Annual probabilities of extreme air temperatures in the past and future climate of Finland

Both extremely low and high temperatures can be relevant for the operation of nuclear power plants (NPPs). For example, temperatures higher than 40 °C might endanger safe shutdown of an NPP (STUK, 2011). The MAWECLI project aims to provide reliable estimates about the likelihood of extreme events affecting NPPs in the changing climate of Finland.

In this study, we evaluate the annual probabilities of low and high temperature extremes in the past and future climate of Finland. To produce estimates for events with low annual probabilities (2%, 1%, 0.2%, and 0.1%), we utilise statistical extreme value analysis, employing both the generalized extreme value distribution with block maxima approach and the generalized Pareto distribution with peak-over-threshold approach. Furthermore, any potential non-stationarity in the time series is addressed by utilising distribution parameter covariates based on linear trends identified from the time series.

For the past, the extreme value analysis is performed for 35 weather stations across Finland. The analysis is based on a 60-year period of observations (1963-2022), which is common for all stations, as well as longer station-specific historic records, starting from 1844 for the longest operational weather station. In addition to instantaneous extreme temperatures, estimates are also calculated for high (or low) temperatures that have prevailed for time periods of six and 24 hours.

For the future projections, bias-corrected daily mean temperature data from 25 CMIP6 global climate models (GCM) is utilised (Ruosteenoja and Jylhä, 2023). To produce extensive data sets for 20-year periods representing current (1999-2018) and mid-century (2041-2060) climatic conditions, 20-year time series of each climate model were combined to create a 500-year ensemble depicting the climatic conditions of each period. The use of this 500-year ensemble enables more robust estimates in comparison to extreme value analysis relying on only 20-year data periods for each individual GCM. Our results show higher warming for extremes compared to changes in mean temperatures, the difference increasing towards smaller annual probabilities, i.e., more rare events.

References:

Ruosteenoja, K., Jylhä, K. Average and extreme heatwaves in Europe at 0.5–2.0 °C global warming levels in CMIP6 model simulations. Clim Dyn 61, 4259–4281 (2023). https://doi.org/10.1007/s00382-023-06798-4

STUK, 2011. European Stress Tests for Nuclear Power Plants. National Report. Finland 3/0600/2011 2011 Tomi Routamo (ed.) Radiation and Nuclear Safety Authority 2011. Available at: https://www.ensreg.eu/sites/default/files/EU_Stress_Tests_-_National_Report_-_Finland.pdf