FairAir

Filling the gaps in air pollution monitoring

Richmond Chevron oil refinery, 2012. Photo via Flickr Creative Commons.

The Big Picture

4 million people die annually due to outdoor air pollution

We can push back on polluters with better air quality monitoring. The governent often places its sensors away from pollution sources. With the price of sensors falling, community groups and residents are starting to buy their own from companies like PurpleAir.com. So far, most sensors wind up in wealthier neighborhoods that face less pollution. Our tool aims to change this.

FairAir predicts which areas on a map face the most air pollution and recommends the best place to put new sensors.

Optimal placement of sensors ensures:
* disadvantaged communities have access to air quality data,
* improves air quality models for everyone,
* saves money from being wasted on redundant sensors.

Use Cases

Activists


Community advocates can use FairAir to help make a case to donors that their community needs sensors. The tool can also help establish a fair way to place the sensors, focusing them on the areas most likely to be impacted by air pollution.

Donors


FairAir can help foundations, governments, and others who donate air pollution sensors to identify the communities most in need of air quality monitoring. We base our recommendations on where we predict air pollution to be the worst and distance from other sensors.

Residents


Live far from a sensor and can't afford one? FairAir offers a free prediction of the air quality in your neighborhood using pollution, weather, and topographic data. No model is perfect, and ours is improved when more sensors are placed optimally.

How It Works

FairAir uses artificial intelligence techniques to predict the levels of air pollution across a region. The prediction combines hourly air and weather sensor data from PurpleAir.com, the U.S. Environmental Protection Agency, and the National Oceanic and Atmospheric Administration. The model also accounts for topography and land use.

The model focuses on one specific type of air pollution: particulate matter that is 2.5 micrometers in diameter, commonly referred to as PM2.5. Pollution particles this small can move deeply into human lungs. Research has linked PM2.5 exposure to many health problems, including asthma, irregular heartbeat, nonfatal heart attacks, and premature death.

Using a combination of static and dynamic inputs, as well as existing PurpleAir sensor data, FairAir's random forest model predicts hourly PM2.5 levels at intervals of 500 meters on a regional grid. Each gridpoint is then scored on two factors: its average predicted PM2.5 level over the past seven days and the number of existing sensors in its vicinity. The user is then shown locations where both sensor coverage is sparse and predicted PM2.5 is high. Adding real-life sensors to such areas would not only provide real-time air quality readings but also improve the accuracy of our model, which includes actual neighbor PM2.5 values as features.

Testimonials

How To Help

If you're impacted by dirty air, organize with your neighbors, community, and local groups to fight for your right to clean air. Get involved with local organizations such as:
* Communities for a Better Environment (Bay Area+)
* West Oakland Environmental Indicators Project (Oakland)
* PODER (San Francisco)
* Asian Pacific Environmental Network (East Bay)
* Greenaction for Health and Environmental Justice (Bay Area+)
* Youth United for Community Action (East Palo Alto)

If you're not directly impacted, support communities who are affected with your time and financial resources.

Team

We are FairAir

Ben Arnoldy
Jake Miller
Sameed Musvee
Mark Paluta
Angshuman Paul