In the discussion of climate change, a lot of people show a lot of graphs. I’m probably one of the more prolific blog graphers myself. There’s one fault with graphs which I think sabotages their very purpose: creating a plot in which the y-axis is so small that all the variation is squeezed into a tiny space. This makes variation a lot harder to see, which is rather nonsensical since the real purpose of graphs is usually to make the changes which are present visible.
A recent example is this graph from a post by Willis Eschenbach at WUWT about UAH TLT temperature data (from UAH, the Univ. of Alabama at Huntsville, for the lower troposphere). He “divided them at the Arctic and Antarctic Circles at 67° North and South, and at the Tropics of Capricorn and Cancer at 23° N & S.” But he decided to put all five latitude bands on the same graph, by stacking them, which requires a very small y-axis for each:
That certainly makes the changes a lot harder to see.
What would you think if I portrayed UAH TLT data for the southern extratropics with this graph?
Even some of the “fawning uncritical thinkers” who visit often might accuse me of trying to hide something. Yet, apparenly on the basis of this graph, Eschenbach says:
“Southern Extratropics? No trend.”
I’d rather look closer (here are annual averages):
I often like to add a bit of analysis:
As it happens, the linear trend is statistically significant at a warming rate of 0.01 deg.C/yr. With that in mind, read Willis Eschenbach’s summary again. True or false?
I’m far less interested in poking holes in Eschenbach’s post (they’re already gaping) than in objecting to the practice of tiny graphs. I know it takes work to make a lot of graphs (boy do I know it!), and I know it takes some care and experience to develop good habits which help readers by making things clear. But, helping readers by making things clear is the point.