### 1. Introduction

_{10}, PM

_{2.5}and PM

_{1}above radiators.

### 2. Materials and Methods

### 2.1. Study Area and Aampling Locations

### 2.2. Sampling Methods

^{2}, there were five sampling points in the adjacent indoor air, otherwise three sampling points. The sampling points in the adjacent indoor air were all 1.5 m above the ground and far away from furnishings. These sampling points were arranged in a line or the shape of a plum blossom.

### 2.3. Analytical Methods

*β*

_{1},

*β*

_{2}, ···

*β*

*represent the regression coefficients,*

_{n}*X*

_{1},

*X*

_{2}, ···

*X*

*are the independent variables and ɛ is the residual error. The approach used in this study was based on the assumption that the environmental parameters were linearly related to the particle mass concentrations above radiators.*

_{n}### 3. Results and Discussion

### 3.1. Particle Size Distributions above Radiators

^{7}/L and the values of dM/dlog (dp) are in the range of 0 to 3300 μg/m

^{3}. The particle numbers as a whole decrease with the particle size (see Fig. 4(a)), which exhibits a similar tendency to some studies on the indoor particle size distributions [38, 39]. The different characters of large and small particles are possibly sufficient to trigger this phenomenon. The gravitational settling is dominant for the deposition of larger particles [40] and the larger particles are difficult to be resuspended. The smaller particles can remain airborne for longer periods of time than larger particles since they are too small for the inertial deposition. However, the particle mass size distributions show different patterns. Larger particles ( > 10 μm) with a lower quantity occupy a great proportion of the particle mass above radiators, as shown in Fig. 4(b). One of the possible causes is that the enhancement of the airflow above radiators makes more large particles resuspension.

#### 3.1.1. Effect of room functions

#### 3.1.2. Effect of temperature, relative humidity and air velocity

### 3.2. Particle Mass Concentrations above Radiators and Effecting Factors

#### 3.2.1. Multiple linear regression models

_{10}above radiators in the forty-two investigated housing units vary from approximately 43.7 to 222.9 μg/m

^{3}. The concentrations of PM

_{2.5}and PM

_{1}above radiators are lower than those of PM

_{10}, which are 33.4–213.1 μg/m

^{3}and 21.3–201.6 μg/m

^{3}, respectively. The multiple linear regression technique was used as a tool for detecting quantitative relations between particle concentrations above radiators and related environmental parameters, assuming a linear relationship. The coefficient of determination (

*R*

^{2}), which was used to describe the strength of the association and the statistical significance of the results, was also evaluated. The multiple linear regression analysis results were reflected by the regression coefficients with their 95% confidence interval and the significance level was 0.05.

_{10}above radiators and the multiple linear regression model is also proposed. The adjusted coefficient of multiple determinations, which is 0.800, gives the proportion of the variation in the PM

_{10}concentrations above radiators explained by the environmental parameters. And that is, 80.0% of the variation in the dependent variable can be explained by the independent variables shown in Table 2. The relationships exhibited in this multiple linear regression equation declare that the concentrations of PM

_{10}in the adjacent indoor air have a significant impact on that above radiators. Also, the results suggest negative correlations between the concentrations of PM

_{10}above radiators and the temperatures. The measurement data shows that a negative correlation exists between the concentrations of PM

_{10}above radiators and the relative humidities. The constant coefficient and three regressions coefficients are all statistically highly significant (

*p*< 0.01)

_{2.5}above radiators are presented in Table 3. When the three variables (concentrations of PM

_{2.5}in the adjacent indoor air, temperatures and the relative humidity) are fitted to the concentration data, the value of the adjusted coefficient of determinations is 0.929. The multiple linear regression results indicate that the concentrations of PM

_{2.5}above radiators are strongly dependent on that in the adjacent indoor air and the β is 0.884. Besides, in this model, there is a significant correlation between the temperatures and the concentrations of PM

_{2.5}above radiators. The multiple linear regression analysis also presents that the concentrations of PM

_{2.5}above radiators are influenced by the relative humidities. The constant and coefficients of regressions are all significant at the 0.01 level (p = 0.000).

_{1}above radiators. For the concentration of PM

_{1}, the value of the adjusted R

^{2}is 0.962, which means that 96.2% of the variation in the concentrations of PM

_{1}above radiators is explained by these four independent variables. The measurement data reveals a good linear relationship between the concentrations of PM

_{1}above radiators and in the adjacent indoor air. The concentrations of PM

_{1}above radiators have a negative correlation with the temperature and the relative humidity. It is worth noting that the air velocities have a positive association with the concentrations of PM

_{1}above radiators (

*p*< 0.05).

#### 3.2.2. Effect of environmental parameters

_{1}above radiators (adjusted R

^{2}= 0.962) is the best. The concentrations of PM

_{10}, PM

_{2.5}and PM

_{1}above radiators are affected by particle concentrations in the adjacent indoor air. The data reported here suggest that the particle concentrations above radiators are closely tied to that in the adjacent indoor air. This result is expected and easy to understand. The zone above radiators connects with the adjacent indoor environment and the particles in these two spaces interact on each other. The concentrations of PM

_{10}, PM

_{2.5}and PM

_{1}above radiators all have negative correlations with the temperatures. The similar relationship between the particle concentrations and temperatures has also been reported by Sarwar et al. [41]. When there is a temperature gradient, small particles suspending in the air will create a fast flow away from the place with higher temperature (opposite to the temperature gradient), which leads to the negative correlation between the particle concentrations above radiators and the temperatures. In accordance with the findings of Fromme et al. [42], the particle concentrations increase with the decreasing relative humidity. One reason for this could be that the small particles prone to condense in the air with high relative humidity. The air velocities are not detected associations with concentrations of PM

_{10}and PM

_{2.5}above radiators. Instead, a weak positive correlation has been found between concentrations of the PM

_{1}above radiators and the air velocities. This most likely arises from the fact that small particles have low inertia and are influenced by airflow pattern more seriously. Besides, the larger air velocities will result in more resuspension of previously settled particles.

### 4. Conclusions

Larger particles ( > 10 μm) with a lower quantity occupy a great proportion of the particle mass above radiators. The functional difference among indoor environments has little effect on the particle size distributions above radiators. The number and mass of larger ( > 10 μm) and smaller ( < 1 μm) particles at locations in different indoor environments have relatively great difference.

The environmental parameters (particle concentrations in the adjacent indoor air, temperatures, relative humidities and air velocities) are related to the particle concentrations above radiators with various degrees. Three multiple linear regression models are proposed to predict the concentrations of PM

_{10}, PM_{2.5}and PM_{1}above radiators.