Climate variability on Fit for 55 European power systems

scenario). Beyond hydropower, other renewable energy sources like wind and solar show rela- tively low variability, as the annual capacity factors for these sources can be predicted with rela- tively high precision [8]. We further measure the relationship, as expressed with a linear regression, between the penetration of renewable electricity sources and emission intensity (Fig 2). This analysis indi- cates how the variability of a specific source of electricity can affect the emissions of the regional power systems. As expected, we observe a strong relationship between the share of renewable energy sources in a region’s electricity generation and the CO2 intensity of that elec- tricity. The magnitude of the effect is stronger for variable renewable energy sources wind and solar than for hydropower, related to the relative (in)flexibility of the generation source and its position in the merit order curve. The results are in line with [10], who focus on the analysis of the variability of European power systems from a similar perspective, especially regarding the variability of CO2 intensity and wind and solar generation. We observe two key areas of divergence in the literature affect- ing these power system variables: 1) the use of different scenarios (for example the European Commission MIX 2030 versus the ENTSO-E visions scenarios from the TYNDP 2016 in [10] and 2) differing approaches to the variability of renewable energy sources (for example the lack of hydropower in [10] introducing major differences in the supply/demand balance in hydropower-rich regions.

3.2 Co-variability We define co-variability as how the change in one region relates to changes in the rest of the continent. Co-variability happens mainly for two reasons: 1) European power systems are inter-connected and can exchange electricity and 2) the meteorological phenomena influenc- ing the power systems (e.g., wind lulls) are often large-scale, large enough to affect multiple regions (if not the entire continent) at the same time. The co-variability between two variables is measured using the Spearman rank correlation coefficient and, where necessary, a confi- dence interval of the value is calculated using a Fisher z-transformation, following the imple- mentation provided by [46]. We analyse the co-variability starting from the year-over-year Fig 2. Linear regression showing the relationship between penetration of electricity sources and the CO2 intensity (measured in gCO2/kWh). Only groups with linear regression p<0.001 are shown. https://doi.org/10.1371/journal.pone.0289705.g002 PLOS ONE Climate variability on Fit for 55 European power systems PLOS ONE | https://doi.org/10.1371/journal.pone.0289705 December 20, 2023 6 / 16