Classification of Indoor-Outdoor using Global Positioning System (GPS), Temperature and Brightness Level for Personal Exposure Assessment

Abstract

Personal exposure to air pollutants is critical for assessment of health effects and identification of control strategy. Personal exposure is affected by concentration at various microenvironments and time spent in the microenvironments. Accurate determination of individuals duration in microenvironments is essential for exposure assessment of air pollution. In previous studies, personal time-activity pattern was determined by questionnaires, observations, recall interviews or time-activity diaries (TAD). However, these methods are highly affected by the accuracy of recall, reliability and compliance of each individual. To overcome the weaknesses, global positioning system (GPS) signal has been suggested to classify personal time-location information. Temperature and brightness level were also introduced for detecting the transition between indoor and outdoor. The purposes of this study were to compare the accuracy of indoor-outdoor classification analyzed with GPS, temperature and brightness level and to determine an optimal way to combine the three indicators. GPS signal, temperature and brightness level were measured in 159 indoor places during summer, autumn and winter in 2015 (July to December). Selection of the 159 indoor places was based on peoples activity patterns. A field technician carried GPS signal receiver, temperature data logger and illuminometer while he visited the 159 indoor places. Temperature and brightness level criteria were determined to detect the transition between indoor and outdoor. Through comparing the sensitivity and specificity, criteria with the highest accuracy were determined for each season. Combinations of GPS signal, temperature and/or brightness level were applied to classify indoor and outdoor. The consistencies with manually written TAD were calculated for each combination. Temperature and brightness level were significantly different between indoor and outdoor levels in three seasons. Temperature criterion for detecting transfer was 2 °C changes in 2 minutes in all three seasons. Brightness level criterion for detecting transfer was 3900 lx in all three seasons. When GPS signal was used, average simple percentage agreements between GPS and TAD were 73.6 %, 72.9 % and 72.1 %, in summer, autumn and winter, respectively. Indoor-outdoor classification was improved when using combinations of GPS, temperature and brightness level. When the three indicators used, simple percentage agreements were improved by 29.6 %, 33.2 % and 34.3 %, in summer, autumn and winter, respectively. The combination of three indicators could replace traditional TAD method classifying individuals time-location.