On the statistical distribution of temperature and the classification of extreme events considering season and climate change—an application in Switzerland

With the increased occurrence of hot spells in recent years, there is growing interest in quantifying the recurrence of extreme temperature events. However, pronounced temperature anomalies occur all year round, and a reliable classification in terms of the time of occurrence in the year is needed. In this study, we present a novel approach to classifying daily air temperatures that take into account the seasonal cycle and climate change. We model the distribution of the daily Swiss temperatures using the skewed generalized error distribution with four time-varying parameters, thereby accounting for non-Gaussianity in daily air temperature, while the climatic trend is modeled linearly with smoothed northern hemisphere temperature as an explanatory variable. The daily observations are then transformed into a standard normal distribution. The resultant standardized temperature anomalies are comparable within a year and between years and are used for quantile-based empirical classification. The approach is suitable to classify historical and current extreme temperatures with respect to the temperature range expected at the time of the event. For example, a heat wave occurring at the end of June is classified as less likely to occur than a heat wave of similar intensity occurring in mid-July, as is shown for the two 7-day heat waves that struck Switzerland in the summer of 2019. Furthermore, climate change has increased the probability of hot events and decreased the probability of cold events in recent years. The presented approach thus allows a fair classification of extreme temperatures within a year and between years and offers new possibilities to analyze daily air temperature.