1. Introduction

For most of the Eurasian continent and North America, the summer 2022 was unparalleled in temperature, characterized by persistent heatwaves and droughts over Europe, the United States, and China (Lu et al. 2022). The summer heatwave over Europe featured record-breaking temperatures in many countries across western Europe and lead to more than 60 000 heat-related deaths (Ballester et al. 2023). The Copernicus Climate Change Service (C3S) ranked it the hottest European summer on record with average temperatures being 1.34°C higher than the 1991–2020 climatology (Copernicus 2022). Many of the 2022 local European heatwaves were characterized by long durations and large spatial extensions with regionally intense heat and drought periods (Imbery et al. 2023; Lentze 2023). The exceptional summer 2022 started off with high temperatures in southern Europe in May, leading to new temperature records, for instance, in France and Portugal (Copernicus Observer 2022). A number of heatwaves followed in June and continued into July and August, spreading from southwestern Europe northeastwards toward the United Kingdom and Scandinavia. Extreme temperatures of over 40°C were recorded in Hamburg, Germany, on the 20 July. This was the first time that temperatures exceeding 40°C were measured north of 53°N in central Europe (Imbery et al. 2023). The World Weather Attribution project states that temperatures of more than 40°C in the United Kingdom were extremely unlikely without anthropogenic influence (Zachariah et al. 2022; Christidis et al. 2020). Persistent high pressure over western Europe in combination with advected hot air masses from North Africa, moving northeastwards from west of Portugal ahead of a trough, caused heatwaves over large areas of western Europe (Copernicus Observer 2022). The summer heatwave 2022 is exemplary for the latest IPCC report results, which state that heatwaves have become not only more intense and persistent but also more frequent in recent decades (Seneviratne et al. 2021). The last decades revealed a rapid shift toward hotter summers in Europe (Lhotka and Kyselý 2022). Future scenarios project a further increase in frequency, duration, and intensification of heatwaves over most land areas [for an overview, see Barriopedro et al. (2023)] due to global warming. It is crucial to understand how human activities have influenced these heatwaves for both mitigation and adaptation.

In this article, we apply a holistic attribution approach combining long-term statistics and event-based storylines, allowing for both an attribution to anthropogenic climate change and a historical classification. We attribute the summer heatwave over western Europe 2022 to anthropogenic climate change using spectrally nudged storylines (van Garderen et al. 2021; van Garderen and Mindlin 2022). With this method, van Garderen et al. (2021) showed that the 2003 heatwave in Europe was 0.6°C and the 2010 heatwave in Russia was even 1°C stronger due to anthropogenic climate change. The storylines enable us to quantify the anthropogenic climate change influence on the event as it was observed, for past, present, and future climate states, despite the presence of possible large-scale natural variability. To put the heatwave into historical context, we analyze ERA5 (Hersbach et al. 2020) reanalysis data and a 2000-yr paleo simulation using a sophisticated clustering method (Philipp et al. 2007; Hansen and Belušić 2021), so that long-term statistics add value to the event-specific attribution method.

2. Heatwave attribution using spectrally nudged storylines

Three storyline ensembles of five ensemble members each were created to simulate the heatwave 2022. They represent a preindustrial, a present-time, and a future +2°C state. The main differences between the storylines are changing sea surface temperatures and greenhouse gases, since these are well known in their response to anthropogenic climate change. The spectrally nudged storylines were simulated with the atmospheric general circulation model ECHAM6 (Giorgetta et al. 2013), which was spectrally nudged toward NCEP–NCAR reanalysis 1 (Kalnay et al. 1996; Kistler et al. 2001). Spectral nudging is applied solely for vorticity and divergence at higher atmospheric layers for large-scale weather patterns. As the dynamic weather situation is constrained by spectral nudging, the method focuses on thermodynamic aspects of climate change. The technique enables us to analyze the very same weather situation in different climate states represented by the storylines. Five ensemble members of each storyline were simulated, deviating just in their spinup starting dates, to show the robustness of the results. For 2022, the present-time and preindustrial storylines show a global mean 2-m temperature difference between their ensemble means of 0.96°C. This is 0.24°C smaller than the observed temperature increase due to climate change since industrialization. As solely sea surface temperatures and greenhouse gases were changed between the storylines, other factors, including natural variability, were neglected, which may have caused the smaller increase. The method can thus be regarded to represent climate change in a more cautious way, as the results are probably not overestimating the anthropogenic part of climate change.

Figure 1 (left) shows near-surface temperatures and geopotential height at 500 hPa for summer (JJA) 2022 and temperature differences between present-time and preindustrial simulations. The heatwave is clearly visible in all months and present-day storylines, showing it spreading from southern Europe northward. The preindustrial storylines show a similar though less intense heatwave pattern. The +2°C scenario shows a similar pattern with an even more intensified heatwave (Fig. S1 in the online supplemental material). The stippling marks grid points where all present-time storyline members are >0.1°C above all preindustrial members. There is large agreement between the individual storyline ensemble members of present-day and preindustrial and between +2°C and preindustrial (Fig. S1) storylines, and most areas are stippled which underpins the robustness of the method.

View Full Size

(left) The left column shows the present-time 2-m temperatures (°C) (shaded) and geopotential height at 500 hPa (m) (black contour lines) of the European heatwave 2022 for JJA. The right column shows the differences in 2-m temperature (°C) between the present-time and preindustrial simulations as shaded fields. Stippling indicates where all present-time members are >0.1°C above all preindustrial members for that grid point. (right) Daily mean 2-m temperature (°C) averaged over Europe (35°–55°N, 10°W–25°E) for 2022 for the present-time (red), preindustrial (blue), and +2°C (yellow) simulations, ERA5 (black stippled line), and NCEP (black solid line) reanalysis data. The climatology (gray shaded area) is the 5th–95th percentile range between 1985 and 2015 calculated with ECHAM6 (Schubert-Frisius et al. 2017).

Citation: Bulletin of the American Meteorological Society 105, 11; 10.1175/BAMS-D-24-0017.1

• Download Figure
• Download figure as PowerPoint slide

The right side of Fig. 1 shows near-surface European summer 2022 temperature time series for all storyline members and ERA5 and NCEP reanalysis in relation to a 30-yr climatology (Schubert-Frisius et al. 2017). The time series show a clear separation between the different storylines and only very small differences between ensemble members. The present-day runs show higher values than the reanalyses for the most extreme summer period. All storylines show a warm bias for global mean temperatures in comparison to observations during JJA. However, the focus of the attribution is here on temperature differences between the storylines and not on absolute values. The heatwave occurred for present day, preindustrial, and the +2°C simulations, indicated by the clear exceedance of the climatology range. This shows that the heatwave would have occurred also without anthropogenic climate change, but it would have been less intense.

3. Long-term statistics

ERA5 data and a 2000-yr paleo simulation (MPI-ESM_past2k; Jungclaus et al. 2014) serve as databases to put the heatwave 2022 into historical context. A cluster analysis [simulated annealing and diversified randomization (SANDRA); Philipp et al. 2007; Hansen and Belušić 2021] for maximum temperature anomalies, related to the 90th percentiles of the years 1950–2022, reveals the most dominant ERA5 clusters for the 2022 heatwave (Fig. 2, left side). These are the western Europe (Eur_W), the northeastern Europe (Eur_NE), the southeastern Europe (Eur_SE), and the northern Europe (Eur_N) clusters. The paleo data were assigned to the ERA5 clusters to make the results comparable, whereby the clusters in the paleo simulation were very similar to the ones of ERA5.

View Full Size

(left) ERA5 daily maximum 2-m temperature anomalies (°C) for the four most dominant European heatwave clusters in summer (JJA) and their cluster occurrences in percent. (right) Boxplots for days of cluster occurrence during JJA, left for ERA5 (1950–2022) and right for MPI-ESM_past2k (1–2014) for no heatwave cases (no) and the four clusters. Symbols show days of cluster occurrence for the extreme heatwaves of 2003 (blue square), 2010 (orange triangle), 2022 (red star), and 2023 (yellow circle). The boxes extend from the lower to the upper quartile, and the orange line shows the median. The range of the 95th percentiles is indicated by the whiskers.

Citation: Bulletin of the American Meteorological Society 105, 11; 10.1175/BAMS-D-24-0017.1

• Download Figure
• Download figure as PowerPoint slide

Most interestingly, the “no heatwave” cases are much less frequent (only 2 days are not assigned to any cluster) in JJA 2022 in comparison to all other years (red star in Fig. 2, right side). Even in the paleo run, there are only two summers with a lower number (1 day) of no heatwave days than in 2022. This accounts to a percentile of 0.15 for the no heatwave cases in the model distribution. The high number of days that can be allocated to heatwave clusters indicates the large spatial extension and duration of this extreme event. The heatwaves with the largest magnitude between 1950 and 2021 (Lhotka and Kyselý 2022), namely, 2003 and 2010 (blue square and orange triangle), were plotted in comparison to the 2022 heatwave (red star). In addition, the heatwave 2023 was plotted (yellow circle). According to this analysis, 2023 was not among the most extreme European heatwaves. However, the method does not differentiate between the highest intensities but has its focus on the number of days with high temperature anomaly percentiles which can be assigned to heatwave clusters. For the four clusters, few individual summers over the past 2000 years show higher frequencies than 2022. It was the most extreme year of the Eur_W cluster, took second place in Eur_N and third place for the Eur_NE cluster in ERA5. In the multimillenial simulation, the summer of 2022 is even exceeding the 95th percentile in all but the Eur_SE clusters.

4. Summary and conclusions

The European summer heatwave 2022 was exceptional in its intensity, duration, and spatial extent (Copernicus Observer 2022). Large-scale and persistent high temperatures, both over land and over sea, in combination with local droughts characterized the extraordinary heatwave. Both methods showed that the heatwave was an extreme event, in terms of temperature differences between present-day, preindustrial, and +2°C times and in its statistics. According to our results, the anthropogenic contribution was crucial for the high temperatures of the extreme heatwave of summer 2022.

The major conclusions of this study are as follows:

• The anthropogenic contribution to the heatwave was 1.25°C on average and regionally up to 5.7°C (30 July over the Balkans) with largest temperature differences over southern Europe in July and even more intense, widespread deviations all over Europe in August. For a +2°C state, the anthropogenic contribution would be 2.21°C and regionally up to 7.68°C (4 August over Poland). The historical perspective shows that the extreme heatwave featured the lowest number of no heatwave days on record in ERA5 reanalysis data. For the four dominant heatwave clusters, very few individual summers show higher frequencies than 2022. The previous most extreme heatwaves 2010 and 2003 exhibit slightly higher values for some of the clusters, but the large spatial extent and persistence of the temperature anomalies in summer 2022 were exceptional.

• A paleo simulation confirms these results. For the “no heatwave” cases and all but one clusters, the summer 2022 even exceeds the 95th percentiles of the multimillennial climate simulation.