A new approach to the characterization of long-term changes in total atmospheric ozone: Determination and application of frequency distributions: Discussion

The extreme value analysis in the subject paper may be summarized as follows: More low values are observed during one 6 yr period than another, so there has been a change in the frequency of low values. The changes are not significant. The situation for distributional analysis is more complex. Had the distributional analysis simply provided information on the changes in the means and standard deviations of the total ozone values and presented the changes in the frequency of various ranges of these values from year to year, it would have been an interesting data article. Instead the authors attempted to relate their results to tropospheric ozone increases and stratospheric decreases. The posited relations are either trivial or not supported by their discussions. The methods have been misapplied and the results of their treatment of the data have been misinterpreted.