AQview in the Wild

I attended and presented at the 2022 Air Sensor International Conference (ASIC) in Pasadena, California hosted by the UC Davis Air Quality Research Center on May 11-13th. It was a pleasure and great learning experience to present on behalf of my team at the California Air Resources Board (CARB) in regards to our work developing AQview.

What is AQview?

California Assembly Bill (AB) 617 passed in 2017, requires community-focused and community-driven actions to reduce air pollution and improve public health in communities experiencing disproportionate burdens from their exposure to air pollution. As part of AB 617, CARB is required to publish this community air monitoring data online. My team is currently developing, AQview, to meet that requirement and more, our plan is to include more than just air monitoring data from AB 617 communities, but also the regulatory network data, research project data, and data from other private and NGO (non-governmental organization) efforts. AQview’s primary goal is to make all this air quality data easier to access in a single web application so that community members can better understand the air quality in their neighborhood. AQview recently launched Phase 1 of the system which includes a straight forward data download tool, and preliminary QC (quality check) process. In future releases, AQview will have a user-friendly map where community members can look at data from specific monitoring stations/sensors.

As I mentioned before, AQview will house a variety of air quality data from low-cost sensors, regulatory monitors, and other monitors such as research-grade, mobile monitoring, and new monitoring technologies. And as we collect these disparate data sets, where the pollutants and measurement technologies are much more variable, these questions arise:

• How do we message differences in data quality?
• To what level do we assess data quality for different instrument types?
• Can we evaluate data from all instrument types on the same quality scale?
• How can we assess whether sensors are appropriately sited for ambient air quality measurements?

The biggest challenge AQview is tasked with now, is figuring out how to display all the data, regardless of quality, together so that it tells a meaningful story for community members. We believe AQview’s developing QA/QC process can address these challenges, which has been outlined in three levels below.

Level 1:

This is what AQview is currently doing with our Preliminary QC – identifying the “low hanging fruit” or obvious anomalies in the data streams through three simple checks. The first is an instrument based upper limit and lower limit check. This is where my team works closely with data providers to understand what monitor model types they are using and those monitor’s limitations. Next is a statistical outlier check, where we’re looking to see if a data value is far above or below the surrounding data points. Lastly is a repeating values check, looking for repeating values over a certain time.

Level 2:

This is the process that the AQview team is beginning to develop. The level two, Enhanced QC, will take a deeper dive into the data and look for the not-so-obvious anomalies in the data streams, taking into account more historical data and further statistical analysis.

Level 3:

This process entails AQview’s long-term goals and future plans to develop an Overall Data Quality Assessment (QA), currently this level three process is in the framework development phase. It would include two data quality elements; the first would be used to assess an individual data stream’s health including data completeness, statistical variability, and statistical stability; the second would include tools and information AQview hopes to gather by working closely with data providers and community members to provide support and assess the health of monitoring networks – this would be most applicable to deployed networks using the same monitors – including things such as the instruments grade, siting attributes, calibration status, and colocation motivated adjustments.

At the end of the day, no matter what AQview does moving forward, all of AQview’s methods will be transparent and defensible for community member’s understanding.

You can view Cheryl's actual presentation session below, it begins around the 26 minute mark.