River water temperature quantiles as thermal stress indicators: Case study in Switzerland

Abstract

The variability of the river water thermal regime has important consequences on the environment and aquatic habitat. In 25 independent and identically distributed stations in Switzerland, local frequency analysis is used to quantify extreme river temperatures. Probability distributions are fitted to the data to estimate maximum water temperatures corresponding to different return periods. The goodness of fit of statistical distributions are evaluated using the Akaike and Bayesian Information Criteria. L-moment ratio diagrams are also used to validate the choices of appropriate candidate distributions. Results show that for high altitude stations the two-parameter Weibull (W2) distribution is the most adequate distribution to represent extreme river water temperatures while for low altitude stations the most commonly selected distributions are the Normal (N) and Inverse Gamma (IG). The L-moment ratio diagrams confirm the results of the local frequency analysis. These results point to the presence of a regional homogeneity in the thermal regime of the study area. River temperature quantiles are compared to know thresholds above which thermal stress occurs for a relatively ubiquitous salmonid species in Europe (Brown trout).

Keywords

• Extreme river temperature
• Probability distribution
• Local frequency analysis
• L-moment ratio diagram
• Discordancy measure
• Goodness-of-fit criteria