US Temperature Trends, In Context

There was some debate a while back around about a temperature chart some Conservative groups were passing around.

Obviously, on this scale, global warming does not look too scary.  The question is, is this scale at all relevant?  I could re-scale the 1929 stock market drop to a chart that goes from Dow 0 to, say, Dow 100,000 and the drop would hardly be noticeable.  That re-scaling wouldn't change the fact that the 1929 stock market crash was incredibly meaningful and had large impacts on the economy.  Kevin Drum wrote about the temperature chart above,

This is so phenomenally stupid that I figured it had to be a joke of some kind.

Mother Jones has banned me from commenting on Drum's site, so I could not participate in the conversation over this chart.  But I thought about it for a while, and I think the chart's author perhaps has a point but pulled it off poorly.  I am going to take another shot at it.

First, I always show the historic temperature anomaly on the zoomed in scale that you are used to seeing, e.g.  (as usual, click to enlarge)

click to enlarge

The problem with this chart is that it is utterly without context just as much as the previous chart.  Is 0.8C a lot or a little?  Going back to our stock market analogy, it's a bit like showing the recent daily fluctuations of the Dow on a scale from 16,300 to 16,350.  The variations will look huge, much larger than either their percentage variation or their meaningfulness to all but the most panicky investors.

So I have started including the chart below as well.  Note that it is in Fahrenheit (vs. the anomaly chart above in Celsius) because US audiences have a better intuition for Fahrenheit, and is only for the US vs. the global chart above.  It shows the range of variation in US monthly averages, with the orange being the monthly average daily maximum temperature across the US, the dark blue showing the monthly average daily minimum temperature, and the green the monthly mean.  The dotted line is the long-term linear trend

click to enlarge

Note that these are the US averages -- the full range of daily maximums and minimums for the US as a whole would be wider and the full range of individual location temperatures would be wider still.   A couple of observations:

  • It is always dangerous to eyeball charts, but you should be able to see what is well known to climate scientists (and not just some skeptic fever dream) -- that much of the increase over the last 30 years (and even 100 years) of average temperatures has come not from higher daytime highs but from higher nighttime minimum temperatures.  This is one reason skeptics often roll their eyes as attribution of 15 degree summer daytime record heat waves to global warming, since the majority of the global warming signal can actually be found with winter and nighttime temperatures.
  • The other reason skeptics roll their eyes at attribution of 15 degree heat waves to 1 degree long term trends is that this one degree trend is trivial compared to the natural variation found in intra-day temperatures, between seasons, or even across years.  It is for this context that I think this view of temperature trends is useful as a supplement to traditional anomaly charts (in my standard presentation, I show this chart scale once and the standard anomaly chart scale further up about 30 times, so that utility has limits).


  1. joe:

    Everyone can play games with the scale of the graph, both sides are guilty - Though responding with a deceptive graph to show how deceptive another graph is has some merit

    My beef is the standard AWG graph is based on temp anomalies instead of showing the prior actual (or connived temps). By only showing the anomaly, you lose the ability to reference back to an older graph to see whether the prior year base lines have changed.

  2. Tim Broberg:

    Now let's see that graph in Degrees Kelvin...

  3. Douglas2:

    I think your chart could be even more improved by showing trend lines not just for average but also for tmin and tmax.

    When the Powerline graph exploded on my facebook page with the various parodies of it, every person:
    • linked to National Review Online's contextless tweet that:
    -- just credited "powerlineUS"and
    -- hotlinked to a marginally-related climate-science article on NRO's own site.
    • called it "National Review's" temperature graph.
    • did so so long after the Powerline blog post that the post giving the context was no-where near the front page of blog posts.
    It's nice that Kevin Drum tracked down the original context. At the time, I concluded that those making fun of the graph didn't want me to see the context because it would contradict their point*.
    When I did track it down my reaction was much as yours.

    *I have proposed "Douglas2's Law" on facebook posts:
    "If you click every link in a post, and every link in those links, and do not in the open pages have either the primary source or a link to it, then the author of that post is lying to you."
    For Science posts, If it takes me more than three clicks of hyperlinks to get to the primary source, I invariably find that the text of the article does not support the headline posted on facebook. I've got access to nearly all academic journals via my workplace, so open-access vs subscription isn't an issue for me.

  4. Shane:

    Very interesting the "volatility" in the low temperature portion of the graphs.

  5. Daublin:

    Many times yes. The daily and monthly changes are much larger than are being discussed in Al Gore's worst case scenario for a 100 year shift, and yet the ecosystem does not fall apart. The year-over-year changes are not larger but are comparable, and again everything does fine.

    It's hard to think of an ecosystem model where these observations would square with a looming catastrophe 100 years from now. If 3 degrees over 100 years is going to ruin the Earth for human habitation, then shoudln't a similar change over 1 year be at least noticeable?

  6. Ike Pigott:

    Has Kevin given you a reason why you can't be reinstated there? A valid one?

  7. Mercury:

    "The problem with this chart is that it is utterly without context just as much as the previous chart."

    It's just possible that 1890-yesterday isn't much context either.

  8. Jim Collins:

    Damn. I think that I might have mentioned this before. I used to work at a small airport in Western Pennsylvania in the early 90's. One of our duties was to record the temperature, barometric pressure, wind speed and direction four times per day. The instruments were in a housing that had been there since the airport opened in the late 1920's. The thermometer was a mercury thermometer that was made in 1928. We wrote these readings in a ledger book and once a month the secretary would take these readings write them on a form and send the form to NOAA.
    While I worked there they installed an Automated Weather Observation Station (AWOS). AWOS recorded all of this information and then some and sent it direct to NOAA. Once AWOS was operational we didn't have to record the info any more.

    A little later we were cleaning out some office space and found a stack of the ledger books dating back to 1931 and we threw them away. I would love to have those books now. Sixty years of temperature records, taken on the same thermometer at the exact same location. Transfer that data to an Excel spreadsheet and see what the graph would look like.

  9. Q46:

    Average Global temperature simply does not exist, no more than the average weight of all the fish in the sea exists.

    That is the joke.

    And think! How ever would you be able to measure the weight of all the fish, and even if you could - so what? - what would an average weight actually mean if it differed from one period to the next?

    What is an 'average' fish? What is an 'average' global temperature? What 'should' either be?

    The second graph with a different scale showing warming of 0.8C, is a joke too, showing a degree of accuracy that most of the instruments used to measure temperatures over the last 130 years do not have.

    Many temperatures included in this fantasy are converted from °F to °C with rounding errors, temperatures extrapolated from instruments hundreds of miles away in the absence of any in the locality allegedly experiencing the temperatures (particularly so in the Antarctic and Arctic), then Urban Heat Island Effect, instrument drop-out, Human error, re-siting and all 'corrected' by an arithmetical process giving results to thousandths of a degree rounded down or up as needed, and done repeatedly over the last few years curiously getting cooler the further into the past the 'adjustments' go.

    And the whole Global Warming panjandrum gyrates around silly people arguing and predicting doom over tenths or even hundredths of a degree of a fiction.

    A joke indeed.

    Next: just how many fairies can you get on the head of a pin?

  10. ColoComment:

    Why, in the unlikely event that there is, or develops, an actual, scientific, and reliably data-evidenced case for global warming, someone, somewhere, needs to start a tally of the possible POSITIVE consequences of warming global temperatures.

    Why do we only hear doomsday claims? Because panic suits the global-warming activist industry.

  11. Matthew Slyfield:

    Kevin isn't in control of who is banned/reinstated. He may not even know know why the Mother Jones admins banned Coyote in the first place.

  12. Matthew Slyfield:

    From 0 to the highest point temp recorded on the face of the earth.

  13. Alex:

    note: this is based on minimal knowledge of climate science.

    I feel like you're deliberately lowering the level of debate by using the context argument. Couldn't an alarmist say "the science looks at how even slightly higher sustained average temperatures can have dramatic impacts. You're not addressing the science. You're using the context argument to say it doesn't **seem** so bad."

    I think your stock market analogy misses the underlying climate science (again, not saying I know what that is). Here's another analogy that's also bad but makes the point: let's say you've driven a car 200,000 miles (around when it breaks down). You and the first chart's author would say, "hey, you drive 20 miles a day. And in the context of 200,000, another few miles is meaningless." But keep driving a few miles and eventually your car dies.

  14. David in Michigan:

    CO2 is the driver of global warming (so we are told). Plot temperature and CO2 on the same chart or as an overlay.

  15. Alex Silverman:

    In fact, as David in Michigan says, it might not even be temperature that matters

  16. sailor116:

    You should add trend lines for TMAX and TMIN, in colors to match. It would make the graph better.

  17. markm:

    The 0-100 Fahrenheit scale is quite relevant, in that it was created for weather observations. In much of the temperate zone, including the area where things like thermometers and temperature scales were invented, 0-100F will cover the daily highs and lows at least 95% of the time. So what that graph shows is that the variation in mean temperature is invisible on a scale that includes most of the weather, but not the worst extremes.

    This at a minimum confirms that climate researchers are trying to find tiny variations by averaging a quantity that varies a hundred times as much as what they are looking for; this makes the average very, very sensitive to the details of how the data is collected and averaged. Adding or removing a few measurements can change the average noticeably. But many measurements have to be "adjusted" before they can be averaged (e.g., estimating noon and midnight temperatures from temperatures recorded at other times), and more data has to be estimated (that is, made up) for areas with no reliable temperature collection. If the persons making those decisions were pure scientists with nothing to prove, and the results were irrelevant to political policies, you'd still have to wonder if these persons - or any human being - was really up to the task. But the persons making those decisions are actually idealogues who claim that the results prove the need for drastic world-wide policy changes...

    Another thing: the resolution of that graph is greater than the accuracy of any thermometer. So one of the big claims is that you improve bad measurements by averaging them. There are situations where this works, but only if there is no systematic bias (like heat islands without corresponding cold islands), and probably not with a data set requiring lots of adjustments.