At Pi School, we always address the most pressing issues of the present, that will contribute to building tomorrow’s world. And what better way to do it than through our artificial intelligence programme, where the brightest minds come together to develop advanced technological solutions.
A challenge for the European Space Agency
One of the School of AI projects, sponsored by the European Space Agency (ESA), explored the available solutions that will enable countries to measure and reduce emissions and increase adaptation efforts. ESA challenged two engineers to work on air pollution forecasting in the context of the SENTINEL-5 mission, whose main objective is to perform accurate atmospheric measurements relating to air quality, climate forcing, ozone and UV radiation, providing a daily global coverage.
Similar to forecasting weather, there are models to predict levels of air pollution and air quality, but these are more complex, as they require mathematical simulations of how pollutants disperse in the air. Over eight weeks, Luka Sachsse and Maximilien Houël teamed up to study and enhance a weather-forecasting model, based on earth observation data and developed with deep-learning techniques.
Today’s dataset is provided by a combination of ground measurements and satellite data. By applying deep-learning methods to satellite imagery, it is possible to produce better data and perform a more accurate analysis. Between pair lessons, mentoring and research, Luka and Maximilien explored different ways to improve the quality of the images and, thus, to achieve more accurate air pollution forecasting.
Forecasting air pollution: a step towards climate action
Today, air pollution is one of the most harmful threats to human health. Forecasting is of high relevance for adopting effective control measures, as well as for informing citizens about preventive measures to take, especially in urban areas. Many cities inform their inhabitants about air quality and advise on when to avoid outdoor activity, for example.
“People feel more concerned about what is closer to them. This project can affect the public understanding of atmospheric pollution as, with better resolution images, it will be easier to pass certain concepts.” Maximilien
More accurate forecasting can also help to provide better answers in terms of environmental disasters, such as the latest fires in Australia or the Taal volcano eruption in the Philippines. “In the context of extreme environmental events such as the fires in Australia, having a prediction of how aerosols or trace gases are spreading would be a determining factor,” says Luka. In such cases, an accurate detection model with space observation data can provide information about where help is needed the most and anticipate how an event will evolve.
“Better-resolution forecasting of pollution variables provides a sharper eye for analysing the atmosphere. It makes it possible to know more precisely where the pollution is, how it is evolving, and what the causes are,” adds Maximilien. A promising tool to be integrated at different levels, from the decision-making process to everyday routine.
"The future applications of this project can allow us to more accurately trace and forecasts air pollutants." Luka
This project casts new light on the use of artificial intelligence in terms of climate change and sustainability. Thanks to the research in this field, it will be possible to take different forms of action, from awareness campaigns to national action plans. Earth observation is one of the most important activities of ESA in contributing to the answer to climate change, and Pi School is proud of having them as a partner in its quest to design the future.
“Humanity is now standing at a crossroads. We must now decide which path we want to take. How do we want the future living conditions for all living species to be like?” Greta Thunberg
Social Media Marketing Specialist