Klimamodelle unterschätzen Rußmengen und deren Erwärmungswirkung in der Arktis

Ruß ist ein klimatisch wärmendes Aersol. Vor ein paar Jahren erkannte man, dass Ruß viel stärker wärmt als angenommen. Daraufhin hätte man eigentlich einen Teil der Erwärmung der letzten 150 Jahre vom CO2 abziehen und auf den Ruß übertragen müssen. Das war politisch natürlich nicht zu machen, also trickste man die Modelle so hin, dass am Ende das CO2 seine vermeintliche Klimakraft behielt. Ein fauler Kompromiss, den wir bereits in unserem Buch “Die kalte Sonne” vor 5 Jahren monierten.

Was gibt es Neues aus der Forschung zum Ruß? Bereits 2013 erschien in PNAS ein Paper von Thomas Painter und Kollegen, die den überaschend frühen und abrupten Gletscherrückzug in den Alpen untersuchten. Bei ihren Eisuntersuchungen stießen sie auf eine starke Zunahme des Ruß aus dieser Zeit, der mit dem Beginn der Industrialisierung Mitte des 19. Jahrhunderts zu tun hatte. Die Autoren schlußfolgerten, dass das Abschmelzen der Alpengletscher stark vom Ruß-Anstieg mitgeprägt war. Hier der Abstract:

End of the Little Ice Age in the Alps forced by industrial black carbon
Glaciers in the European Alps began to retreat abruptly from their mid-19th century maximum, marking what appeared to be the end of the Little Ice Age. Alpine temperature and precipitation records suggest that glaciers should instead have continued to grow until circa 1910. Radiative forcing by increasing deposition of industrial black carbon to snow may represent the driver of the abrupt glacier retreats in the Alps that began in the mid-19th century. Ice cores indicate that black carbon concentrations increased abruptly in the mid-19th century and largely continued to increase into the 20th century, consistent with known increases in black carbon emissions from the industrialization of Western Europe. Inferred annual surface radiative forcings increased stepwise to 13–17 W⋅m−2 between 1850 and 1880, and to 9–22 W⋅m−2 in the early 1900s, with snowmelt season (April/May/June) forcings reaching greater than 35 W⋅m−2 by the early 1900s. These snowmelt season radiative forcings would have resulted in additional annual snow melting of as much as 0.9 m water equivalent across the melt season. Simulations of glacier mass balances with radiative forcing-equivalent changes in atmospheric temperatures result in conservative estimates of accumulating negative mass balances of magnitude −15 m water equivalent by 1900 and −30 m water equivalent by 1930, magnitudes and timing consistent with the observed retreat. These results suggest a possible physical explanation for the abrupt retreat of glaciers in the Alps in the mid-19th century that is consistent with existing temperature and precipitation records and reconstructions.

Dank Satellitenmessungen haben wir heute ein genaueres flächendeckenderes Bild der Gletscheralbedo. Allerdings gbt es auch hier einige “Fallen”, denen es auszuweichen gilt. So räumte das Dartmouth College im Oktober 2015 ein, dass vermeintliche Veränderungen der Albedo im nördlichen Grönland gar nicht real seien, sondern auf eine Verschlechterung der Satellitensensoren zuückging, sozusagen ein Messartefakt. Im März 2016 folgte dann eine genauere Auswertung eines Teams von Marco Tedesco und Kollegen in The Cryosphere. Sie fanden für den Zeitraum 1981-2012 einen leichten Rückgang der Albedo in Grönland, wobei für den Zeitraum 1981-1996 allerdings kein Trend auszumachen sei. Rußmessungen aus dem nördlichen Finnland geben Hoffnung. In den letzten 40 Jahren ist die Rußkonzentration in der Atmosphäre dank sinkender Emissionen stetig zurückgegangen.

Das japanische Forschungszentrum RIKEN bemängelte im Mai 2016, dass die Klimamodelle viel zu wenig Ruß in den arktischen Regionen annahmen. In der Realität gäbe es viel mehr Ruß:

Current atmospheric models underestimate the dirtiness of Arctic air

Black carbon aerosols—particles of carbon that rise into the atmosphere when biomass, agricultural waste, and fossil fuels are burned in an incomplete way—are important for understanding climate change, as they absorb sunlight, leading to higher atmospheric temperatures, and can also coat Arctic snow with a darker layer, reducing its reflectivity and leading to increased melting. Unfortunately, current simulation models, which combine global climate models with aerosol transport models, consistently underestimate the amount of these aerosols in the Arctic compared to actual measurements during the spring and winter seasons, making it difficult to accurately assess the impact of these substances on the climate.

To find out if these inaccuracies could be mitigated, a team of scientists decided to use the Japanese K computer to perform fine-grained simulations of how black carbon aerosols are transported to and distributed in the Arctic region. By using smaller grids—with spacing of just a few kilometers rather than several tens of kilometers as in conventional current models—they were able to show that they could more realistically model the amount of black carbon aerosols, mitigating the underestimation in more coarse-grained models. Their finest model used 3.5 kilometer grids broken up vertically into 38 layers, so that it required 1.6 billion grids to cover the globe. The simulation, done on the 10-petaflop K computer, still required 17 hours to perform the two week simulation.

According to Yousuke Sato of the RIKEN Advanced Institute for Computational Science (AICS), “this research shows that powerful supercomputers, by performing more fine-grained simulations, can help us to model weather and climate patterns in a more realistic way. We have to note, however, that while our model reduced the underestimation, it did not completely eliminate it. Further generations of even more powerful computers will allow us to run simulations that may be able to make even more realistic simulations and help us to understand the mechanism through which these aerosols are transported.”

“It is also known,” continues Sato, “that current models do not realistically model the vertical distribution of the aerosols, and we believe that finer measurements could help there as well. Unfortunately there were no vertical measurements taken in November 2011, the time we chose to model, so we plan in the future to do simulations for time periods for which actual measurement data exist.”

The research, published in Scientific Reports, was carried out by AICS in collaboration with the University of Tokyo, the National Institute of Environmental Studies, Kyushu University, and the Japan Aerospace Exploration Agency.

Reference: Yousuke Sato, Hiroaki Miura, Hisashi Yashiro, Daisuke Goto, Toshihiko Takemura, Hirofumi Tomita, and Teruyuki Nakajima, “Unrealistically pristine air in the Arctic produced by current global scale models”, Scientific Reports, doi: 10.1038/srep26561

Die Konsequenz dieser Modellfehlannahme ist klar: Weniger Ruß in den Modellen bedeutet auch weniger Erwärmung durch Ruß. So kann die beobachtete Arktis-Erwärmung bequem vor allem dem CO2 angelastet werden. Wenn man den Ruß jetzt in den Modellen erhöht, muss dem CO2 ein Teil der Erwärmungswirkung wieder abgenommen werden, die CO2-Klimasensitivität sinkt entsprechend. Wo man auch hinschaut, stets nimmt sich das CO2 mehr vom Klimakuchen als ihm eigentlich zusteht. Ein Fehler mit System.