The dog is the weather

Update January 27: There is also another recent dog-based animations from Victoria (southeast Australia) explaining some of the key drivers of our climate and how some are changing.

A TV series that ran on Norwegian TV (NRK) last year included a simple and fun cartoon that demonstrates some important concepts relative to weather and climate:

In the animation, the man’s path can be considered as analogous to a directional climatic change, while the path traced by his dog’s whimsical movements represent weather fluctuations, as constrained by the man’s path, the leash, and the dog’s moment-by-moment decisions of what seems important to investigate in his small world. What might the leash length represent? The man’s momentary pause? The dog’s exact route relative to concepts of random variation? The messages in this animation are similar to the recent results of Grant Foster and Stefan Rahmstorf in ERL (see post here).

We’d also like to praise the TV-series ‘Siffer‘, hosted by an enthusiastic statistician explaining how most things in our world relate to mathematics. The series covers a range of subjects, for instance gambling theory, the Tragedy of the Commons, anecdotes about mathematical riddles, medical statistics, and construction design; it even answers why champagne from a large bottle tastes better than that from a smaller one. There is also an episode devoted to weather forecasting and climate.

Success in understanding our universe often depends on how the ‘story’ about it is told, and a big part of that often involves how mental images are presented. Mathematics and statistics can describe nature in great detail and “elegance”, but they are often difficult and inaccessible to the average person. Conversely, the man-and-dog animation is intuitive and easy to comprehend. Similarly, Hans Rosling’s Fun with Stats provides some very nice demonstrations of how to convey meaning via the creative display of numbers.

68 comments on this post.
  1. Kevin McKinney:

    #50–“Sometimes barking at the moon makes you feel like your doing something…

    but it may also be irritating the neighbors.”

    Hence the (mostly British) expression “barking mad.” (OK, maybe not.)

  2. Jbar:

    It’s a fair analogy. Reality is that global temperature varies +/- 4C from the mean, and has since at least 1900, and the deviation of monthly temperature from the mean fits a normal bell-curve distribution (i.e. seemingly random variation).

    The interesting part is that the mean isn’t tracking greenhouse gases exactly. Relative to CO2 concentration, the mean took a trip to the cold side between 1900 and 1920, and deviated to the warm side in 1935 to 1946. The “dog” in that case was still wandering around the “man”, limited by the same length of the leash, but during those time periods the “man” was himself wandering away from the greenhouse gas drivers.

    Anybody know what explains those deviations from the GHG driver?

  3. Kevin McKinney:

    Well, Foster & Rahmstorf (2011) didn’t directly address the record before 1979, and so doesn’t apply directly to your question, jbar.

    But over the period which they did examine, ENSO, TSI and aerosol loading accounted for most of the difference between ‘dog and man.’ It seems likely that those three factors would also have been a big part of the picture during the times you mention. (And I know there is quite a bit of literature examining the questions.)

    Foster & Rahmstorf (2011) summarized:

  4. vukcevic:

    # 50 & #51
    “barking mad.”
    barking mad, indeed ?
    Dr. Mann’s article was helpful in identifying what appear to be the source of the AMO, one of the most important natural temperature oscillations:
    perhaps reading Dr. Mann may be more useful than impersonating Statler and Waldorf

  5. Hank Roberts:

    > isn’t tracking greenhouse gases exactly


    This is a talking-poing strawman argument that’s been rebunked recently.
    Who’s pushing it these days?

  6. Kevin McKinney:

    #55–“. . .perhaps reading Dr. Mann may be more useful than impersonating Statler and Waldorf. . .”

    Or perhaps not? The limiting factor, after all, is not the potential usefulness of the former. . .

    Besides, I flatter myself that I do Statler and Waldorf rather well.

  7. Ray Ladbury:

    Hmmm, 1900-1920, 20 years; 1935-1946…11 years. Climate-a trend of 30 years or longer. I think I see your problem.

  8. Dan H.:

    Using your 30-year designations with the GISS global data yields the following deviations from the long-term trend line: +0.06C prior to 1896, -0.02C from 1896-1925, +0.01C between 1926 and 1955, -0.10C from 1956-1985, and +0.07 since. Regardless of the timeframe chosen, there are real deviations from the trendline. The dog is not totally random in his walk, but deviating over regular intervals. Maybe there is an oak tree on opposite sides of the street every other block (to reference your earlier analogy).

  9. John P. Reisman (OSS Foundation):

    #58 Dan H.

    What part of ‘it’s an analogy’ didn’t you understand?

    In the real world shorter-term natural variation is dictated by oceanic heat content overturn along with solar variation. There are no oak trees.

    If I am understanding you correctly, it seems your comment is hypocritical because it looks like you are attempting to mix a fantasy analogy of ‘oak trees’ with real world data and variation, while apparently simultaneously attacking by inferring analogies are weak because they are not real based on ‘your view’ of random vs. regular intervals.

    Looks like another pot meet kettle moment. It looks like you need to study attribution and ocean cycles though.

    and do look into google scholar.

  10. Dan H.:


    The oak tree analogy was in reference to Ray’s squirrel analogy. You may have missed that.

    I fully understand that the variations correspond to ocean cycles.

  11. John P. Reisman (OSS Foundation):

    #60 Dan H.

    Okay, let’s play. What’s your position on human induced global warming?

  12. Dan H.:

    Humans have been responsible for at least half of the observed warming since 1880. I expect future trends to mimic those of the past.

  13. John P. Reisman (OSS Foundation):

    #62 Dan H.

    How will the future trend mimic that of the past if humankind has increased the radiative forcing. Solar variation tends to be around 0.1 W/m2 and human induced positive forcing is 1.66 W/m2 (IPCC AR4 WG1) 1.8 (NASA)?

    Or are you saying while we continue warming (natural cycle would have us relatively flat to cooling), natural variation will continue to meander like the dog analogy along the new warming path?

  14. Hank Roberts:

    Dan H is just posting a talking point there.

    Try an expert assessment:
    “somewhere between 80 to 120% of the warming.”

    You know how to look it up.

  15. Dan H.:

    Basically, yes.

  16. vukcevic:

    No random dog walk in the Atlantic hurricane activity, as usual there is a cause and the consequence.
    According to the NOAA’s assessment the Atlantic hurricane activity is directly related to the Equatorial Atlantic’s SST; neither of which is predictable.
    However that not may be the case.
    Comparing the NOAA’s Atlantic Accumulated Cyclone Energy (ACE) index with the ‘Atlantic Hurricane probability index’ based on the North Atlantic other historical data (also available from the NOAA) it could be concluded that the hurricane activity will (on average) stay just above the normal for at least a decade.

  17. DS:

    Over what time period is it meaningful to talk about climate changes? Or put another way, how many years would it be necessary to average out the effects of natural variation (say we assume natural variation causes the global average temperature to vary in a gaussian pattern around the climate trend).

    [Response: It depends on the forcings and the climate metric you are looking at. For the current situation, you need between 15 and 20 years for the global mean temperatures, a little less for Arctic minimum sea ice extent, a lot more for a local or regional temperature signal. – gavin]

  18. DS:

    Thanks Gavin–I really appreciate you and the others helping me learn more about this stuff. Do you deduce the minimum time from looking at the data (like a 15 year filter seems to produce a smooth trend), or is this known from, say, that relevant short term climate oscillations tend to be less than 15-20 years?