Noise pollution in cities has major negative effects on the health of both humans and wildlife. Using iPhones, we collected sound-level data at hundreds of locations in four areas of Boston, Massachusetts (USA) before, during, and after the fall 2020 pandemic lockdown, during which most people were required to remain at home. These spatially dispersed measurements allowed us to make detailed maps of noise pollution that are not possible when using standard fixed sound equipment. The four sites were: the Boston University campus (which sits between two highways), the Fenway/Longwood area (which includes an urban park and several hospitals), Harvard Square (home of Harvard University), and East Boston (a residential area near Logan Airport). Across all four sites, sound levels averaged 6.4 dB lower during the pandemic lockdown than after. Fewer high noise measurements occurred during lockdown as well. The resulting sound maps highlight noisy locations such as traffic intersections and quiet locations such as parks. This project demonstrates that changes in human activity can reduce noise pollution and that simple smartphone technology can be used to make highly detailed maps of noise pollution that identify sources of high sound levels potentially harmful to humans in urban environments.

We collected sound measurements within four different urban sites in Boston, Massachusetts. Working in small teams of 2-4 people, we used the mobile app SPLnFFT to collect sound level data in A-weighted decibel readings using smartphones. We exclusively used iPhones for data collection for consistency in hardware and software. Before each collection, we calibrated each iPhone to the same standard, which was used for every collection outing. We recorded the L50 value (the median sound level) for each recording because the L50 value is less affected by short bursts of loud sound than the mean reading. Recordings ran for approximately 20 seconds each. We recorded all sound measurements between 9 am and 5 pm on workdays to avoid the influence of rush-hour traffic, and only collected data on days without rain, snow, or strong wind to prevent inaccuracies due to weather. Within these conditions, we collected sound measurements over multiple days and at different times to ensure representative data. We followed these procedures for both collection cycles (2020 during lockdown and 2021 after lockdown had been lifted). The 2017 data were collected for an unrelated noise pollution project conducted by previous members of the Primack Lab and were not collected with the exact parameters established for the 2020 and 2021 collections. However, we found these noise data to be valuable given that they could be used to compare lockdown sound levels to the soundscape before the COVID-19 pandemic.

We used R Studio to create sound maps from the individual data points in a way that allows for spatial visualization of the soundscape before, during, and after the pandemic lockdown. To test for statistically significant differences in sound level between years, we performed Welch’s t-tests on the raw data for all sites comparing lockdown (2020) measurements to pre (2017) and post (2021) lockdown measurements. Given the hypothesis that 2020 would have lower sound levels at each site, we report the results of one-tailed t-tests.