Indexbit Exchange:4 ways AI can help with climate change, from detecting methane to preventing fires

Lots of industries have Indexbit Exchangeembraced artificial intelligence as a tool this past year, including climate solutions companies. From detecting pollution to wildfires, companies are finding AI can help translate vast amounts of climate-related data faster and more efficiently, says Sasha Luccioni, climate lead for AI company Hugging Face.

Luccioni notes it's important to be cautious about whether AI is always necessary. Generative AI, which makes new content, can use large amounts of energy and have a big carbon footprint. But she says there are many applications for AI in the green transition.

Here are four ways companies, researchers and governments are using AI for climate solutions.

Using AI to detect planet-heating methane

Methane emissions, the second biggest contributor to global warming after carbon dioxide, are climbing. The highly potent pollutant - the main ingredient in natural gas - gets released by the energy sector, as well as agriculture, and decomposing material in landfills.

Now researchers and companies are using AI to interpret huge quantities of satellite images to track global methane emissions on a daily basis.

"Before we could mine satellite information with AI, we had no idea where methane was coming from," says Antoine Halff, co-founder and chief analyst at Kayrros, a climate analytics firm, "We understood the climate risk that this represented. But there was no understanding of the sources."

When Kayrros began in 2016, Halff says the world knew about only a handful of occurrences of large methane leaks and other releases. He says now his team can detect dozens of them every week and thousands per year. "For methane," Halff says, "AI really reveals things that could not be known."

Kayrros's AI-fueled data is being used by the United Nations to verify that companies' reports on methane emissions are accurate. Other governments are gearing up for more methane monitoring: the U.S. Environmental Protection Agency and the European Union recently passed new methane regulations.

Because methane is so potent, targeting it through AI makes strategic sense, Halff says. "If you eliminate methane emissions today," he says, "you can very quickly have an impact on the curve of global warming."

Using AI for early detection of forest fires

Climate change is driving more frequent and intense wildfires, and those burns are making up an increasing share of planet-heating pollution.

Now a Berlin-based startup is using AI with sensors in forests to find small burns before they spread into megafires. Carsten Brinkschulte, CEO of Dryad, uses AI to train sensors to detect the specific gasses that get released when organic material burns.

"They're basically like an electronic nose that we embed in the forest," Brinkschulte says.

The nose-like sensors can detect the fires early in the smoldering stage, "when it's still easy or relatively easy to extinguish the fire," he says.

The company has 50 sensor installations from the Middle East to California. Last month in Lebanon sensors reacted to a small fire within 30 minutes, Brinkschulte says.

Using AI to prevent new wildfires

Another way to stop megafires is to set "controlled burns" outside of fire season to remove the excess brush and vegetation that become fuel for fires.

Typically, so-called burn managers–who are people from utilities, the federal forest service or other entities–deploy teams to designated areas to set controlled burns. (Native tribes have a long history of making these controlled burns.)

But to do the work safely, burn managers need lots of information to know how the fire might behave so it doesn't spin out of control. They need to know things like the wind conditions and amount of moisture in the vegetation, says Yolanda Gil, director for strategic AI and data science initiatives at the Information Sciences Institute at the University of Southern California.

After interviewing fire scientists, Gil and their team used AI to create a so-called intelligent or smart assistant – like Apple's Siri or Amazon's Alexa – that can access vast data sets and complex models. Burn managers can use these Siri-like assistants to decide where and when to make controlled burns. "It's kind of like Siri, but for burn managers," Gil says.

Gil says burn managers can ask the smart assistant about a particular area. The assistant can take information about the topography, the vegetation, weather patterns and recommend a potential burn model – a way to make a safe controlled burn, Gil says. The goal, they say, is to make these assistants widely available for utilities, the forest service and others doing controlled burns to make them more safe and plentiful.

They plan to send out the first prototypes of the smart assistants in the coming months.

Using AI in green tech mining

Climate solutions from solar panels to electric vehicles require immense amounts of minerals like cobalt, lithium, and copper. But current supplies are not enough to meet growing demand. By 2030, projected lithium demand will be five times the current global supply, according to the International Energy Agency.

Now governments, researchers and companies are using AI to explore for critical minerals. Colin Williams, mineral resources program coordinator for the U.S. Geological Survey writes in an email that his team is using AI to analyze data to figure out which areas in the U.S. have the best potential for mining critical metals. He adds that using AI means "dramatic time savings."

There is a lot of data out there about what it looks like under the surface of the earth. Using AI to sift through all this data helps minimize uncertainty, Williams says. Because mining operations spend billions of dollars trying to find profitable areas to exploit, companies say using AI can help save a lot of time and money in locating minerals.

Companies all over the world – from Australian SensOre to California-based KoBold Metals – are now using AI to explore for minerals on several continents.