Which would you choose?

One of the main accusations leveled in a recently discussed video was that the data set was chosen, deliberately for the purpose of making the warming look bigger.

Suppose you needed to plot mid-troposphere temperature, but your real goal was to make the warming look as big as possible. Which data set would you choose?

An obvious choice would be satellite data from the Univ. of Alabama at Huntsville, sometimes called “UAH TMT” data. But that doesn’t start until December 1978. Is there anything prior to that?

Yes, there is. We have data for temperature at various altitudes from balloon-borne radiosondes. These have been collected and combined into a global average to form various data sets, including the well-known HadAT2 data.

But, the HadAT2 data gives temperature at very specific altitudes. What we want is a time series that represents the troposphere. We could get that by computing a weighted average of the measurements at specific altitudes, but then we’d have to know the right weights to apply to get an “equivalent temperature” corresponding to the satellite channels. Well, the HadAT2 people have been kind enough to provide those weights so we can do the calculation. But we don’t even have to do that — because they’ve already done the calculation for us.

So: there’s at least one other choice if you want to plot temperature in the troposphere, a choice which doesn’t require satellite observations so it goes further back in time than 1978. It starts in 1958.

But the question remains: which would you choose if your purpose was to make the warming look as big as possible?

Let’s plot the HadAT2 data for “T2-equivalent temperature,” and the UAH TMT data, on the same graph. I’ll offset the HadAT2 data to align it with the UAH data, and I’ll add trend lines estimated by least-squares regression:

Remembering that the purpose is to make the warming look as big as possible … which would you choose?