Could you plot year-on-year temperature change vs. CO2?
No. I use MATLAB/Octave, and Excel/CALC for that, each pairing being proprietary/open-source. MATLAB and Octave are more or less interchangeable, but I think it’s worthwhile avoiding features that work in one but not the other, especially since MATLAB has millions of users.
Excel/CALC are good because MS Office is widely deployed and Open Office is a free download, and almost everyone else has or can easily download MS’s free Excel reader if they just want to look at other people’s work.
MATLAB produces the nicest graphs if you use embedded postscript, but Excel reaches the largest audience when distributing stuff for examination (terrific for transparency, the whole world can see all the formulas and values creating the graphs, unlike GCM’s where the math is totally opaque, an intrinsic problem with GCM’s).
I would say the question of how to “do it right” with such software is open. My presentation at the AGU meeting in SF on Dec. 8 gives a couple of ways of going about it, along with how the choice of way can influence the answers to questions like “How hot will it get by 2100?”, “Between years Y and Y’, how much of the temperature variation was our fault?”, “How should climate sensitivity be defined and what is its value with each definition?”, “How important is it to separate the various natural contributors to climate change?” (what difference does it make if they’re all lumped together) and so on.
The answers depend on the method, suggesting that the questions can’t be answered precisely without a reason for preferring one method over another, which can quickly lead to arguments as we see here all the time.
But the extent to which they depend is of interest, because questions for which the extent is small have more precise answers than those whose answers depend very heavily on the method.