Odd Citizen
An Odd Citizen’s Search For Vanishing Freedoms

Nov

23

Eee-Gad! A Simpler Climate Model That Actually Works?

Filed Under Uncategorized, Wisdom

While stumbling around the climate blogosphere I came across this, just published report:
Simple Model Leaves Expensive Climate Models Cold

The report describes a simple climate model that seems to actually work and compares it to the complex U.N. IPCC models that don’t work.

Incidentally, this new model predicts 0.5 degree C. rise in temps over the next 100 years, but says this may be an artifact of measurement error. Nothing to worry about in any case.

Maybe someone will be able to elaborate on the technology and mathematics involved. (See link in cited article, above.)

Just thought you might be interested.

Comments

One Response to “Eee-Gad! A Simpler Climate Model That Actually Works?”

  1. bkalafut on November 24th, 2009 1:50 am

    I think I can sum this one up for you in plain English:

    Green et al are comparing the cumulative relative absolute error (vertical distance from forecast curve, summed over all years–equal to the area bounded by the graph of measurement and the graph of the forecast curve, with vertical cuts at both ends–you could actually compute this with paper, a scale, and an x-acto knife!) of a horizontal line and a linear projection contained in the 1992 IPCC report.

    Green et al find that the cumulative error for their flat line is 5% greater than that for the 1992 IPCC linear projection.

    There’s nothing more to it than that. Compare integrated (summed) absolute errors of one curve to integrated absolute errors of another.

    This review process for this paper appears to have been soft. It would never have made it into a natural science journal in its current form, and one suspects that the journal to which it was submitted didn’t select reviewers who could be critical about the natural science content. Were I refereeing this paper, I’d recommend that the authors re-submit it after major revisions for the following reasons

    (1) Comparison to the 1992 projection line is worthless; that result has long been superseded. (It’s also far off the AR3 and AR4 results!) The authors ought to compare their result to either the “multimodel ensemble” of the IPCC AR4 or not perform this exercise at all. Work, in other words, lacks timeliness or relevance.

    (2) The literature review is insufficient. It covers neither the foundations of the 1992 projection line nor more recent work. Most importantly, it contains no discussion of the difference between purely empirical projections and physical models. Indeed Green et al could be said to be arguing from equivocation–and they’ve been called out on this before.

    (3) Argument for the flat-line benchmark is fatuous. It is clear that the “noise”, if we want to treat it as such, in the 800,000 year curve is extremely autocorrelated, over periods of tens of thousands of years. A reasonable literature review or even application of “freshman”-level science knowledge would have revealed that this is to be expected. On the one hand, this means that a flat line is inappropriate over a long time scale. On the other, it means that a flat line may be appropriate over short time scales, but this is not the argument the authors make.

    (4) Authors justify testing only against the 1992 line by stating “To test any forecasting method, it is necessary to exclude data that were used to develop the model; that is, the testing must be done using out-of-sample data. The most obvious out-of-sample data are the observations that occurred after the forecast was made.”

    A real, responsible literature review would have revealed that temperature data are not used to develop the physical contained in the IPCC reports, aside from setting the initial state. This concern is thus superstition, taking us back to point 1: Authors could have compared their baseline to any physical model over the entirety of the reliable global instrumental record.

    (5) Use of “benchmark” in this context is obscurantist. Would be better to say that they are comparing to a no-change hypothesis or something similar.

    (6) Footnote 3 is irrelevant in the context of natural sciences; the authors either do not understand the distinction between physical and statistical models or are arguing from equivocation.

    (7) The entire 1851-onward discussion is based on an obvious false premise: that of a constant warming rate, which would be a foolish projection.

    (8) The Tierney anecdote is largely irrelevant, perhaps misdirection, and ill suited to the tone and style of a formal scholarly journal.

    (9) Discussion of the role of possible systematic error of measurement (terminating in a citation of McKitrick and Michaels) needs to be done in a quantitative manner.

Leave a Reply

You must be logged in to post a comment.