Scientists have long known that the world is running out of time to meet international climate goals. Artificial intelligence is now coming to a similar conclusion.
Groundbreaking new AI research finds it will take humans about a decade to beat the optimistic goal of limiting global warming to 1.5 degrees Celsius.
This is the same conclusion that scientists have arrived at using traditional climate modeling methods, but AI research is a testament to climate scientists and policy experts that the world will almost certainly exceed the 1.5°C target. Adds more evidence to the growing conviction betweenclimate wire11 November 2022).
Policy makers continue to strive to keep the global alert below 2 degrees even if the 1.5 degree target is exceeded. But even that goal is in jeopardy, according to AI research. We find that the 2C threshold may be approached even faster than previous studies predicted.
AI research suggests that the 2C threshold could be reached around the middle of this century, even with relatively tight cuts in greenhouse gas emissions over the next few decades. This is decades earlier than conventional climate models generally suggest under the same hypothetical low-emissions scenario. The UN’s Intergovernmental Panel on Climate Change admits that the world could cross the 2°C threshold by the end of the century in that scenario, but it is ‘unlikely’. also said.
This does not mean that there is no hope of meeting Paris’ climate targets.
The aggressive emission reduction scenarios used in this study are not necessarily the best the world can do. Nonetheless, it assumes that the world will reach net zero emissions sometime after the middle of this century. Meanwhile, dozens of countries around the world have set net-zero timelines, with many aiming for 2050. This is a little earlier than the scenario envisioned in the new study.
The IPCC reports that achieving the 1.5°C target would require the world to achieve net-zero emissions by 2050, while the 2°C target would require net-zero emissions by around 2070. . But AI research suggests that net zero by 2050 may even require a less ambitious threshold of 2C.
“AI predicts that they [pledges] We may need it to avoid 2 degrees,” said Noah Deffenbaugh, a climate scientist at Stanford University who co-authored the new study with Elizabeth Barnes, a climate scientist at Colorado State University.
Conventional climate research usually makes climate predictions using computer models that simulate the physical processes responsible for global warming. A new study uses a unique approach to address the prevailing climate problem of our time.
Researchers investigated using artificial neural networks, a form of machine learning. Neural networks provide a way for computers to process large amounts of data and recognize patterns in the information presented to them. It can then be trained to make predictions based on learned patterns.
Researchers first trained a neural network using inputs from traditional climate model simulations. Next, enter the global map of actual current temperature anomalies. These are places around the world where temperatures are above or below average. Next, they asked the neural network to predict how quickly the 1.5 C and 2 C targets would be reached under various hypothetical future emissions scenarios.
A neural network predicted that the 1.5 C target would be reached between 2033 and 2035. He also found that the 2C goal is likely to be reached between 2050 and 2054.
AI does not entirely rule out the possibility that the world could avoid the 2C threshold under the studied low-emissions scenario. However, we consider that result unlikely.
“Given how much warming has already happened on the map of global temperature anomalies in recent years, the AI is fairly confident that 2°C is a real possibility in low-forcing scenarios,” says Defen. Bo said, “If it takes another half-century to reach net zero, the AI predicts that we will likely reach 2 degrees Celsius.”
According to Amy McGovern, a scientist at the University of Oklahoma and director of the National Science Foundation’s AI Laboratory on Trusted AI in Weather, Climate and Coastal Oceanography, the research is “definitely new and innovative.”
McGovern was not involved in the new research, but he is familiar with it. Burns, Defemboau’s co-author of the new study, works for her at the NSF AI lab.
According to McGovern, AI is rapidly gaining traction as a new tool in weather and climate science. It can be used to complement traditional modeling techniques in many ways, from short-term weather forecasting to modeling cloud formation and other complex climate-related phenomena.
Climate models are generally very accurate. However, they require enormous computational power and cannot always adequately represent all the granular processes that make up the world’s climate system, especially on a global scale.
AI can replace certain subtle physical processes in climate models, so they can run faster. It also helps you process large amounts of data more easily.
“There’s really been a revolution in the amount of data available today,” says McGovern. “But right now there is more data than humans can really handle. AI can help us understand what we can focus on.”
AI does not necessarily replace traditional climate and weather modeling techniques. However, it can help strengthen the model and improve its limitations, opening up new possibilities for climate research.
“We really think we are on the cusp of a revolution in how AI is being used to predict weather and climate,” McGovern said. “It’s going to really change the way we improve our forecasts.”
Reprinted from E&E News with permission of POLITICO, LLC. Copyright 2023. E&E News provides essential news for energy and environmental professionals.
