Atmospheric pollution is a given in South Africa’s industrial heartland, the Vaal Triangle. An innovative project that leverages Big Data and the Internet of Things will help understand the problem – and maybe clean things up for all of us

Urban air pollution generated by vehicles, industries and energy production kills approximately 800 000 people a year, according to a World Health Organisation report on environment and health in developing countries. It’s fair to infer from this that an uncomfortable proportion of those occurs in Johannesburg, the home to 4,5 million souls – and Africa’s most polluted air. But moves are under way to address that. A pilot project driven by a partnership between Johannesburg, IBM and the CSIR aims to combine Big Data, the Internet of Things and learning technologies to help the city deliver on its air quality management plan.

This forms part of the global rollout of IBM’s Green Horizons initiative. That kicked off a year ago in China, whose meteoric, relentless growth in the past two decades has come at a severe cost to air quality. The country’s massive pollution problem hit headlines in early December 2015 when it declared its first-ever Red Alert as a result of severe smog in Beijing. The city ground to a standstill, with restrictions on transport use, school closures and clampdowns on industry. After a mild respite, the alert had to be reimposed later that month.

China hopes to strike a better balance between economic growth and emissions strategy by means of scenario modelling. Current measures include temporary restrictions on urban traffic and construction activity and long- term measures applicable to industry. On his recent state visit to South Africa, Chinese president Xi Jinping pledged $60 billion of investment in environmental innovation on the continent. Unsurprisingly the researchers on the Johannesburg project will be working very closely with the Chinese. They, too, will be exploring the leveraging IoT technology and the number-crunching power of cognitive computing.

Their hoped-for outcomes: insights and recommended actions to improve air quality and health prospects for Johannesburg’s citizens – and, in the longer term, the rest of the country. Elsewhere in the world, similar ideas are being pursued. In the Indian capital of Delhi, for example, technologies are being put to use to understand the connection between the city’s traffic patterns and air pollution. The tie-up between the global IT company and Johannesburg is the second in less than a year. (Last October, IBM’s Watson social media analytics capacity tapped into citizens’ thoughts on the EcoMobility World Festival 2015.) Research scientist Tapiwa Chiwewe leads the air quality pilot project. He describes the tools that will be used as game-changing technologies that can help us transform our understanding of air pollution.

To start with, the Johannesburg project will examine both historical and real-time data with the aim of understanding what air pollution is, what causes it, and how we can deal with it. It’s worth noting the impact of history, in addition to current-day activities, on Johannesburg’s pollution scenario. Decades of mining and the resultant mine dumps helped generate minute airborne soot and dust particles – the size and kind most harmful to human health. That’s aggravated by increasing urbanisation and its traffic and industries, as well as the burning of wood and coal for domestic use. Next comes the really interesting bit: high-accuracy air pollution forecasting.

How will atmospheric forecasting help? “That will be one of the biggest developments,” Chiwewe says. An unprecedented level of forecasting will allow trends to be predicted possibly weeks in advance. “This has never been seen before.” Using sources as disparate as economic data and traffic flow, predictive models will be developed. Why Gauteng? “It’s the economic hub of the country. Most of the polluted areas are located in the Vaal Triangle and the City of Johannesburg has the most pollution in the country. Ekhuruleni is heavily polluted.”

The intention is to extend the project not only to the rest of South Africa, but also to Africa as a whole. Currenty, facilities are situated in Kenya and South Africa. The local branch will share space with Wits University in the Tshimologong Precinct technology hub in Braamfontein, Johannesburg. It forms part of a 10-year government-supported investment programme through the Department of Trade and Industry.

For all the concern about Johannesburg’s air quality, Chiwewe describes the city as progressive in its approach to the problem. “Monitoring is quite good,” he explains. “Johannesburg created its air quality management plan as far back as 2002. That’s before it was even needed.” But the monitoring network is sparse. The project’s researchers will have a mere eight locations in and around Johannesburg to work with. And, admittedly, the monitoring that’s been done was put in place mainly for compliance. Not much was done with the data. But that will change in a future scenario: predictive analysis will be applied to data obtained using IoT technology to add to the existing data from monitoring stations. Satellite and weather data as well as GPS inputs could be integrated with IBM’s analytic systems using specialised algorithms to interpret and handle the information provided.

“We also have some technology that allows us to use existing stations and form a kind of ‘virtual’ station,” he says. There’s an intention to introduce mobile monitoring stations, to get to pollution at source if possible. Says Chiwewe, this will provide good inputs into sustainable development. “We want to be able to empower the authorities to make decisions when it comes to planning.” That would cover everything from future industry to power-generation facilities and roads. It would be useful to address more immediate, short-term needs, too. “To warn of excessive pollutants and issue alerts to citizens, for instance.” Althought the insights generated as a result of the project and others like it will be aimed largely at governmental and institutional users, there’s no reason it should not be avail-able to the public as part of their decision support system.

For now, the aim is to concentrate on the air quality part of the project – the What and When elements. “We must make sure there is greater situational awareness. Later, we will focus more on decision support, including air quality forecasting.” (The When, How and Why parts.) “What we envisage is renewable energy forecasting,” says Solomon Assefa, director of IBM’s SA research lab. What’s the benefit there? “For solar stations, we will be able to forecast the yield they will get.” Although this has obvious relevance at a local level where the solar collectors are located, it also has implications for overall management of an electricity grid. For example, according to IBM, UK energy giant SSE is piloting the company’s technology to help forecast power generation at its wind farms in Great Britain. The system is able to forecast energy for individual turbines and includes visualisation tools to show expected performance several days ahead. One solar monitoring system in Japan involves managing and controlling energy from a plant’s 890 000 solar panels. In the USA, renewable energy forecasting technology is being made available to government agencies, utilities and grid operators to support supply and demand planning.

Collaborations in China include working with the country’s largest wind power solution provider to use IoT, cloud computing, big data analytics and other advanced technologies to drive innovation and elsewhere, using cognitive forecasting techno-logies to help integrate more energy into the grid. In Beijing, one of the world’s most advanced air quality forecasting and decision support systems is able to generate high-resolution kilometre-square pollution forecasts 72 hours in advance and pollution trend predictions up to 10 days into the future. It models and predicts the effects of weather on the flow and dispersal of pollutants as well as the airborne chemical reactions between weather and pollutant particles. The city is well on its way to reaching its goal of a one-quarter reduction in ultra-fine particulate matter by 2017.

We can’t leave it until tomorrow to act on air pollution. It’s the world’s single largest environmental health risk, agrees Assefa. For the moment, breathing Johannes-burg’s air is not as hazardous as doing so in some of the world’s megacities. But things change – and we need to be ahead of the curve. What brings solutions within reach is the combined power of the Internet of Things and cognitive computing. What does it mean for IBM, taking part in projects such as these? “We are developing platforms that can be used all around the world,” says Assefa. “For now, concepts such as IoT and machine learning may seem like remote concepts. But we want to show that they can be used for this purpose. “We are working on problems that matter.”

Cognitive technologies  understand data and use it to tune a predictive model that shows where the pollution is coming from, where it will likely go, and what will be its potential effect, allowing more informed decisions about how to improve air quality.

Machine learning technologies  self-configure, improving in accuracy and automatically adjusting predictive models to different seasons and topographies. They blend various predictive models, including traffic flow, weather forecasting, air and atmospheric pollution and economic data to help explore various “what if” scenarios and better understand the consequences of certain actions, such as optimising or changing traffic flows, relocating industry, switching to renewables and even introducing more green areas into a city. (Source: IBM)

This article was originally published in the April 2016 issue of Popular Mechanics magazine.