2013 was a record year in Australia: The average temperature reached an all-time high. To even be able to depict it, meteorologists had to introduce a new color. Since then, magenta has come to visually represent temperatures over 50°C. A new research project uses both digital and art history methods to look into the history and variety of these and many other climate images on the internet. Who creates climate images on the web? Who uses them, and for what purpose? Which images gain currency and how? These questions are explored by researchers at the University of Potsdam, the Potsdam University of Applied Sciences (FH), and the Potsdam Institute for Climate Impact Research (PIK).
A polar bear stranded on a melting ice floe, flooded cities, dried up riverbeds, or overturned wind turbines – pictures are used to tell us something about the climate. To this end, they use a certain visual language. A Potsdam research team is analyzing the rhetoric of climate images and the strategies of various stakeholders. “We use image recognition software that works like a search engine,” explains Sebastian Meier of the Interaction Design Lab at FH Potsdam. “It automatically crawls the internet and collects information on images and their contexts.” How many illustrations does an environmental organization like Greenpeace, for instance, use on its website? Are climate charts and diagrams or photos and comics being used? What colors, shapes, and contents are depicted? What year are the images from? “We are looking for commonalities among climate images on the internet,” summarizes media ecologist Prof. Dr. Birgit Schneider of the University of Potsdam.
The color ranges chosen is also a part of visual language. Global warming depicted in a bright red may look very threatening, whereas cool blue tones might convey something less dramatic. “Behind every visual statement, there is a desire to convince,” Schneider explains. When visualizations distort scientific findings and manipulate the public, however, that becomes a problem, since a lay audience might not notice it. “Scientific images of climate change are often very complex,” says Dr. Thomas Nocke of the PIK. The Intergovernmental Panel on Climate Change, national environmental agencies, and research institutions are among those creating such complex visualizations. But if politicians, journalists, or bloggers modify and simplify them, scientific correctness may fall by the wayside. And climate change deniers use other means than climate change activists. “A typical strategy of deniers is to cherry-pick,” the researchers say. They tend to lean on excerpts to cast doubt on global warming. “Sometimes certain datasets are simply omitted, or certain regions or periods are selected to support an argument,” Nocke says. Obsolete scientific theses are often used to depict climate change as a lie.
Climate images are used to negotiate trust in science
Uncertainty is a challenge for climate images. It poses a risk, particularly to the credibility of the presented information, since leeway in forecasts can be difficult for lay audiences to understand. “Probability plays into the hand of deniers,” Nocke explains. “The ethos of research, however, is supposed to be objective,” Schneider adds. “Yet researchers admitting what they are not sure about comes across as a sign of weakness to climate change skeptics.” For Schneider, climate images, therefore, reflect a society’s trust in science.
The researcher has dealt with climate images for a long time. For about eight years, she has been compiling relevant representations in science, art, entertainment, and advertising in a database, where they – 3000 images in total – are searchable by keyword: futures, curves, anomalies, apocalypse, world on fire, or skeptics. “As an art historian and media ecologist, I asked myself whether processing the images digitally would make sense.” A major advantage would be that with quantitative methods researchers would be able to quickly analyze millions of pictures, whereas the qualitative analysis of just a few pictures by art historians takes considerably longer.
Furthermore, the researchers are explicitly interested in a critical review of digital methods. “We want to test the suitability of algorithms that computer scientists use every day using methods from humanities,” Meier explains. For as objective as digital methods may seem, their results cannot be trusted without examination. Faulty, insufficient data or wrong search parameters can distort results. A poorly chosen training dataset – with which software “learns” – may bias the results. This became obvious with the image recognition algorithms of Google and flickr: People with dark skin were not detected properly, since the training dataset was calibrated mainly on people with fair skin. So how can digital methods be more than just supporting tools? This is another question the team will be addressing in the coming years.
Researchers analyze communication on climate change
“Even now, computer science is everywhere in our lives,” Thomas Nocke finds. Algorithms are making major decisions: They are indispensable in training driverless cars, investigative authorities use image analysis methods for crime detection, and dating sites match couples using algorithms – with far-reaching ethical consequences. “We believe that computer science can benefit from including humanities perspectives.” Climate images are, thus, well suited for testing the common application of humanities and digital methods as such. And in times of digital humanities and big data, critical views are all the more important.
Climate change is a global challenge – that is why the Potsdam researchers are focusing on how it is communicated. Climate images need to convey scientific findings in a way everyone can understand and as neutrally as possible. So there will be not only publications, workshops, and meetings at the end of the project, but also interactive online exploration tools. “We are not talking about an Excel sheet listing all characteristics of climate images,” says Meier. The tools will be descriptive and reveal new correlations. Last but not least, the researchers hope their work improves communication on climate change. “If all goes well, we will be able to provide political decisionmakers, the scientific community, and the arts scene with new ideas on how to better understand the global challenge of climate change,” Schneider says.
The Researchers
Prof. Dr. Birgit Schneider studied art theory and media theory. Since 2015, she has been Professor of Media Ecology at the University of Potsdam.
Universität Potsdam
Institut für Künste und Medien
Am Neuen Palais 10
14469 Potsdam
birgit.schneideruuni-potsdampde
Dr. Thomas Nocke studied computer science and computer graphics. He works at the Potsdam Institute of Climate Impact Research.
Potsdam-Institut für Klimafolgenforschung
Telegrafenberg A56
14473 Potsdam
nockeupik-potsdampde
Sebastian Meier studied communication design and interface design. Since 2012, he has been a member of the Interaction Design Lab.
Fachhochschule Potsdam
Kiepenheuerallee 5
14469 Potsdam
meierufh-potsdampde
Between April 2017 and March 2020, Volkswagen Foundation will be funding the collaborative project “New potentials for analyzing network images – Similarity as a criterion for comparing images in image studies, computer and visualization science using the example of climate images on the web”. Two doctoral theses will be written on the subject, and a colloquium, workshops, publications, and a website are planned.
http://www.uni-potsdam.de/medienoekologie/index/digitale-analyse-vernetzterklima-bilder.html
Text: Jana Scholz
Translation: Monika Wilke
Published online by: Alina Grünky
Contact for the online editorial office: onlineredaktion@uni-potsdam.de
Read this and other articles on research at the University of Potsdam in our research magazine Portal Wissen.