Non-stationary Generalized Extreme Value Distribution Analysis of Temperature Extremes for the Effects on Electrical Power Infrastructure

Power infrastructure, like power plants, power stations, overhead lines, etc., might be strongly altered or destroyed  by the effect of surrounding air temperature. Either affecting the efficiency of power generation, in case of power plants, or its subsequent transport by influencing the thermal rating of power lines. Extreme temperature events, e.g., the annual maximum surface air temperature, are of interest, since they are representative of the maximum thermal stress from the environment that infrastructure should ideally be capable of withstanding. Additionally, they are usually events that coincide with exceptionally high energy demand, too, due to cooling by air conditioning or electric heating in case of annual temperature minima.

As a result of a changing climate towards hotter average air temperatures, knowledge of the statistics of temperature extremes is relevant to ensure reliable operation of existing infrastructure and to asses the operation environment of potential future assets. The methodology followed in this study is purely statistical [1]. It is the best available methodology for predicting the statistics of extremely rare events based on both observation datasets and the best available climate model outputs. It involves the fit of a non-stationary generalized extreme value distribution (GEV) using the software package ANKIALE [2] using a Bayesian setup. The parameters of the GEV distribution are determined as follows in this setup: First, an a priory distribution from data of 28 CMIP6 models is created. Next, using measurement records, namely the E-OBS dataset in version 31.0e [3], this a priori estimate is then constrained by observations to obtain the final a posteriori distribution of the GEV parameters. It is worth noting that the employed Bayesian approach provides uncertainty or error estimates on the obtained parameters, too, allowing to make statements about the reliability of the predictions.

Using this method, a comprehensive dataset for the non-stationary GEV distribution parameters over Europe was created encompassing the period from 1850 to 2099. For future years, data for the climate scenarios SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5 is available. To facilitate an easy evaluation of the data, a complementary data viewer software was developed. The software visualizes spacial maps of Europe for summary statistics of the GEV distribution including their uncertainty. A relevant quantity for power infrastructure of the GEV distribution is for example the upper bound, which can be interpreted as the most extreme temperature that is statistically possible.

[1] Robin, Y. and Ribes, A.: Nonstationary extreme value analysis for event attribution combining climate models and observations, Adv. Stat. Clim. Meteorol. Oceanogr., 6, 205–221
[2] Robin, Y., Vrac, M., Ribes, A., Barbaux, O., and Naveau, P.: A Bayesian statistical method to estimate the climatology of extreme temperature under multiple scenarios: the ANKIALE package, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2025-1121, 2025.
[3] Cornes, R. C., van der Schrier, G., van den Besselaar, E. J. M., & Jones, P. D. (2018). An ensemble version of the E-OBS temperature and precipitation data sets. Journal of Geophysical Research: Atmospheres, 123, 9391–9409.