Recently, climate change, species extinction, degraded soil, and excessive demands on freshwater systems have risen to the forefront of environmental concerns. However, air pollution is still a problem that needs our attention and action.

The World Health Organization says that between 3 and 8 million people die too soon every year because the air they breathe often has dangerous chemicals in it that can hurt the lungs, cause inflammatory diseases, or weaken the immune system.

Even though there are a lot of rules in place to try and lower the amount of air pollutants that are released into the air and keep concentrations below certain levels, measurements taken all over Europe still show that concentrations are often higher than the levels that are safe for people's health and food production.

Challenges

There are several unique difficulties associated with using AI for this purpose. In the 1990s, researchers began experimenting with AI techniques to improve the accuracy of local air quality forecasts. Since algorithms and processing capability were a million times less influential today, machine learning results were only modestly better than traditional statistical approaches.

After 2012, convolutional neural networks improved classic AI tasks, reviving atmospheric scientists' interest in AI. Since 2018, multiple research studies have shown that high-quality local air pollution forecasts may be generated using advanced machine learning algorithms. Alternative and computationally far cheaper solutions to regional air pollution forecasting may soon be available using machine learning models. 

AI-powered data models can anticipate street-level air pollution and regulatory violations. Wearable sensors make air quality data accessible to everyone, allowing them to assess their air pollution exposure.

Case studies:

Some parts of the world have even worse problems. For example, the smog in the megacities of Southern and Eastern Asia, Africa, and South America is sometimes so bad that it's hard for people to work or get around.

New York

Scientists employed models with many data points to predict air pollution. Complex models were inaccessible to non-experts. Cornell engineers created a simple model. They show street-level air pollution with AI. Traffic exhaust causes fine particle pollution (PM2.5), which damages human health. The new AI program learns traffic-related air pollution concentration simulations from traffic data, topology, and weather.

City planners and regulators can utilize the machine learning model to analyze hyperlocal air pollution better. This data helps policymakers create hyperlocal air pollution reduction projects. Governments can use hyperlocal data to assess the health consequences of new transport and infrastructure developments for smarter planning. 

Barcelona

Barcelona scientists developed an AI algorithm that predicts urban NO2 exceedances using machine learning. Barcelona Supercomputing Centre researchers incorporate data from multiple sources: 

• CALLIOPE-Urban predicts air pollution at high resolutions, different heights, and anywhere in the city.
• official air quality stations and cheap sensors
• building density and other geolocated data
• meteorological data

Africa

Rapid urbanization and industrialization in Africa's big towns have made the air quality worse in recent years, leading to higher death rates and higher healthcare costs for respiratory illnesses. Different countries are working hard to clean up the air, which has put collecting data and doing studies at the top of the list. It has been hard for people in Africa to get data and tools because they are expensive and complex. Machine learning is an attempt to fill this gap. It also lets cutting-edge technology be used in places where data have yet to be officially covered, and it can be used to help researchers find errors more easily in studies about air quality in Africa.

Delhi-NCR

The Commission for Air Quality Management in NCR and Adjoining Areas (CAQM) has decided to tap the technical/academic expertise of renowned scientific institutions working on air pollution to prevent, control, and decrease air pollution in Delhi-NCR. 

In addition, projects have been proposed by the Automotive Research Association of India (ARAI) in Pune, the Institute of Information Technology (IIT) in Delhi, Swachh.io, and SASTRA University in Thanjavur. For the Commission to be more effective in its fight against NCR's air pollution issue, the Projects aim to build better capacities for air quality monitoring/ demonstrate field implementable solutions/technologies.

Image source: Unsplash