Credit: Unsplash/CC0 public domain
Artificial intelligence is being developed by researchers to assess the tipping point for climate change. Deep learning algorithms could be used to warn of climate change and prevent it from getting worse.
Chris Bauch, an assistant professor of applied mathematics at University of Waterloo is the co-author of a paper that reports on results of the deep-learning algorithm. Bauch stated that the research examines thresholds at which a system experiences irreversible or rapid change. Bauch stated that the new algorithm was capable of predicting tipping points better than other approaches and also providing information about the state beyond the tipping point. "Many of these tipping point are undesirable and we would like to stop them if possible."
Melting Arctic permafrost could release large amounts of methane, spurring further rapid heating, and could also cause the breakdown of oceanic currents systems which could result in almost immediate changes to weather patterns. Ice sheet disintegration could also lead to rapid sea level change.
Researchers claim that the AI was designed to learn about all tipping points.
This approach combines AI and mathematical theories about tipping points. It is more effective than either of them could alone. The researchers trained the AI using a "universe" of tipping points, which included around 500,000 models. They then tested the AI against specific tipping points in real-world systems, including historical climate cores.
Timothy Lenton, co-author of the study and director of the Global Systems Institute at University of Exeter, said that "our improved method could raise red flags whenever we're near a dangerous tipping point." "Providing better early warning of climate tipping point could help societies adapt to and reduce their vulnerability, even if it is not possible,
Deep learning has made great strides in pattern recognition, classification, and classification. Researchers have converted tipping point detection into a pattern recognition problem for the first-time. This is to detect patterns that are occurring before tipping points and create a machine-learning algorithm that can tell if a tipping tip is approaching.
Thomas Bury, a McGill University postdoctoral researcher and one of the co-authors, said that tipping points are well-known in climate systems. However, tipping points can also be found in ecology, epidemiology, and stock markets. "What we have learned is that AI can detect tipping points in complex systems and is very adept at finding them.
Madhur Anand (another researcher on the project, and director of Guelph Institute for Environmental Research) said that the new deep learning algorithm is "a game-changer for being able to anticipate big shifts including those related with climate change."
The team has now learned how tipping point functions and is currently working on the next stage. This is to provide the data necessary for current trends in climate change. Anand warned that such knowledge could lead to disaster.
She said, "It certainly gives us a head up." It's up the humanity to decide what they do with this information. These new findings should lead to fair and positive change, I hope.
Bauch, Lenton and Bury, Anand published the paper "Deep Learning for Early Warning Signals of Tiping Points" in the journal Proceedings of the National Academy of Sciences.
Continue reading. Breaching tipping points could increase the economic cost of climate change impacts
Further information: Thomas M. Bury and colleagues, Deep learning to detect tipping points early, Proceedings of the National Academy of Sciences (2021). Information for Journal: Proceedings of the National Academy of Sciences Thomas M. Bury and colleagues, Deep learning to early warning signs of tipping points. (2021). DOI: 10.1073/pnas.2106140118
