Lies, Damned Lies, Statistics … and Graphs

Guest Post by Willis Eschenbach

I got to thinking about how the information about temperatures is presented. Usually, we are shown a graph something like Fig. 1, which shows the change in the US temperatures over the last century.

Figure 1. Change in the US annual temperatures, 1895-2009. Data from the US Historical Climate Network (USHCN DATA) [Yes, it's in Fahrenheit, not Celsius, but hey, it's US temperature, and besides I'm doing it in solidarity with our valiant allies, all the other noble countries that are bravely fighting a desperate rear-guard action against the global metric conspiracy ... Liberia and Myanmar ...]

Whoa, this is obviously a huge and scary change, look at the slope of that trend line, this must be something that calls for immediate action. So, what’s not to like about this graph?

What’s wrong with it is that there is nothing in the graph that we can compare to our normal existence. Usually, we don’t even go so far as to think “Well, it’s changed about one degree Fahrenheit, call it half a degree C, that’s not even enough to feel the difference.”

So I decided to look for a way to present exactly the same information so that it would make more sense, a way that we can compare to our actual experience. Fig. 2 is one way to do that. It shows the US temperature, month by month, for each year since 1895.

Figure 2. US yearly temperatures by month, 1895-2009. Each line represents the record for a different year. Red line is the temperature in 2009. Data source as in Fig. 1. Photo is Vernal Falls, Yosemite

Presented in this fashion, we are reminded that the annual variation in temperature is much, much larger than the ~ 1°F change in US temperatures over the last century. The most recent year, 2009, is … well … about average. Have we seen any terrible results from the temperature differences between even the coolest and warmest years, differences which (of course) are much larger than the average change over the last century? If so, I don’t recall those calamities, and I remember nearly half of those years …

To investigate further, Fig. 3 looks at the decadal average changes in the same way.

Figure 3. US decadal average temperatures by month, 1900-2009. Red line is the average for the decade 2000-2009. Photo is Half Dome, Yosemite.

Most months of the year there is so little change in the decadal averages that the lines cannot be distinguished. The warming, what there is, occurred mostly in the months of November, December, January, and February. Slightly warmer temperatures in the winter … somehow, that doesn’t strike me as anything worth breathing hard about.

My point in all of this is that the temperature changes that we are discussing (a global rise of a bit more than half a degree C in the last century) are trivially small. A half degree change cannot be sensed by the human body. In addition, the changes are generally occurring in the winter, outside of the tropics in the cooler parts of the planet, and at night. Perhaps you see this small warming, as has often been claimed, as a huge problem that “vastly eclipses that of terrorism” (the Guardian). Maybe you think this is a pressing concern which is the “defining issue of our era” (UN Secretary Ban Ki-Moon).

I don’t. I’m sorry, but for me, poverty and injustice and racial prejudice and totalitarian regimes and recurring warfare and a lack of clean drinking water and torture and rampant disease and lack of education and child prostitution and a host of other problems “vastly eclipse” the possibility of a degree or two of warming happening at night in the winter in the extra-tropics fifty years from now.

Finally, the USHCN records are not adjusted for the urban heat island (UHI) effect. UHI is the warming of the recording thermometers that occurs as the area around the temperature recording station is developed. Increasing buildings, roads, pavement, and the cutting down of trees all tend to increase recorded temperatures. Various authors (e.g. McKitrick, Spencer,  Jones)  have shown that UHI likely explains something on the order of half of the recorded temperature rise. So even the small temperature rise shown above is probably shown somewhere about twice as large as it actually is …

My conclusion? Move along, folks, nothing to see here …
Watts Up With That?