholose’s sage advice.

      The findings by Wentz et al., if correct, are not room for celebration.
      At the end of the day we know that global temperatures are increasing, that atmospheric water vapour levels are increasing in response and in turn extreme precipitation events are increasing.

      As Daniel Bailey pointed out though, is is not getting wetter everywhere (it is getting drier over large tracts of land), and the science reported in the IPCC does not claim that either.

    • “Above the greenhouse gases” also seems to be a fail–water vapor declines much more rapidly with increasing altitude than CO2, right?

      • Yes, definitely true for WV and IIRC, CO2 is well mixed above the boundary layer. Also, they seem to be talking about thunderstorms transporting the moisture higher up in to the troposphere, but as we know, not all precipitation falls from thunderstorms.

  5. Michael Hauber

    From my memory US is expected to get wetter in the far north, and drier in the South, particularly SW? It seems that the places that are expected to get drier aren’t obviously losing rainfall (yet) and only getting drier through the heating effect. This mirrors Australia where the rainfall stats for the north seem to be very defnitely trending upwards and stats for the south where it is predicted to get drier are mostly flat.

    And one issue that many ‘sceptics’ forget when the discuss the possibility that negative feedbacks from clouds or water vapour changes may save us is that these negative feedbacks involve changing the hydrological cycle. Perhaps we will be ‘saved’ by negative feedbacks, and the disruption to the hydrological cycle will be a worse cure than the illness….

    • There’s a lot of east-west variation between models but the general pattern, at least for North America is less precipitation in the south and more precipitation in the north.

      Bit the “transition zone” between drier and wetter encompasses most of the lower 48, so it’s difficult to have much confidence about what’s going to happen at any given location, at least south of the 45th parallel.

  6. The part of the country where I grew up, eastern Dakotas, appears to be getting wetter. My parents are elderly, so I go back a few times a year. These are some thing I’ve noticed. When I was a kid we harvested corn in the fall. As far as I can remember we did not have to wait until the ground froze. Now they do. I’m told they have no choice. The equipment will drop right to the axles if they attempt to go out onto the fields before the ground is solidly frozen.

    When I was a kid we had a sump pump in our basement. It ran in the spring and that was it. We unplugged the thing. Now my parents have three sumps in their basement, and they run year round. When I was there last fall we had to backfill the sump holes with gravel as large cavities were wallowed out under the basement floor. Each time the sump runs, it takes a little dirt with it. Basically there was a lake under the floor. About 6 years ago I replaced all of the wooden posts that are holding up the house. They were fine when they bought the house in the 1970s. The house was built in 1889. They were completely rotted out below floor level. When I dug the new holes they immediately filled with ground water. If the ground water had been as today, the original posts would not have made it to the Roaring Twenties.

    Many fields have become persistently flooded. Fields that have been worked since homesteading. Many farmers have taken to entering into conservation easements to recoup some value for their persistently flooded land.

    The Red River flooding and the Devils Lake story make national news, but I think the whole area has changed.

  7. Joel,

    I’m very skeptical of the Wentz results, and there’s other work showing that there might not be a significant long-term precipitation increase as well (e.g., Gu et al., 2007, Journal of Climate). Trying to integrate a “global precipitation” variable over space, time, etc is not very easy, and might be pointless.

    There are also good theoretical constraints involving the tropospheric (or probably better yet, the surface energy budget) for global precipitation/evaporation anomalies, which go up much less rapidly than the column integrated water vapor via Clausius-Clapeyron. Even further, the water vapor increase that results in excess precipitation vs. that which causes a positive radiative feedback are entirely different things, and it’s best not to discuss them interchangeably.

  8. Bernard J.

    At the risk of being excessively flippant, perhaps Watts could revisit his surfacestations photo album and discern whether the areas of increased precipitation correspond with sites that have recetnly had overhanging trees cut down…

    Of course, this is predicated on the assumption that the original placement of such stations ignored the presence of such blocks to precipitation, but given the gymnastics of reason to which Watts and his ilk are prone I wouldn’t be surprised if they ignored this obvious point and went ahead anyway.

    He might even get a second paper out of it, proving yet again that climatologists are competent in their work.

    On a more serious note, is there any way to tease out factors that might be affecting the spacial distribution of areas that display the recent precipitation increase? I’m curious about the overall spacial distribution hockey stick defined by the regional precipitation hockey sticks, and can’t help but wonder what might be determining that pattern.

  9. Why 5×10 instead of 5×5?

    • TrueSceptic

      I assumed that this to compensate for the difference in distance covered by a degree of longitude and a degree of latitude as you move away from the Equator, but it doesn’t really work out.

      A degree of latitude is always equal to the same distance (10,000/90 = 111 km approx). Lines of longitude converge as you increase latitude, however, so a degree covers an increasing small distance. At the Equator you get the same 111 km but at the poles you get zero. At 49° N, a degree of longitude covers about 73 km. 5° x 5° (lat x lon) is 555 x 365 whereas 5° x 10° is 555 x 730. The first is not square (ratio 1:1.52) as you would expect, but the second isn’t either (ratio 1:1.32). It’s better, but not by much. (I chose 49° N because that’s the northern border of the contiguous USA; south of that makes 5° x 10° even less square.)

      So, can anyone explain the use of 5° x 10°, or have I got the numbers wrong?

      [Response: It’s an arbitrary choice, entirely for convenience.]

  10. W Scott Lincoln

    I know that it was frequently discussed in my agriculture-related courses and hydrology program that streamflow in Iowa had increased on most rivers… sometimes by a factor of 2. The going talk was that the ~3″ of increased yearly rainfall (~35″ today compared to ~32″ 50 years ago) was causing the huge increase. Seems to be a bit of an error in the mass balance at first, but after taking out the amount of water that goes into evaporation/transpiration as well as the amount of water that infiltrates to the water table, the runoff amount you are left with may only be 5-10″. Add 3″ to that and you’ve made a substantial percent increase to runoff.

  11. Thanks, Tamino.

    Precipitation seems to be quite a bit harder to characterize than temperature, and that’s distressing as it may be a more serious threat as well. And a lot of ‘regular folks’ out there haven’t (yet) been exposed to anything explaining that precip changes may be a very big deal indeed as the climate warms.

  12. Horatio Algeranon

    “Hot…and wet”??

    You wouldn’t be trying to boost your traffic from searches, would you?

  13. [Tasteless quips suppressed.]

  14. MrCannucistan

    Of course the $64,000 Question is always still, “What is causing all these changes?” Any chance you can update your graphs with trend lines?


  15. Different analysis technique, similar result:
    We don’t discuss this in the blog or the paper, but none of the OLS trends are statistically significant in raw form. The positive trends from Texas through the Midwest and to New England become significant (p < .05) if the data are aggregated into 5-year bins to remove the ENSO cycle.

  16. Philippe Chantreau

    Horatio, you’ve been playing with words for too long…

  17. Interesting. I work for a water company in the UK and have written our climate change adaptation report (commissioned by the government).
    I’d like to replicate your analysis; we have monthly data from 1931-present. A couple of questions. Firstly, did you calculate your monthly mean from the entire 111 years to the present day, or did you use another time period within it? For info, the UK climate projections use 1961-1990 as their reference period. Secondly, is your smoothing something like a 20 year moving average, or something else? Thanks.

    [Response: For monthly means (used to define anomalies) I used the entire time span. You’ll probably want to stick with a fixed baseline period (1961-1990).

    The smoothing is a “modified lowess smooth,” and gives results comparable to a 15-year moving average.]

  18. rationale for choosing 5 by 10 degrees might even be justified by physical phenomena as the heat/water vapor transfer is from tropics to pole (90 degrees) or from the closest ocean to continents on the NH (Pacific and Atlantic cut the 360degrees in half, 180degrees), but of course the true spans of oceans vs. continents in degrees are different. Then that is further complicated by the Indian ocean which messes a clean picture in the heat transfer by being forced to south by African coast, or to north by monsoonal variations. This leads me to question whether deepening the Suez canal, facilitating heat transfer between Mediterranean and Red Sea, could have an effect on the near east climate ;-), perhaps driving some cyclones to Arabian peninsula? Sorry for going into (a bit off-topic) speculations, anyway I don’t really think cutting the N-S distance in half in climate models would produce any improvement onthe accuracy of those.

  19. This overview is non-trivial, although I’m biased since in Maine we haven’t really seen the sun for 2 weeks. In Maine, the past 5 years, excepting 2010, have been exceedingly wet during the spring and summer, shattering numerous longstanding (150-year old) instrumental records for spring and summer precipitation and river flow. Tuesday, 5/17, the Kennebec River at North Sidney hit its highest level recorded in 25 years; the previous record was in 2007. The USGS gage data are interesting.,00060

  20. I suggest a new title for this thread: “Hot, Wet, and Throbbing.”

    Sorry about that. It’s late. I’m tired.

    • Seems N/W Europe is missing the Wet part and the North Sea being considerable warmer than normal. Stronger winds are drying out the lands even further.

      Interestingly, though the Baltic Sea had a record freeze over last winter and snow packs never seen before, it’s all pretty much gone and given the pink on the NSIDC chart, anomalously negative as of yesterday. The FOGT site has a good analogues picture up to the 3 monkeys:

      Well, who’d worry… October 5 has been set as the new date and this time we all go in a one day bang :P

  21. > throbbing

    “… Fung focused on the seasonal fluctuations evident in the data. Concentrations of carbon dioxide in Earth’s atmosphere reach their highest levels in May, before the growing season begins, when photosynthesizing by new foliage draws the levels down. “We look at these records in great detail,” Fung explains, “to derive everything we can about the biosphere. It is like you can see the Earth breathing.”

    Before long, Fung’s detailed data analysis helped her build a large-scale computer model to represent the geographic and temporal variations of CO2 sources and sinks. …. recently, Fung has coupled her carbon-cycle model onto existing large-scale computer climate models to project how land and ocean carbon sinks are likely to change …