# Honey, I broke the graph

Well, I did it.

NOAA released their global temperature data for September, so I thought it was time for an update. But I was trying to update the NOAA temperature data graph in a way that would really tell the story. So first I graphed the yearly average global temperature up to last year, but left out this-year-so-far. Then, I figured I’d add an extra point, marked as an asterisk, in red, to emphasize this year’s value (so far). Problem is — it broke the graph:

Dang that pesky global warming.

### 29 responses to “Honey, I broke the graph”

1. Magma

The more rational explanation is that warmists are shrinking the graphs. Clearly such a feat is child’s play for a group of conspirators willing to melt nearly a trillion tonnes of ice a year to advance their plans.

• Moses

It’s a warmist conspiracy against geometry. They have an interest in controlling our axes (so we can’t chop down trees).

2. Martin Smith

I read a blog yesterday by a guy who claims that the graph of the increase in CO2 since 1880 is misleading because the range on the y-axis is restricted to 300 ppm to 400 ppm. He says he always starts his y-axis at 0.

• Oh, you mean like this:
http://denialdepot.blogspot.ie/2010/11/how-to-cook-graph-skepticalsciencecom.html
Graph y-axes should always start at zero or else they will mislead people.
We can learn a lot from from the masters :-)

• michael sweet

Does anyone know why Dr. Inferno has not posted for so long? I am worried that the warmistas have knocked him off.

• Malcolm

Was about to say ‘only a matter of time’ but I’d guess somewhere there is already a denier who insists temperature graphs should all start at absolute zero.

3. That’s not a totally invalid point, Martin. I believe you should always include 0 in the range of an axis if (a) the axis is on an absolute scale where 0 has meaning and (b) the data are “reasonably close” to zero, in the sense that you won’t have to compress the variation in the data by “too much” to include 0. My rule of thumb for “too much” is a factor of 5-10. Certainly both choices are defensible, however when that happens I think it does make sense to show the absolute one first, and then show one that is zoomed in on the area of interest. Even much larger factors can be justified when the variation in the data is known to be “small”, for example you wouldn’t want a plot with the range [900,1000] showing a rise over a 10 year period from 913 murders per year to 986.

• kinimod

I would add, that it depends on which range has an effect. It makes a big difference, whether somebody has 36 or 41 °C body temperature – on an absolute scale this is 309 and 314 K, just about 1.7 %, barely to notice on a plot in absolute scale. The same of course holds for average global surface temp. 2 K, just 0.7 %, makes already a big difference. So a scale of some degrees is completely justified.

• Martin Smith

But the author was criticizing the standard graph of CO2 increase in the atmosphere. 0 ppm of CO2 in the atmosphere means there is no life on earth. For millions of years of evolution, CO2 stayed in the range of 200 to 300 ppm, so it makes no sense to include 0..200 on the y-axis for that graph. The author also criticized a global average temperature graph because it didn’t include 0 F, but that isn’t meaningful for earth either. The author’s complaint was that “alarmists” make graphs look “alarming” because they don’t start the y-axis at 0, even when 0 makes no sense: http://www.climatechangedispatch.com/the-only-global-warming-chart-you-need-from-now-on.html

• Yeah, the temperature complaint is nuts. The CO2 complaint is worth about 1 point on the scale from 0 to 10, I would say. The choice of 250-400 can be justified, including 0 can be justified, and really the ideal is switching to a log axis (in which case there is no 0, so the issue of including 0 goes away) and including enough range to show the preindustrial value. Plotted any of these ways it is alarming, so the author really has no point at all. Even his preferred plot is alarming, look at how short the range covered by his x-axis is. He says he is plotting “the same data” but he silently changes the range of the x-axis from [1750, 2010] to [1959, 2009] (to hide the increased rate of change), and yet the result is still alarming.

• The two indisputable principles for graphing CO2 honestly:

1. Since it’s a concentration, zero is the relevant baseline, so always include zero.
2. Since radiative forcing depends on the log of CO2, always use a log scale.

As long as you satisfy those two principles, you can’t go wrong! ;)

4. Wait till you see October!

• JCH

But, if it’s above the line, then it didn’t happen.

• +1 internets for that one, JCH. Good thing I just finished my tea.

• The real life example is the converse: if it’s below baseline, then it can’t be counted as ‘warming’ because it’s a cooler than average temperature… yes, let me say again that that is a real example.

Very convenient because it generally lets you cut a warming trend in half right from the start!

5. I’m certain that next year will be cooler than this one and we’ll start a brand new “hiatus decade” starting at 2015.

6. I feel compelled to mention Denial Depot’s “How to Cook a Graph SkepticalScience.com Style”?

7. Patrice

Darling… fix it ;)
Sorry honey, it will be just a temporary fix :(

• Magma

Accounting for a ~0.1°C shift in the GISTEMP baseline from 2008 to now, looks like the bet still open, but closing fast. (2015 point is Jan-Sep)
[IMG]http://i61.tinypic.com/4pt7wk.jpg[/IMG]

• Magma

An update of Tamino’s January 2008 main figure, with 2 sigma error/uncertainty bounds based on a standard deviation of 0.09°C calculated from the detrended 1975-2015(to date) GISTEMP anomalies.

[IMG]http://i58.tinypic.com/2lu54j5.jpg[/IMG]

8. Nah. You didn’t break the graph. You are just thinking outside the box. The box being the imaginary walls that surround the people who are in denial that human induced global warming exists.

9. Jim Eager

Tamino, graphs like this can rightly be dismissed as misleading, *unless* they use 12-month averages ending in the last data month, September in this case. What is the 12 month period of this graph?

I can not understand why anyone would use a calendar 12-month average period and simply append the 9 months to date of the current year. It opens one to well deserved ridicule by those we constantly demonstrate to be lying with statistics.

[Response: There’s nothing wrong with putting a 9-month year-to-date figure on an annual average graph, when that is clearly stated. For a lot of people, it’s far less confusing than using a year which isn’t a calendar year. Your accusation of “well-deserved ridicule” is out of line.]

• Magma

In particular when the data consists of global temperature anomalies that remove seasonal variations, why not? It is very likely that the 2015 GISTEMP anomaly will end up somewhere around 0.81 or 0.82°C, very close to the year-to-date value of 0.80°C.

2015 GISTEM LOTI (°C x 100)
Jan 82
Feb 88
Mar 90
Apr 74
May 79
Jun 77
Jul 73
Aug 81
Sep 81

• Jim Eager

OK, the “well deserved” is out of line. The rest of my last sentence stands.

10. barry

I’ve been curious about the el Nino effect on satellite lower trop data. Doesn’t seem to be showing up over the last few months, even with the 2-3 month lag I’ve read about (is that right?).

I would have thought we’d have seen el Nino reflected in global LTT (it shows up in the tropics), as we have with other significant ENSO events. Would this suggest something is out of whack with the satellite records? Has this been discussed anywhere?

Official UAH data plotted at woodfortrees: version 5.6
http://www.woodfortrees.org/plot/uah/from:1990

Beta version 6.0 plotted at Roy Spencer’s site:
http://www.drroyspencer.com/2015/10/uah-v6-0-global-temperature-update-for-sept-2015-0-25-deg-c/

11. barry

[Didn’t see my post awaiting moderation, and wondered if the links broke it. here’s the post without links]

I’ve been curious about the el Nino effect on satellite lower trop data. Doesn’t seem to be showing up over the last few months, even with the 2-3 month lag I’ve read about (is that right?).

I would have thought we’d have seen el Nino reflected in global LTT (it shows up in the tropics), as we have with other significant ENSO events. Would this suggest something is out of whack with the satellite records? Has this been discussed anywhere?

• chrisd3

It’s far from the only thing that suggests “something is out of whack with the satellite records”, IMO.