Background Community noise pollution can adversely impact health, yet noise has rarely been systematically measured in United States (U.S.) cities for epidemiological research. Objective Collaborating with the Multnomah County Health Department, we developed an exploratory measurement campaign to systematically capture community noise in Portland, Oregon, U.S. to inform environmental health research and practice. Methods We identified short-term measurement locations using weighted probability sampling and developed a protocol for deploying Class 1 sound level meters at identified sites to measure sound levels continuously for at least five days. We calculated daytime, nighttime, and daily-average noise metrics including day-night average sound levels (DNL), day-evening-night levels (Lden), intermittency ratios (IR), and 10th- and 90th-percentile noise levels (L90, L10). We evaluated noise metrics by built environment and sociodemographic characteristics at the census tract level and performed machine learning-based cluster analysis to identify locations with similar exposure profiles. Nine additional locations were sampled continuously for one year to assess agreement between short- and long-term noise measurements. Results DNL ranged from 49.6 to 86.7 decibels across short-term sites (n = 217). DNL exceeded U.S. Environmental Protection Agency guidelines at 78% of sites, and nighttime noise exceeded World Health Organization guidelines at 90%. Short-term sites in census tracts with higher median income and proportion of white population had lower DNL compared to lower median income and proportion of white population census tracts. Cluster analysis revealed four noise profiles: low LAeq/moderate IR sites usually occurring in residential neighborhoods, high LAeq/moderate IR sites adjacent to major roads, moderate LAeq/high IR sites within 1–2 city blocks of major roads, and high LAeq/low IR and low LAeq/low IR sites near highways or parks, respectively. impact <UnorderedList Mark="Bullet"> <ItemContent> This study reveals a high prevalence of potentially harmful community noise exposure levels in a medium-large city in the United States, particularly in lower-income and racially diverse neighborhoods. By identifying groupings of sites with similar noise exposure profiles, we establish a foundation for exploring built environment drivers of noise and differential health impacts of multidimensional noise exposures. The measurement protocol and database of noise measurements collected provides tools for researchers and communities (available upon request) to investigate noise exposure patterns, environmental justice concerns, and associated health impacts, with further applications for predictive modeling to estimate individual-level exposures in epidemiologic studies. </ItemContent> </UnorderedList>

Systematic measurement and machine learning-based profile characterization of community noise in a medium-large city in the United States

期刊论文
发表日期：

2025年7月19日

查看全文

Carson Mowrer,
Andrew Larkin,
Charlotte Roscoe,
Stephanie T. Grady,
Junenette L. Peters,
Brendon Haggerty,
Perry Hystad,
Matthew Bozigar

摘要

Background

Community noise pollution can adversely impact health, yet noise has rarely been systematically measured in United States (U.S.) cities for epidemiological research.

Objective

Collaborating with the Multnomah County Health Department, we developed an exploratory measurement campaign to systematically capture community noise in Portland, Oregon, U.S. to inform environmental health research and practice.

Methods

We identified short-term measurement locations using weighted probability sampling and developed a protocol for deploying Class 1 sound level meters at identified sites to measure sound levels continuously for at least five days. We calculated daytime, nighttime, and daily-average noise metrics including day-night average sound levels (DNL), day-evening-night levels (L_den), intermittency ratios (IR), and 10^th- and 90^th-percentile noise levels (L₉₀, L₁₀). We evaluated noise metrics by built environment and sociodemographic characteristics at the census tract level and performed machine learning-based cluster analysis to identify locations with similar exposure profiles. Nine additional locations were sampled continuously for one year to assess agreement between short- and long-term noise measurements.

Results

DNL ranged from 49.6 to 86.7 decibels across short-term sites (n = 217). DNL exceeded U.S. Environmental Protection Agency guidelines at 78% of sites, and nighttime noise exceeded World Health Organization guidelines at 90%. Short-term sites in census tracts with higher median income and proportion of white population had lower DNL compared to lower median income and proportion of white population census tracts. Cluster analysis revealed four noise profiles: low LA_eq/moderate IR sites usually occurring in residential neighborhoods, high LA_eq/moderate IR sites adjacent to major roads, moderate LA_eq/high IR sites within 1–2 city blocks of major roads, and high LA_eq/low IR and low LA_eq/low IR sites near highways or parks, respectively.

impact

This study reveals a high prevalence of potentially harmful community noise exposure levels in a medium-large city in the United States, particularly in lower-income and racially diverse neighborhoods. By identifying groupings of sites with similar noise exposure profiles, we establish a foundation for exploring built environment drivers of noise and differential health impacts of multidimensional noise exposures. The measurement protocol and database of noise measurements collected provides tools for researchers and communities (available upon request) to investigate noise exposure patterns, environmental justice concerns, and associated health impacts, with further applications for predictive modeling to estimate individual-level exposures in epidemiologic studies.