NZSM Online

Get TurboNote+ desktop sticky notes

Interclue makes your browsing smarter, faster, more informative

SciTech Daily Review

Webcentre Ltd: Web solutions, Smart software, Quality graphics


Returning Drought

In a recent letter [October 1992], Kieran Devine suggests my article "The 100-Year Drought?" [August 1992] contains an over-estimate of the error involved in the return period of the recent drought, due to the fact that I ignored the supposed information contained in smaller droughts in the available record. My argument that it is only the data of extreme droughts that tells us anything about extreme droughts is as follows:

Firstly, the facts speak for themselves. Evidently, prior to 1992 Electricorp estimated that droughts of the 1991/92 magnitude had a 100-year return period. Presumably the "information" contained by the smaller droughts was incorporated in this.

Subsequent to this drought actually happening, its return period was revised downward (it was not stated to what value, but the revised estimate would be unlikely to be much more than 50 years). This indicates (i) as of now, just about all the information relating to droughts of the 1991/92 size is contained by the single data value of the 1991/92 drought, (ii) the pre-1992 return period estimate was subject to something like a 50% error (and there is no reason why the revised estimate should not be subject to a similar percentage error).

Secondly, the concept of using smaller droughts to gain information about extreme droughts is based on the unjustifiable assumption that we have a unique means of extrapolation beyond the largest droughts in the record. Having adopted some extrapolation mechanism (such as a probability distribution), the apparent narrowing of the uncertainty in the return period estimates of extreme events is simply an illusion generated by the statistical packages.

This false accuracy arises because it is not possible to incorporate the human factor implicit in the personal selection of an extrapolation technique -- different extrapolations can yield very different estimates of the return periods of extreme events. The approach in the article may have been simple (not simplistic), but it has the distinct advantage of meeting the important scientific criterion of objectivity.

Actually, Electricorp were being a little misleading by insisting throughout the electricity crisis that they were dealing with the consequences of a 100-year event, which implies little more than bad luck. Had the revised return period been announced earlier, as it could have been, then questions might have been raised as to whether Electricorp's operating procedures took into account the large statistical error associated with their original return period estimate.

Earl Bardsley, Hamilton