Extreme Value Analysis of a Fish/Temperature Field Database

Abstract

Extreme water temperatures limit the presence of fish species in streams and lakes. Upper extreme water temperatures and their uncertainties are determined by several statistical methods from a large field database. There are over 140,000 weekly mean fish/stream temperature matched pairs in the database. Three different techniques are employed to estimate upper extreme habitat temperatures of 24 fish species. To quantify the uncertainty of the estimated extreme temperatures the bootstrap method, the method of moments and the residual method are applied. The data above the maximum growth temperature matched well by a type III extremal or a three-parameter lognormal distribution. Standard error of the estimated extreme habitat temperatures depends on species and varies from O.lC to 0.6C at the 95% cumulative probability of occurrence.