13 August 2009

Changing Perilous Assumptions to Suit the Analysis

In 2007 Michael Mann and colleagues published a paper (PDF) critical of work suggesting an undercount in storms from historical records, claiming that it was “perilous” to assume that there is a “fixed” relationship between landfalling and total hurricanes in the Atlantic basin:
Of course, estimation of undercount based on the assumption of a fixed relationship between total TC counts and the number of landfalling storms is perilous. Such an approach assumes, in particular, that the large-scale atmospheric steering which determines the trajectories of TCs once they’ve formed is constant, when there is in fact strong evidence that it is highly variable over time . . .
Now Mann and another set of colleagues (PDF) make what appears to be the exact opposite assumption in a paper just out in Nature, that landfalls are “in rough proportion” to overall basin activity:
We compared the sediment-based record against the above statistical estimate of basin-wide tropical cyclone activity (Fig. 3), guided by a working assumption that an appropriately weighted composite of regional landfalling hurricane activity varies, at multidecadal and longer timescales, in rough proportion to basin-wide tropical cyclone activity.
What is troubling is that the analysis in the second paper depends to some degree upon the first (that is, if there is a significant undercount, which Mann dismisses, then the nature of the relationships used in the second paper changes). I note that the recent paper shows a dramatic uptick in storm activity that has been convincingly refuted by "strong evidence that there has been no systematic change in the number of north Atlantic tropical cyclones during the 20th century." It would be interesting to see Mann's analysis run with observational data properly adjusted for undercount and short-duration storms, or at a minimum considering these factors as part of the uncertainties in the analysis.

At the minimum, the two Mann et al. studies rely on highly inconsistent assumptions, yet one analysis depends upon the other. Not good.