A New Procedure of Statistical Inference on Extreme Values

Abstract

Various methods of statistical inference on the extreme values in geophysical phenomena have been proposed by many authors, but not many statistical procedures to treat errors attendant to such inference have yet been offered.
In this pape r, the author proposes two simplified procedures to treat such problems statistically and applies these methods to meteorological data.
Distrib u tion of the largest order statistic and stochastic extrapolation of truncated transformation curve for normalization are quite suggestive for such problems of inference and so technical modifications of these statistical methods were tried in their application to meteorological data.