Better Air Quality Forecasting through Satellites

Monitoring air quality accurately has always been a challenge, its possible solutions filled with far too many shortcomings. Large regulatory-grade sensors are accurate, but they can only detect pollutants for a single location. Since air quality can vary significantly over a distance of a few blocks, these traditional units give an oversimplified view. Internet of Things (IoT) air quality stations can grant a higher level of resolution with satisfactory accuracy, though their high cost is problematic. Thanks to technologic improvements in air quality components and low power IoT communications networks, units that can be mounted to buildings and light poles are now significantly cheaper. However, many sensors are still chemically based and require frequent maintenance by trained personnel.

Some companies, like Aclima, are pioneering mobile air monitoring by mounting sensors to vehicles. While these provide a more varied view of pollution levels, data collection suffers from validity issues, including lack of samples from the same point over time.

Then there is the air quality modeling approach, which asks, What if we didn’t have to deploy units at all? What if you could accurately predict the air quality for any point on earth instantaneously? These are questions that a group of software companies like EarthSense, BreezoMeter, and IQAir are trying to answer. Their approach is bold: take data from the few points we currently have and fill in the gaps with a model that considers other information like traffic conditions, weather, economic data, zoning ordinances, and historical data. For instance, they may take data from two government air quality stations 10 miles apart and use their predictive model to estimate the air quality for every point in between.

While an ambitious approach, there’s one large, lingering issue: data starvation. Even as air quality becomes an increasing concern for society and fosters the development of more tools to measure it, we don’t have enough data to feed into these models. Clever modelers can use several tricks, like computational fluid dynamics and AI, to try to solve air pollution’s three-dimensional puzzle. Even with these tricks of the trade, the lack of overall data points continues to be an obstacle.

Measuring Your Street’s Air from Orbit

With the launch of a new satellite network in February of 2020, scientists are aiming to bring more air quality data into the realm of public availability. South Korea’s Geostationary Environment Monitoring Spectrometer was launched first and will be followed by satellites from NASA and the European Space Agency. This “constellation” of geostationary satellites will take air quality readings of the surface every hour, as opposed to older space-based instruments that measure once a day. Collectively, the network will provide accurate data for large swaths of the planet’s surface and potentially reduce the need for individual sensors.

This type of data stream can help overcome data starvation. With their appetites satiated by this satellite network, companies will be able to develop much more accurate and defensible predictions. At present, the information being modeled remains interesting in a general sense but not rigorous enough to inform policy decisions. A more robust air quality modeling process could change this scenario while significantly lowering the financial barrier for many cities.