The findings arrive against the backdrop of Chiang Mai’s well-documented air pollution crisis. The northern Thai province experiences severe haze episodes every year, driven in large part by seasonal biomass burning across Southeast Asia, and it holds an unenviable distinction: it has more road underpasses than any other province in Thailand. These semi-enclosed, poorly ventilated structures are designed to move traffic efficiently beneath intersections and communities, but they also act as traps for vehicle exhaust and airborne particles. Despite their ubiquity, the dynamics of PM2.5 pollution inside such tunnels and the health risks they pose to the people who pass through them have remained largely unstudied, a gap the new research set out to close.

To build a detailed picture of pollution inside the tunnel, the research team deployed five low-cost DustBoy sensor stations along the Don Kaew underpass and collected data continuously from August 2022 to July 2023. The monitoring campaign yielded an extraordinary 385,622 records spanning a broad suite of pollutants and environmental variables, including carbon monoxide, carbon dioxide, nitrogen dioxide, sulfur dioxide, ozone, PM10, PM2.5, total volatile organic compounds, formaldehyde, temperature, relative humidity, air pressure, wind speed, wind direction, and rainfall. Data were obtained at no cost through collaborations with Thailand’s Pollution Control Department, the Department of Highways, and the Climate Change Data Center at Chiang Mai University, allowing the researchers to assemble one of the most granular datasets ever compiled for a Thai road underpass.

The analytical core of the study was a multiple linear regression model, built using a stepwise selection procedure in the R statistical environment, designed to forecast PM2.5 concentrations inside the tunnel. The researchers deliberately stratified their analyses by weekday versus weekend and by daytime versus nighttime periods, recognizing that traffic patterns and meteorological conditions shift systematically across these categories. The resulting model achieved a remarkably high goodness of fit, with an R-squared value of 0.9995 and a p-value below 0.001. Among the significant predictors were PM10, temperature, relative humidity, and monthly and daily time variables, indicating that coarse particle levels and local weather conditions carry most of the explanatory power for fine particle variability within the tunnel environment.

Correlation analysis revealed tight coupling among several pollutants. PM2.5 and PM10 moved almost in lockstep, with a correlation coefficient of 0.9995, suggesting that within the tunnel both size fractions share common sources, most plausibly vehicle exhaust and road dust resuspended by passing traffic. Formaldehyde and total volatile organic compounds correlated at 0.9999, as did formaldehyde and carbon dioxide, pointing to combustion-related emissions as a dominant driver of the tunnel’s chemical profile. In contrast, temperature and relative humidity were strongly negatively correlated at minus 0.7967, and relative humidity showed a negative relationship with ozone at minus 0.5939, reflecting the physical constraints of the semi-enclosed space where normal atmospheric photochemistry is suppressed and humidity responds inversely to heat generated by traffic and trapped solar energy.

The temporal patterns the model uncovered carry clear implications for public health planning. PM2.5 concentrations inside the underpass were significantly higher on weekdays than on weekends, with the weekly peak occurring on Tuesdays, and levels were elevated during daytime hours and during the dry season months when regional haze blankets northern Thailand. This rhythm mirrors the commuting cycle and the annual agricultural burning calendar, meaning that the tunnel’s pollution burden is not constant but pulses with human activity and seasonal weather. For urban planners and public health officials, the message is that exposure mitigation efforts could be targeted in time as well as in space, concentrating ventilation interventions and public advisories on weekday daytime hours and the dry season.

To translate concentrations into health consequences, the team applied a comparative risk assessment framework grounded in Thai and World Health Organization exposure factor standards. They calculated inhalation rates and daily intake for different population groups, then compared the most vulnerable scenario, infants inside the underpass, against a reference group of healthy adults at the surface. The result was a relative risk of roughly 3.5 times for the infants, driven by the combination of higher pollutant concentrations in the tunnel and the physiological characteristics of early childhood, including higher breathing rates relative to body weight and developing respiratory systems. Even accounting for the brevity of a typical two-minute crossing, the per-trip intake of PM2.5 inside the tunnel was 1.55 times higher than at the surface, and the statistical comparison between tunnel and control site concentrations was highly significant, with a t-statistic of 8.432 and a p-value below 0.001.

The study’s methodological choices also carry lessons for environmental monitoring in resource-constrained settings. Low-cost sensors like the DustBoy units used here have sometimes been viewed with skepticism because of accuracy concerns, but the researchers argue that dense networks of such devices, carefully deployed and interpreted, can capture the fine spatial and temporal structure of pollution in microenvironments that fixed regulatory stations miss entirely. The near-perfect fit of the regression model suggests that within a confined tunnel environment, PM2.5 behavior is highly predictable from a small set of co-measured variables, which opens the door to real-time forecasting systems that could warn tunnel users or trigger ventilation on demand. Similar multiple linear regression approaches have been successfully applied to particulate forecasting in Malaysia, the Greater Klang Valley, and Chiang Rai Province, situating this work within a growing regional tradition of accessible statistical air quality modeling.

The broader context makes the findings more urgent rather than less. Air pollution has emerged as one of the world’s leading health risk factors, and chronic respiratory disease imposes a substantial and growing burden across Southeast Asia. Chiang Mai itself has been the subject of extensive research linking PM2.5 exposure to hospital admissions, respiratory illness, and even lung cancer mortality patterns, with studies documenting the chemical composition, oxidative potential, and biomass burning origins of the province’s haze. What the new study adds is the recognition that the microenvironments people move through daily, such as road underpasses, can amplify exposure far beyond what ambient monitoring stations would suggest, and that these amplifications fall hardest on those least able to tolerate them.

The authors conclude that even brief transit exposures in unventilated underpasses generate disproportionate health burdens for vulnerable populations, and they call for improved mechanical ventilation and health-informed urban planning as Chiang Mai’s underpass infrastructure continues to expand. The implications extend well beyond a single tunnel in northern Thailand. Cities across Asia and beyond are building grade-separated road infrastructure to ease congestion, and each semi-enclosed structure represents a potential exposure hotspot for pedestrians, motorcyclists, cyclists, and vehicle occupants alike. As the researchers demonstrate, a year of affordable sensor data and a well-specified regression model are enough to reveal these hidden risks with precision, giving municipal authorities the evidence base they need to design ventilation systems, schedule maintenance, and issue health advisories that protect infants, the elderly, and others whose lungs pay the highest price for two minutes of urban convenience.

Subject of Research: PM2.5 exposure and respiratory health risk assessment for vulnerable populations in an urban road underpass in Chiang Mai, Thailand, using multiple linear regression forecasting

Article Title: Assessment of PM2.5 exposure and respiratory health risks for vulnerable populations in urban underpass: a predictive MLR approach in Chiang Mai, Thailand

Article References: Kongpitak, P., Chetiyanukornkul, T., & Nonkrasaesin, C. (2026). Assessment of PM2.5 exposure and respiratory health risks for vulnerable populations in urban underpass: a predictive MLR approach in Chiang Mai, Thailand. Environmental Science and Pollution Research, 33(27), 13989-14005. https://doi.org/10.1007/s11356-026-38171-3

Image Credits: AI Generated

DOI: 10.1007/s11356-026-38171-3

Keywords: PM2.5, air pollution, road tunnel, Chiang Mai, Thailand, multiple linear regression, health risk assessment, vulnerable populations, low-cost sensors, respiratory health, urban planning, ventilation