Skip to Content

Can Machine Learning Help Lift China’s Smog?

As China’s air quality sinks, IBM’s research group is using AI techniques to predict pollution levels for every square kilometer 10 days in advance.
March 28, 2016

From the street, through Beijing’s heavy smog, it can sometimes be hard to make out IBM’s Chinese headquarters: a towering office building with a distinctive undulating architectural flourish and a large company logo at the top.

But just a short distance away, on the northeast outskirts of the capital, IBM computer scientists are using artificial intelligence to develop what they think will be a way to manage China’s notorious and chronic pollution problem more successfully.

The team is using complex computer models and machine learning to calculate how pollution will spread across the city. The researchers can now produce pollution forecasts, with a resolution of a kilometer square, up to 10 days in advance.

These predictions can also tell the government how it might act to avoid the worst scenarios—for instance, by shutting certain factories, or by reducing the number of cars on the road.

When MIT Technology Review visited the offices of IBM Research–China last November, the air was particularly bad. Cold weather had increased demand for electricity, forcing nearby coal plants to boost output. This, combined with the usual traffic chaos, had produced some truly lung-scorching smog. Pollution is measured in terms of the amount of fine particulate matter per cubic meter. For a developed city, the World Health Organization recommends that this figure not exceed 25. During my visit, it reached almost 250. The modeling system, called Green Horizon, was being used to predict the spread of pollution; but it was not clear whether the government had decided to limit either factory output or the number of cars on the road. The need for heating seemed to be outweighing the ill effects.

A day of heavy pollution in Beijing, December 25, 2015.

The Beijing project, which uses data captured from pollution sensors around the city, involves complex modeling both of specific sources of pollution and of weather and air movement to predict how bad pollution will be in different neighborhoods. Previous readings are used to refine predictions using an approach known as machine learning. This makes it possible to create new predictions from these combined factors, says Xiaowei Shen, director of IBM Research–China.

“Everyone’s talking about big data, but we all know the traditional IT technologies we have developed will not be sufficient to handle all the big data,” Xiaowei says.

IBM runs complex simulations of the economic impact of shutting down factories due to pollution levels, says Jin Dong, a distinguished engineer at IBM Research–China and leader of the project. Various government bodies make those decisions.

The Chinese government may need to make some hard decisions concerning energy production in order to mitigate both the short-term health consequences and the long-term climate effects of air pollution. Sarah Williams, an assistant professor in MIT’s Department of Urban Studies and Planning and director of the Civic Data Design Lab, who studied Beijing’s pollution problem during the 2008 Olympics, says that IBM’s effort could be very valuable if it helps show the Chinese government how limited an impact on particulate matter short-term fixes like taking factories offline can have—and how necessary more extensive environmental regulation may be.

“Unless the government uses that data and data visualization to enact changes, it will have little net benefit,” Williams says.

IBM’s modeling system is being used in two other Chinese cities with major pollution problems: Baoding and Zhangjiakou. Meanwhile, related technology created at IBM is being used to study the relationship between traffic and pollution in Delhi, India, and the effectiveness of air pollution control measures in Johannesburg, South Africa.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.