The Earth's climate is changing at an alarming rate, resulting in more frequent and intense heatwaves globally. Western Europe has experienced a disproportionately rapid increase in summer temperatures and heat extremes over the past two decades, exceeding the warming observed in other mid-latitude regions. This accelerated warming has led to several unprecedented heatwaves, including the 2003 European heatwave, the record-breaking temperatures in Northwestern Europe in 2018, and the intense heatwaves of 2019 and 2022, which saw all-time temperature records broken in numerous locations. While climate projections indicate the possibility of such extreme events, the observed pace of increasing heatwave magnitude in Western Europe is generally not reflected in these models. This discrepancy between observed trends and climate model simulations poses a significant challenge to accurately projecting future heat extremes and developing effective adaptation strategies. This research aims to investigate the causes of this mismatch, focusing on the role of atmospheric circulation changes in amplifying the trend of extreme heat in Western Europe.

Previous research has explored various explanations for the disproportionate warming trends in Western Europe. For mean summer temperatures, studies have considered factors such as changes in mean atmospheric circulation, aerosol effects, and changes in early summer soil moisture and associated feedbacks. Regarding extreme heat, the increased frequency and persistence of split mid-latitude jet states, potentially linked to a weakening of the mean summer zonal circulation, have been identified as a potential contributor to amplified heatwave intensity trends. Changes in atmospheric circulation patterns favorable to heat accumulation over Western Europe, such as a positive trend in a dipole structure with low pressure over the Eastern Atlantic and high pressure over the Mediterranean extending towards Central Europe, have also been highlighted. However, studies on blocking events over Scandinavia haven't yielded robust conclusions, and the role of Rossby waves remains debated due to sensitivity in their definition. Existing studies indicate potential roles of dynamical factors, but a comprehensive understanding of the mechanisms driving the observed accelerated warming of extreme heat in Western Europe remains elusive.

This study employs a methodology based on circulation analogues to assess the contribution of atmospheric circulation changes to trends in maximum and mean daily maximum temperatures (TXx and TXm). The analysis uses ERA5 reanalysis and E-OBS interpolated observations for the period 1950-2022. Regional atmospheric circulation patterns are characterized using the 500 hPa streamfunction. Circulation analogues for a given day are identified by searching for other summer days with similar anomaly characteristics, measured by the anomaly correlation coefficient (ACC). This allows for the calculation of statistics conditional on a given circulation and assessment of the role of dynamical changes in circulation-conditioned variables. To estimate the dynamical contribution to TXx and TXm trends, the daily temperature field for each day is replaced with the temperature field from a day with an analogous circulation pattern. This process, in principle, removes thermodynamical effects of warming, although a correction for non-stationarity based on regression between original and analogue time series is applied. The dynamical contribution is then evaluated using linear regression with the Global Mean Surface Air Temperature (GSAT). The study also employs a second method, "dynamical adjustment," using ridge regression to estimate the contribution of circulation to temperature variability, providing a validation for the analogue approach. The sensitivity of the results to the analogue approach is also examined by varying the analysis domain. Additionally, climate model simulations from CMIP6 are analyzed to compare the observed trends with modeled trends. Analogue techniques are applied to all available CMIP6 simulations with available 500 hPa fields, and the dynamical trends are compared to the observations. A multiple testing procedure, the False Discovery Rate (FDR), is used to evaluate the statistical significance of the findings. Finally, the thermodynamic contribution to the trends is calculated as a residual by subtracting the dynamical trend from the total trend.

The study's key findings highlight a significant discrepancy between observed and modeled trends in extreme heat over Western Europe. The analysis of ERA5 and E-OBS data reveals that TXx trends in Western Europe exceed 5°C/GWD (degrees Celsius per global warming degree) in northern France and Benelux, reaching an average of 3.4°C/GWD over the broader study area (45°N-55°N, 10°W-15°E). The dynamical contribution to TXx trends, estimated using the circulation analogue method, is substantial, reaching approximately 1.5°C/GWD in several areas and averaging 0.8 °C/GWD (0.2°–1.4 °C) across Western Europe. This substantial dynamical contribution is significantly underestimated in all 170 CMIP6 climate model simulations analyzed. The dynamical adjustment method provides consistent results, although with slightly weaker dynamical contributions. Only a small fraction of model simulations (less than 1%) show dynamical TXx trends comparable to the observations, indicating a significant model-observation mismatch. While thermodynamic contributions to trends are generally consistent between models and observations for TXm, there is a tendency for an underestimation of TXx thermodynamic trends in the models. The observed rapid increase in the frequency of southerly flow patterns, which contribute to increased heat, is largely not captured by the CMIP6 simulations. This indicates that the large observed warming of extreme heat in Western Europe is not well represented by the current generation of climate models, largely due to an underestimation of the dynamical contributions. The observed trends are statistically significant based on the FDR procedure.

The findings of this study indicate that current climate models significantly underestimate the rapid warming of extreme heat in Western Europe. The large observed dynamical contribution to the trend, which is not captured by the simulations, explains a significant part of the model-observation discrepancy. The mismatch between observed and simulated dynamical trends could result from either an underestimation of the forced regional response of models to greenhouse gases, or an underestimation of low-frequency variability in the models. If the mismatch is due to an inaccurate representation of the forced response, then future projections may be overly conservative. If the mismatch stems from underestimation of internal variability, the level of uncertainty regarding the future pace of summer heat in Europe remains high. Therefore, caution is needed when interpreting climate model projections for adaptation and resilience planning in Western Europe. These findings extend previous research highlighting similar model-observation discrepancies for wintertime weather in Europe, underscoring the need for improved model representation of atmospheric circulation dynamics.

This research demonstrates a significant underestimation of the observed rapid warming of extreme heat in Western Europe by CMIP6 climate models. This underestimation is primarily attributed to the models' failure to capture the substantial dynamical contribution from changes in atmospheric circulation, specifically the increased frequency of southerly flows. The results emphasize the need for improved model representation of both forced responses and internal variability in atmospheric circulation to accurately project future heat extremes. Further research should focus on identifying the underlying causes of this model-observation mismatch and improving climate model representation of these critical processes to enhance the reliability of future climate projections.

While this study provides valuable insights into the mismatch between observed and modeled heat extremes in Western Europe, some limitations should be noted. The analysis is focused on a specific region, and the findings may not be generalizable to other parts of the world. The methodology relies on the assumption of linear relationships between variables, which may not fully capture the complexity of climate dynamics. The study's findings are based on currently available climate models, and future improvements in model resolution and physics may yield different results. The selection and quality of circulation analogues may also introduce uncertainties into the analysis.