TWH – A new study published Monday in Proceedings of the National Academy of Sciences uses machine learning primed on climate models to predict the time remaining until the United Nations Paris Agreement targets will be reached. The findings suggest that the world is already experiencing the impacts of 1.1°C (1.98°F) to 1.2°C (2.16°F) of warming. It also finds that passing 1.5°C or 2°C above preindustrial levels could dramatically increase climate event risks to society and ecosystems.
Image credit: Reto Scheiwiller from Pixabay
Machine learning in this study refers to artificial neural networks that are trained on climate model data output. Machine learning is a broad term for a field of inquiry that uses data to improve performance on tasks. It tries to mimic intelligent human behavior and “learned” without being programmed to perform a task. AI pioneer Arthur Samuel defined it as “the field of study that gives computers the ability to learn without explicitly being programmed.”
The machine learning algorithm takes data and makes predictions training itself on its own successes and failures in prediction against the data. When it guesses and gets a prediction wrong, it tweaks itself to improve the accuracy of its results. Over time and millions of bits of trial and error, the algorithm refines its predictive capacity.
The authors of the new research note that “given their policy relevance, and the strong scientific evidence for accelerating impacts, the time remaining until these global warming thresholds are reached has generated considerable interest in the scientific literature.”
In the current study, the researchers from Stanford University and Colorado State University note that their approach with this method of machine learning allows them “to make truly out-of-sample predictions of that timing, based on the spatial pattern of historical temperature observations. Our results confirm that global warming is already on the verge of crossing the 1.5 °C threshold, even if the climate-forcing pathway is substantially reduced in the near term. Our predictions also suggest that even with substantial greenhouse gas mitigation, there is still a possibility of failing to hold global warming below the 2 °C threshold.”
Using data from 1980 to 2021, their algorithm was honed on predicting rising temperatures. They also offered historical measurements to continue priming the model to predict the temperature benchmarks.
The researchers found that we have about a decade left until the 1.5-degree target is reached and then exceeded. These findings are consistent with prior research on when the particular target from the Paris Accords would be reached and exceeded.
But this study also found that even in the best possible scenario with the lowest fossil fuel emissions from steep cuts in their use over the next few decades, the chances of breaching the 2°C target remain high. The study finds that hitting the 2°C target will occur between 2043 to 2058, with a central likelihood in 2050. The model found a nearly 70% chance that the two-degree Celsius threshold would be crossed between 2044 and 2065.
The findings may be a warning, but they are not a reality.
The iterative process of machine learning relies not only on the quality of the data but the quality and training of the algorithm. They represent an estimate but not a perfect substitute for climate models. Still, the findings suggest that there is a very evident likelihood that climate warmth targets will be exceeded leading to new climate challenges in the near future.
The researchers note the limitations of AI in climate modeling. Those limitations introduce different biases in the conclusions of the algorithm.
Nevertheless, the researchers note, “The fact that our central estimate for the time until 1.5 °C lies between 2033 and 2035 in the High, Intermediate, and Low forcing scenarios confirms that global warming is already on the verge of crossing the 1.5 °C threshold, even if the climate forcing pathway is substantially reduced in the near term.”
“We have very clear evidence of the impact on different ecosystems from the 1C of global warming that’s already happened,” said Stanford University climate scientist Noah Diffenbaugh, who co-authored the study with atmospheric scientist Elizabeth Barnes. “This new study, using a new method, adds to the evidence that we certainly will face continuing changes in climate that intensify the impacts we are already feeling.”
They added that their prediction is consistent with other models that even with “substantial greenhouse gas mitigation,” the possibility of exceeding Paris targets by mid-Century are high.
The World Economic Forum describes the consequences of the target temperatures in the model. They include a 70% – 90% decline in coral reefs, ice-free arctic summers, and a 6% and 8% loss of insect and plant species respectively.
Another serious consequence is the effect on human health. The WEF notes that in the worse of the two scenarios, that is if the 2°C target is exceeded, as much as 37% of the world’s human population will experience severe heat 1 in every 5 years.
“Our AI model is quite convinced that there has already been enough warming that 2C is likely to be exceeded if reaching net-zero emissions takes another half-century,” said Diffenbaugh. “Net-zero pledges are often framed around achieving the Paris Agreement 1.5C goal,” he added. “Our results suggest that those ambitious pledges might be needed to avoid 2C.”
Still, Diffenbaugh hopes that this work will help motivate decision-makers rather than breed dismay. “Stabilizing the climate system will require reaching net zero, he said. “There are a lot of emissions globally – and it’s a big ship to turn around.”
The Wild Hunt is not responsible for links to external content.
To join a conversation on this post:
Visit our The Wild Hunt subreddit! Point your favorite browser to https://www.reddit.com/r/The_Wild_Hunt_News/, then click “JOIN”. Make sure to click the bell, too, to be notified of new articles posted to our subreddit.