Attribution of the 2017 Southeast Europe summer heat
This method describes the analysis performed for the writing of this article in The Conversation.
Early August has seen extreme heat across the southeast of Europe in what's been dubbed the "Lucifer" heatwave. At the time of writing, this heatwave is ongoing, so performing a rapid attribution analysis poses its own challenges. Here I describe the analysis I have done following the methods of previous peer-reviewed literature, specifically Lewis and Karoly (2013) and King et al. (2015) for the attribution component, and King et al. (2017) for the 1.5°C and 2°C analysis.
Firstly, I used the daily E-OBS dataset v15.0 (Haylock et al. 2008) for 1950-2016 to construct the observed series. I interpolated the dataset onto a regular 2° grid. I then found the hottest summer (June-August) day at each gridbox for each year and took an area average over the southeast Europe region (8°E-24°E, 36°N-50°N). This method bears the closest similarity to the calculation of the TXx index (representing the hottest daily maximum temperature in each summer) that was also computed in the CMIP5 model simulations (Sillmann et al. 2013a,b).
The region was chosen to include the area most strongly influenced by this heatwave and includes Italy, Greece, the Balkans and extends up to southern Poland and the Czech Republic.
Temperature anomalies from a 1961-1990 climatology were calculated in both the E-OBS series and the model simulations (relative to that period in the historical runs).
Nine CMIP5 models with all required model data available for the TXx variable were evaluated against the E-OBS dataset using a Kolmogorov-Smirnov test. Models where at least two-thirds of historical simulations were not statistically dissimilar (i.e. p>0.05) to the E-OBS series for the common 1950-2005 period were used in further analysis. Eight models passed this test (CanESM2, CNRM-CM5, CSIRO-Mk3.6.0, GFDL-CM3, IPSL-CM5A-LR, IPSL-CM5A-MR, MIROC-ESM, and NorESM1-M) with one failing (MRI-CGCM3).
The attribution analysis was conducted using three thresholds based on the E-OBS series: the 90th percentile (equivalent to a historical one-in-ten year event), the 95th percentile (equivalent to a one-in-20 year event), and a the existing record of 2007 (corresponding to a new record). These thresholds correspond to temperature anomalies of +2.85°C, +3.06°C, and +3.86°C respectively. As the event is ongoing it is not known which threshold is best to use at this time.
The likelihood of extreme heat exceeding each of these thresholds is compared for the natural world simulations (histNat; 1901-2005) and the current world (RCP8.5; 2006-2026). To estimate the change in likelihood, and associated 90% confidence interval, the pairs of complete CMIP5 simulations are 50% bootstrap resampled 10000 times, and the change in likelihood of extreme heat is recalculated. A conservative 10th percentile value is used to say that extreme heat events like this one are at least four times as likely due to human-induced climate change.
This process is repeated for the 1.5°C and 2°C global warming worlds (see King et al. 2017 for details).
Overall, a clear climate change signal is found in this event.
Early August has seen extreme heat across the southeast of Europe in what's been dubbed the "Lucifer" heatwave. At the time of writing, this heatwave is ongoing, so performing a rapid attribution analysis poses its own challenges. Here I describe the analysis I have done following the methods of previous peer-reviewed literature, specifically Lewis and Karoly (2013) and King et al. (2015) for the attribution component, and King et al. (2017) for the 1.5°C and 2°C analysis.
Firstly, I used the daily E-OBS dataset v15.0 (Haylock et al. 2008) for 1950-2016 to construct the observed series. I interpolated the dataset onto a regular 2° grid. I then found the hottest summer (June-August) day at each gridbox for each year and took an area average over the southeast Europe region (8°E-24°E, 36°N-50°N). This method bears the closest similarity to the calculation of the TXx index (representing the hottest daily maximum temperature in each summer) that was also computed in the CMIP5 model simulations (Sillmann et al. 2013a,b).
The region was chosen to include the area most strongly influenced by this heatwave and includes Italy, Greece, the Balkans and extends up to southern Poland and the Czech Republic.
Temperature anomalies from a 1961-1990 climatology were calculated in both the E-OBS series and the model simulations (relative to that period in the historical runs).
Nine CMIP5 models with all required model data available for the TXx variable were evaluated against the E-OBS dataset using a Kolmogorov-Smirnov test. Models where at least two-thirds of historical simulations were not statistically dissimilar (i.e. p>0.05) to the E-OBS series for the common 1950-2005 period were used in further analysis. Eight models passed this test (CanESM2, CNRM-CM5, CSIRO-Mk3.6.0, GFDL-CM3, IPSL-CM5A-LR, IPSL-CM5A-MR, MIROC-ESM, and NorESM1-M) with one failing (MRI-CGCM3).
The attribution analysis was conducted using three thresholds based on the E-OBS series: the 90th percentile (equivalent to a historical one-in-ten year event), the 95th percentile (equivalent to a one-in-20 year event), and a the existing record of 2007 (corresponding to a new record). These thresholds correspond to temperature anomalies of +2.85°C, +3.06°C, and +3.86°C respectively. As the event is ongoing it is not known which threshold is best to use at this time.
The likelihood of extreme heat exceeding each of these thresholds is compared for the natural world simulations (histNat; 1901-2005) and the current world (RCP8.5; 2006-2026). To estimate the change in likelihood, and associated 90% confidence interval, the pairs of complete CMIP5 simulations are 50% bootstrap resampled 10000 times, and the change in likelihood of extreme heat is recalculated. A conservative 10th percentile value is used to say that extreme heat events like this one are at least four times as likely due to human-induced climate change.
This process is repeated for the 1.5°C and 2°C global warming worlds (see King et al. 2017 for details).
Overall, a clear climate change signal is found in this event.
References
Haylock, M.R., N. Hofstra, A.M.G. Klein Tank, E.J. Klok, P.D. Jones and M. New, 2008: A European daily high-resolution gridded dataset of surface temperature and precipitation. J. Geophys. Res. Atmos., 113, D20119, doi:10.1029/2008JD10201
King, A. D., D. J. Karoly, and B. J. Henley, 2017: Australian climate extremes at 1.5 and 2 degrees of global warming. Nature Climate Change, doi:10.1038/nclimate3296.
King, A. D., G. J. van Oldenborgh, D. J. Karoly, S. C. Lewis, and H. M. Cullen, 2015: Attribution of the record high Central England Temperature of 2014 to anthropogenic influences. Environ. Res. Lett., 10, 054002, doi: 10.1088/1748-9326/10/5/054002.
Lewis, S. C., and D. J. Karoly, 2013: Anthropogenic contributions to Australia's record summer temperatures of 2013. Geophys. Res. Lett., 40, 3705–3709, doi:10.1002/grl.50673.
Sillmann, J., V. V. Kharin, X. Zhang, F. W. Zwiers, and D. Bronaugh, 2013: Climate extremes indices in the CMIP5 multimodel ensemble: Part 1. Model evaluation in the present climate. J. Geophys. Res. Atmos., 118, 1716–1733, doi:10.1002/jgrd.50203.
Sillmann, J., V. V. Kharin, F. W. Zwiers, X. Zhang, and D. Bronaugh, 2013: Climate extremes indices in the CMIP5 multimodel ensemble: Part 2. Future climate projections. J. Geophys. Res. Atmos., 118, 2473–2493, doi:10.1002/jgrd.50188.
King, A. D., D. J. Karoly, and B. J. Henley, 2017: Australian climate extremes at 1.5 and 2 degrees of global warming. Nature Climate Change, doi:10.1038/nclimate3296.
King, A. D., G. J. van Oldenborgh, D. J. Karoly, S. C. Lewis, and H. M. Cullen, 2015: Attribution of the record high Central England Temperature of 2014 to anthropogenic influences. Environ. Res. Lett., 10, 054002, doi: 10.1088/1748-9326/10/5/054002.
Lewis, S. C., and D. J. Karoly, 2013: Anthropogenic contributions to Australia's record summer temperatures of 2013. Geophys. Res. Lett., 40, 3705–3709, doi:10.1002/grl.50673.
Sillmann, J., V. V. Kharin, X. Zhang, F. W. Zwiers, and D. Bronaugh, 2013: Climate extremes indices in the CMIP5 multimodel ensemble: Part 1. Model evaluation in the present climate. J. Geophys. Res. Atmos., 118, 1716–1733, doi:10.1002/jgrd.50203.
Sillmann, J., V. V. Kharin, F. W. Zwiers, X. Zhang, and D. Bronaugh, 2013: Climate extremes indices in the CMIP5 multimodel ensemble: Part 2. Future climate projections. J. Geophys. Res. Atmos., 118, 2473–2493, doi:10.1002/jgrd.50188.
