Paired Student's *t*-test, † *p* < 0.05, ‡ *p* < 0.01. § Reference periods were the one-hour period before cooking periods.

Table 5 shows the association between air pollutants concentrations (24-h average concentration of air pollutants in each house as dependent variable), and household characteristics by using the generalized estimating equations model. This study revealed that CO concentrations were positively associated with the number of occupants, cleaning, smoking, incense burning, mosquito coil burning, and negatively correlated to cooking with a statistical significance. Indoor CO2 concentrations were positively associated with the number of occupants, air-conditioning use, smoking, incense burning, and negatively correlated to mosquito coil burning with a statistical significance. In addition, significantly higher NO2 levels were found in the homes with smokers than homes without smokers. There were significantly positive associations between indoor SO2 concentrations and smoking and incense burning. In terms of O3, indoor O3 concentrations were positively associated with the window opening and negatively correlated to the number of occupants, incense burning, and essential oil use with a statistical significance. For PM2.5, it was positively associated with cleaning and incense burning with a statistical significance.


**Table 5.** Association between air pollutants concentrations (24-h average concentration of air pollutants in each house as dependent variable), and household characteristics: generalized estimating equations.

Generalized estimating equations (GEE) † *p* < 0.05, ‡ *p* < 0.01.

#### **4. Discussion**

Our results showed that the outdoor concentrations of O3 and PM2.5 were significantly higher than indoor concentrations. The Kaohsiung City is a city with intense traffic and heavy industries, and previous studies believed SO2, NOx, PM2.5, and CO were the major conventional air pollutant in steel plants, oil refineries, and vehicular exhaust emissions [12–14,24]. In addition, outdoor O3 might be formed by the photochemical reaction of nitrogen oxides absorbing sunlight, and VOCs [25,26]. According to the PSI database from 2010 to 2012 of Taiwan EPA, only O3 and total suspended particulate (TSP) would exceed the standard [27]. This may be the reason why outdoor PM2.5 and O3 concentrations were higher than indoor concentrations. In our study, outdoor median PM2.5 levels (90 μg/m3) were higher than both the National Ambient Air Quality Standards of UAS and Taiwan EPA with the 24-hour standard for PM2.5 of 35 μg/m3. In addition, the median value of indoor PM2.5 concentrations (40 μg/m3) was also higher than the criteria of indoor air quality (IAQ) standards of Taiwan EPA (35 μg/m3/24 h). In our study, indoor CO, CO2, and NO2 levels were significantly higher than outdoor levels. The number of occupants and human activities such as cooking, smoking, etc. might be the factors affecting indoor pollutants whereas liquefied petroleum gas (LPG), not electric stoves, was the main cooking way in Kaohsiung City [28,29]. In addition, most of the houses were just by the roads and very close to the mobile sources in Kaohsiung City, which was thought of as a traffic-intensive city with the number of cars and motorcycles of approximately 430,000 and 1,230,000, respectively, in 2010 [30]. Thus, the main combustion products of vehicular engines such as CO, NOX, etc. entering the houses through cracks and windows might be the reason why the indoor concentrations of CO, CO2, and NO2 were higher than outdoor concentrations [15,16].

In comparison with traffic, industrial, and general areas, the highest household CO concentration was found in the traffic area among the three areas. According to the previous study, the greatest source of CO (more than 90%) in cities was motor vehicles [24]. The high traffic flow in the traffic area might be the reason for the observation. For CO2, our study indicated that the lowest household CO2 level was in the general area among the three areas. The main source of CO2 was from human respiration [24,31]. The number of residents might be one possible reason since the number of residents > 4 people in the traffic area, the industrial area, and general area were 40.40%, 57.01%, and 34.41%, respectively. We also found both household NO2 and O3 concentrations of the industrial area were

lowest among the three areas, which was not consistent with the observations of previous studies that ambient NO2 was related to industrial activities [24], and outdoor O3 might be formed by the photochemical reaction of nitrogen oxides absorbing sunlight, and VOCs [25,26]. We believed these may be related to Taiwan EPA' s policies and efforts to control air pollution from stationary sources after that the "Stationary Pollution Source Air Pollutant Emissions Standards" was passed in 1992, and the "Air Pollution Control Act Enforcement Rules" was also implemented in 2003.

In regard to the effects on the window opening, our study displayed that household NO2 and PM2.5 concentrations during window opening periods were significantly higher than that during reference periods. NOX and PM were related to traffic emissions [24,32], and most of the houses in Taiwan were adjacent to roads, so window opening might increase indoor NO2 and PM2.5. For the influence of cooking, there were many simulated experiments exploring the air pollutant emissions of cooking-related fuel combustion [29,33–36], and they demonstrated that CO, CO2, NOX, and PM2.5 would be emitted by the process of the experiments. Although CO also was produced by cooking, it was revealed that combustion of high-grade fuels (such as natural gas, and LPG which contained propane, butane, etc.), the main fuel-burning stoves use in Taiwan households usually produce much less CO than combustion of low-grade fuels [29,33]. In the previous study, Delp et al. revealed the residential cooking exhaust hoods could not completely capture the pollutants and their efficiency was highly variable [37]. Our results showed that indoor CO2, NO2, and PM2.5 levels during cooking periods were significantly higher than during reference periods, but the indoor CO level during cooking periods was lower than during reference periods, possibly indicating that the emission rate of CO2, NO2, and PM2.5 might be higher than the capture rate of the exhaust hood and the emission rate of CO might be lower than the pollutants capture rate of the exhaust hood.

In terms of influence factors, we found there were significantly positive correlations between the number of occupants and CO and CO2 concentrations. Our study was consistent with the observations of the previous study that CO2 was produced by human respiration [24,31]. In addition to the combustion, the indoor CO also was related to the status of residents; the previous studies revealed either a smoking person or person with inflammatory diseases exhaled higher CO levels than control group [38,39]. We also found smoking was significantly positively associated with household CO in our study. According to previous studies, smoking, incense burning, and mosquito coil burning were significantly positively associated with CO, CO2, SO2, NOX, and PM [40–42], and these results were consistent with our observation. The cleaning behavior would increase indoor PM2.5 and CO levels; it was consistent with the previous study that indoor PM2.5 and PM5 levels could be elevated by the cleaning behavior of dry dust, and vacuuming [43]. In addition, commercial cleansers and disinfectants contain VOCs [44], and El Fadel et al. found VOCs concentration was positively correlated with CO concentration [45]. We also revealed that air-conditioning use was positively associated with indoor CO2 concentrations with a statistical significance, which was consistent with a previous observation that CO2 levels were higher in mechanically ventilated buildings than in naturally ventilated buildings [46]. There was a significantly negative association between essential oil use and O3 concentration. The commercially available essential oils contain many VOCs (e.g., D-limonene, α- pinene, etc.) [47], in addition, a study displayed that indoor VOCs level had increased significantly after burning essential oils [48]. O3 was one of the indoor oxidants [49,50], and Waring et al. demonstrated that 68% of all O3 reactions were with D-limonene, and 26% of all O3 reactions are with α-pinene [50]. This might be the reason why the essential oil use could decrease the O3 level. Finally, by questionnaire, it was found that window opening was significantly correlated with increased O3 concentration, which was not consistent with the results from the time–microenvironment–activity-diary that only NO2 and PM2.5 levels during the window opening periods were significantly higher than that of reference periods. We believed O3 was a major component of photochemical pollution, so it is more relevant to outdoor sources than indoor sources. Thus, compared with the households which closed the windows, the households which opened the windows had a significantly higher 24-hour average concentration of O3. When comparing the window opening periods with the reference periods (two one-hour periods before and after window

opening periods), there was no significant variation in atmospheric O3 concentration in a short time (within three hours). For PM2.5 and NOX levels, there was no significant difference between households which closed and opened the windows, the possible reason might be that PM2.5 and NOX could come from both indoor (cooking) and outdoor (traffic) sources.

#### **5. Conclusions**

This study explored the concentration of indoor air pollutants in different areas including traffic, industrial, and general areas within an industrial city. Moreover, this study also revealed household NO2 and PM2.5 concentrations during window opening periods were significantly higher than that of the reference periods with increased concentrations of 18.71 ppb, and 7 μg/m3, respectively. For the influence of cooking, indoor CO2, NO2, and PM2.5 levels during the cooking periods were significantly higher than that of the reference periods with increased concentrations of 26.17 ppm, 5.40 ppb, and 5 μg/m3, respectively.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-3417/9/20/4306/s1, Table S1: Descriptive statistics of 24-h average concentration of indoor air pollutants in the houses of traffic, industry, and general areas.

**Author Contributions:** Conceptualization, C.-Y.Y.; formal analysis, Y.-C.Y.; writing—original draft preparation, Y.-T.C. and Y.-T.C.; writing—review and editing, P.-S.C. and K.D.M.

**Funding:** This research was funded by "Wang Jhan-Yang Public Trust Fund", grant number 108-002-6.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
