AI Being Used to Confront, Mediate Climate Change

Researchers at UMass Amherst have introduced a framework for designing machine learning algorithms that aims to make it easier to specify fairness. (GETTY IMAGES)

AI is being applied to the biggest challenge facing the planet – climate change. Early results are encouraging.

Machine learning can be deployed in energy production, CO2 removal, education, solar geoengineering and finance, among 13 relevant answers according to a paper titled “Tackling Climate Change with Machine Learning, present at a workshop in June as a way to focus research, according to David Rolnick, a postdoctoral fellow at the University of Pennsylvania and one of the authors.

“It’s surprising how many problems in machine learning can meaningfully contribute to,” said Rolnick, quoted in an account in National Geographic. Possible outcomes include more energy-efficient buildings, new low-carbon materials, better monitoring of deforestation and greener transportation.

Three specific areas where AI research could focus were suggested: better climate predictions, showing the effects of extreme weather and measuring where the carbon is coming from.

Climate predictions can be enhanced by climate informatics, a discipline at the intersection of data science and climate science. It covers a range of topics including extreme events, reconstructing of past climate conditions, and large-scale models to be used for predictions. Climate modeling is progressing, with complex climate simulations having potential to unlock new insights.

One project is using machine learning algorithms to combine the predictions of some 30 climate models used by the Intergovernmental Panel on Climate Change.

Researchers at the Montreal Institute for Learning Algorithms (MILA), Microsoft and ConscientAI Labs are using General Adversarial Networks (GANs) to simulate what homes will look like after being damaged by rising sea levels and more intense storms. Plans include release of an app to show individuals what their neighborhoods and homes might look like with different climate change scenarios.

Banking Industry Also Studying Climate Change

A London-based not-for-profit consultancy called Caron Tracker is researching the impact of climate change on financial markets. It generates data by monitoring coal plant emissions with satellite imagery. Carbon Tracker is working to fulfill a UN goal of preventing new coal plants from being built by 2020. A grant from Google is expanding the effort to include emissions from natural gas plants, to help identify where pollution is coming from.

“This can be used worldwide in places that aren’t monitoring,” said Durand D’souza, a data scientist at Carbon Tracker. “And we don’t have to ask permission.”

Climate Change AI is an organization of volunteers from academia and industry discussing how computational science can mitigate climate change. Participants include Andrew Ng, co-founder of Google Brain, Deis Hassabis, a founder of DeepMind and Jennifer Chayes, managing director at Microsoft Research, according to an account in BBVA, serving the banking sector.

AI is seen as helping improve the energy sector, where automated distribution networks can perform real-time smart assessments to fine tune supply and demand of electricity. Smart homes and intelligent operations and logistics in the construction industry, have the potential to lower the carbon footprint. Algorithms and machine learning make it possible to anticipate electricity demand of a city or a manufacturing plan months in advance. Power can potentially be distributed to small local populations more efficiently as a result.

Google operates a fleet of wind farms in the US. Algorithms developed by Alphabet’s Deepmind researchers are able to predict wind farm energy 36 hours in advance, using advanced weather forecast technologies, now available.

In transportation and logistics, data from the Intergovernmental Panel on Climate Change (IPCC) shows that between 1970 and 2004, the sector increased its greenhouse emissions by 120 percent. The potential is there for transportation companies to more accurately predict demand and avoid risks. DHL, the global logistics services provider, has developed software that can juggle up to 58 parameters to define optimal schedules for cargo airplanes days in advance. This should result in few flights.

UK Government Backing New Doctoral Research on Climate Change

The UK government is backing more research into climate change at Cambridge University, with a new doctoral training program on the Application of AI to the Study of Environmental Risks, to be led by Prof. Simon Redfern, head of the Department of Earth Sciences.

The scientific community has access to larger datasets than ever before to help conduct this research. “These datasets represent a transformation in the way we can study and understand the Earth and environment, as we assess and find solutions to environmental risk,” said Redfern in an account in liwaiwai.com, a site aimed at programmers. “Such huge datasets pose their own challenges, however, and new methods need to be developed to tap their potential and to use this information to guide our path away from environmental catastrophe.”

Projects underway include the use of satellite observations to chart the distribution and pathways of whales through oceans, large datasets to understand biodiversity changes in woodland habitats, machine learning to understand earthquake risk,and the use of drones to monitor hazards at active volcanoes.