IBM researchers are developing a computer system that can predict Beijing’s air pollution 72 hours in advance. China is committed to improving urban air quality by 10 per cent by 2017.
Pollution levels in urban centres are difficult to predict because they depend on various factors including industrial activity, traffic and weather conditions.
The system being developed by IBM uses adaptive machine learning, or something the company calls ‘cognitive computing’: it extracts insights from huge amounts of data from several different models and then combines them to predict the severity of air pollution in different parts of the city up to 72 hours in advance, reports MIT Technology Review.
“Our researchers are currently expanding the capability of the system to provide medium- and long-term (up to 10 days ahead) as well as pollutant source tracking, ‘what-if’ scenario analysis, and decision support on emission reduction actions,” says Xiaowei Shen, director of IBM Research China.
For instance, the system could offer specific recommendations to reduce pollution, such as closing certain factories or temporarily restricting road traffic.
According to Shen, the predictions are high resolution and 30 per cent more accurate than those derived through conventional approaches. A similar system is being developed for a city in the Hebei province, and IBM hopes to export this technology to other cities and countries where pollution is a growing problem.
The average level of airborne particulates measured in 360 Chinese cities in April this year was more than two and a half times the limit recommended by the World Health Organization, explains the article.