Multivariate Monitoring and Local Interpretation of Indoor Air Quality in Seoul's Metro System

Abstract

To control indoor air quality (IAQ) of the pollutants in subway systems, several key air pollutants are measured and monitored offline and also online by telemonitoring systems. Most monitoring methods of air quality in subway stations are based on a univariate approach that considers only a single pollutant such as particulate matter (PM2.5 or PM10). In this study, new multivariate monitoring and local interpretation of IAQ in an underground space has been presented to tackle this correlation problem of the air pollutants. The proposed method consists of three main components: (1) principal components analysis to reduce its dimensionality; (2) global multivariate modeling to monitor real-time data for air pollutants and to diagnose the status of IAQ in a subway station during a 1-year period; and (3) local multivariate models to precisely monitor the air quality in a subway station, including seasonal models that account for variations in pollutant concentrations throughout the year in Korea. Monitoring results in Seoul metro system illustrate that it can monitor the IAQ with a multidimensional aspect and interpret the effective variables of the cause of contaminated IAQ. Also, seasonal monitoring models show more reliable and more meaningful interpretation results than a global multivariate model, because local models better isolate air quality characteristics that vary seasonally and allow for more specific monitoring of air pollutants. Overall, we expect that the use of a multidimensional monitoring method will help analyze the effects of various air pollutants on the passengers' health and will help the operation of indoor air control in subway stations with the operator's decision making and with adequate ventilation system.