How we collected the data

EDF partnered with researchers from the University of California, Berkeley and Berkeley Lab, along with citizen scientists from the West Oakland Environmental Indicators Project to deploy a network of 100 low cost black carbon sensors across West Oakland for 100 days in the summer of 2017 (from May 19 to August 26). The study was described in detail in a peer-reviewed paper published in Environmental Science & Technology in June 2019.

Development of the sensors

The black carbon sensors used in this project were designed and built by the Berkeley team–a detailed description of the technology and performance of the sensors was published in 2018. Prior to deployment, the researchers validated the sensors by operating them outdoors alongside high-grade black carbon monitors maintained and operated by the Bay Area Air Quality Management District (BAAQMD). All of the sensors used in the 100×100 study agreed closely with the BAAQMD's monitors.

Installation of the network

The Berkeley team installed the black carbon sensors across the West Oakland neighborhood outside of homes, local businesses, community organizations, and industrial sites. They placed several sensors at two BAAQMD air monitoring sites to allow for continuous validation throughout the campaign. They also placed sensors on Maritime Street along the Port of Oakland boundary as well as at three sites near the San Francisco Bay that are typically upwind of the Port of Oakland and other sources of pollution in West Oakland. The measured black carbon concentrations at these upwind sites represent "background" concentrations on top of which local sources of pollution build.

Data collection and analysis

Sensors transmitted data from the 100×100 network wirelessly to a central database and averaged to 1-minute time resolution. In total, the network collected approximately 22 million measurements of 1-minute average black carbon concentrations. The Berkeley research team remotely monitored sensor performance throughout the campaign and conducted regular maintenance to maximize data quality and completeness. They replaced and repaired sensors experiencing problems as necessary. To prepare a database of valid black carbon concentrations, erroneous data resulting from technical problems were removed. Hourly average black carbon concentrations were calculated for all hours with at least 48 minutes of valid data, resulting in more than 200,000 hours of hourly average black carbon concentrations spanning the 100 days and 100 sites (84% data completeness). The Berkeley team then analyzed the final data set, evaluating daily and weekly trends across the network and identifying locations where pollution patterns differed.