On the Issue of Informativeness of Mass Spectral Diagnostic Relations and Criteria Based on Them for Testing Geochemical Hypotheses

Abstract

The quantitative evaluation of the errors of the 1st and 2nd kind for mass spectral criteria is obtained. These errors have been earlier proposed for distinguishing oils generated in carbonate and other rocks for the Domanic formation of the northern and central Volga−Ural regions. It is shown that none of them work: the best criteria have the 1st and 2nd kind errors close to 50%. Consequently, for all regions where carbonate oil source rocks are possible, a work similar to this should be done primarily, and only after such study, the analysis of oils becomes meaningful. It is also necessary to obtain quantitative estimates of the quality of all the other criteria used in petroleum geochemistry, for which such assessments are technically possible. New criteria should be designed only on the basis of those composition parameters for which the value of the Mann–Whitney test is close to zero.