Lukas Brunner

@lukasbrunner.bsky.social

774 Followers 337 Following

Climate scientist @uni-hamburg.de: future projections, extremes, uncertainties | Formerly @ethz.ch @univie.ac.at | PhD @uni-graz.at | Python nerd | Hiking enthusiast #SystemChangeNotClimateChange #OpenScience More: https://lukasbrunner.github.io

Screenshot der Startseite von "Der Standard": Ein Foto zeigt eine ausgetrocknete, vertrocknete Wiese mit Bäumen im Hintergrund. Die Schlagzeile lautet "Neuer Hitzerekord: Erstmals 40,1 Grad in Wien", mit Hinweis auf extreme Hitze auch an den Folgetagen und der höchsten Wetterwarnstufe im Osten Österreichs. Darunter folgen Kacheln mit weiteren Nachrichten.
Screenshot of the title page of the article entitled: Three decades of simulating global temperature patterns with coupled global climate models
Fraction of models closest to each reference. Shown is the fraction of models for which each reference dataset (shown on the x-axis) leads to the lowest RMSD.
Root-mean-square distance (RMSD) to the full reference range for all 176 models individually.
Model performance as a function of model resolution. RMSD to the full reference range as a function of model resolution for all 165 CMIP and 11 km-scale models.
Difference between maximum and minimum RMSD divided by the mean RMSD. The RMSD is calculated for each model using each of the 10 different reference datasets in turn. Difference in RMSD when using one of two references: ERA-Interim or its successor ERA5.
Fraction of Earth’s land surface area for which the models' 20-year mean surface temperature falls within the range of the 10 reference. Two models are highlighted: IFS-FESOM from nextGEMS cycle 4 and from EERIE phase 1.
Fraction of Earth’s land surface area for which the models' 20-year mean surface temperature falls within the range of the 10 references and area-weighted root-mean-square distance (RMSD) relative to the reference range.
Book cover: Welt ohne Morgen by Tom Roth
https://meetingorganizer.copernicus.org/EGU25/session/52390
Figure from: O’Reilly C., Brunner L., Qasmi S., Nogherotto R., Ballinger A., Booth B., Befort D., Knutti R., Schurer A., Ribes A., Weisheimer A., Coppola E., and McSweeney C. (2024): Assessing observational constraints on future European climate in an out-of-sample framework, npj Climate and Atmospheric Science
Abbildung, die den Jahresgang der Maximaltemperatur der 30 Jahre von 1961 bis 1990 zeigt. Der Grenzwert für Extreme, basierend auf der 90. Perzentile ist ebenfalls zu sehen. Tage an denen die Maximaltemperatur den Grenzwert überschrietet sind als extrem gekennzeichnet. Im unteren Teil der Abbildung sieht man die Frequenz von Extremen für jedes Monat. In einigen Monaten weicht die Frequenz deutlich von den 10% erwarteten Extremen ab. Im Juni um -75%.
Abbildung einer Weltkarte mit der Ausprägung des Fehlers im Jahresmittel. Einige Regionen zeigen so gut wie keinen Fehler, in anderen Regionen übersteigt der Fehler 30%.
Screenshot of paper text, reading: "To conclude, we strongly warn against the use of long-running windows without correction when calculating extreme thresholds. The use of such a biased method is never advisable, even though the impacts on derived metrics might not always be strong or immediately apparent. Many studies, naturally, adopt methods from previous work, so avoiding the use of biased methods also helps to prevent their spread. Crucially, the same biased method may have a negligible effect in one setting and a large impact in another."
Figure showing a global map of extreme temperature bias after applying the bias correction mentioned in the tweet. The bias is limited to less than 5% for this case.
Figure showing a global map of summer land heatwave changes between 1961-1990 and 2071-2100.
Figure showing a global map of temperature extreme changes between 1961-1990 and 2071-2100 as the ratio between future and past.
Figure showing a global map of extreme temperature bias.
Figure showcasing how the interaction between a 31-day running window and the seasonal gradients leads to a 90th percentile extreme threshold that is not exceeded by a single value.
Figure showing the seasonal cycle of maximum temperature in the North Atlantic as well as the instances when temperature exceeds the 90th percentile threshold. In the bottom part, the mean frequency per month is shown. In most months the frequency deviates clearly from the expected 10%.
Spaker Katharine Hayhoe in front of one of her slides. The text on the slide reads: "People want solutions that give us a btter life and a btter world, not a worse one.