- Email: [email protected]
Nonlinear prediction model of noise reduction by greenbelts
Urban Forestry & Urban Greening 14 (2015) 282–285
Contents lists available at ScienceDirect
Urban Forestry & Urban Greening journal homepage: www.elsevier.com/locate/ufug
Nonlinear prediction model of noise reduction by greenbelts Jing Zhang a , Xiaoping Guo a,∗ , Chuangang Zhao b a Key Laboratory for Soil and Water Conservation Desertification Combating of Ministry of Education, College of Soil and Water Conservation, Beijing Forestry University, Beijing, China b College of Information Science and Technology, Beijing Forestry University, Beijing, China
a r t i c l e
i n f o
Keywords: Model of noise reduction Populus tomentosa greenbelt Traffic noise
a b s t r a c t Based on the noise-reduction mechanism of greenbelts, a noise-reduction model of greenbelts is proposed along with their sound-propagation characteristics. Two Populus tomentosa greenbelts suitable for noise experiments were selected. A noise-reduction prediction model was established for P. tomentosa. The parameters for this model were the distances between the measuring point and the sound source and the frequencies of the sound source. The model error was 4.1 dB, approximately 9.8% of the average noise attenuation. When the noise frequency was 500 Hz (frequency equivalent to the road traffic noise), the average error was minimized. The experimental method and the model may be extended to noise reduction studies for other greenbelts and used to provide guidelines for road greenbelt construction. © 2015 Elsevier GmbH. All rights reserved.
Introduction Prevention and control of traffic noise primarily employs three approaches (Ding et al., 2004; Li and Tao, 2002): prevention and control of the sound source, cutting the route of transmission and protection at the point of sound reception. For cutting the transmission route, the principal methods are building sound barriers and planting greenbelts. However, sound barriers are expensive and most of them have monotonous appearances. A greenbelt can reduce noise while providing landscaping. So in the case of general noise reduction target, reducing noise by greenbelts is an effective and practical method (Ozer et al., 2008; Islam et al., 2012). The noise reduction effect of greenbelts is mainly related to noise source frequencies and the structural characteristics of greenbelts. For these reasons, the development of a greenbelt noise-reduction model that may be used to guide greenbelt construction for traffic-noise prevention and control is of great significance. Existing noisereduction prediction models may be divided into univariate and multivariate models (Xie, 2003; JTGB03-2006 China, 2006; Du et al., 2007; Yuan et al., 2009; Yuan, 2009; Zhang, 2009; Zhang et al., 2009; Guo et al., 2009). However, according to research methods, due to background conditions and the complexity of noise attenuation by the greenbelt, greenbelts of different structures have different effects on acoustic waves. The sound transmission losses under
∗ Corresponding author. Tel.: +86 13889886097. E-mail addresses: [email protected] (J. Zhang), [email protected] (X. Guo), [email protected] (C. Zhao). http://dx.doi.org/10.1016/j.ufug.2015.01.007 1618-8667/© 2015 Elsevier GmbH. All rights reserved.
the canopy are associated with ground impedance (Arnold and Michelle, 2009; Arnold, 2010). With the aim of establishing a more precise and practical greenbelt noise-reduction model, separate models should be developed for greenbelts composed of different plants. In this study, the common greenbelt species Populus tomentosa was selected as the research subject. Noise attenuation of various noise frequencies at several measuring points was measured and a noise-reduction prediction model was developed for P. tomentosa greenbelts. Measured data were used for validation with the aim of providing a method and basis for later designing of noise-reduction greenbelts and a guide for the construction of road greenbelts. Materials and methods Selection of sound source and sample area Given the relationship between the noise reduction effect of greenbelts and the noise frequency (Margaret et al., 1988; Tarrero Femandez and Gonzalez, 2002; Yuan, 2008), seven different frequencies of pure noise were selected as experimental sound sources: 125, 250, 500, 1000, 2000, 4000 and 8000 Hz. Pure noise as octave-band centre frequency noise, was generated by the Adobe Audition software (Adobe Audition 1.5, Adobe Audition(C)1992·2004 Adobe System, Incorporated), which is often used in the acoustic field. The selected noise frequencies included the main frequencies of traffic noise (Guo et al., 2009). The experimental sound source was a point sound source transmitted through loudspeakers as pure noise of different frequencies. The
J. Zhang et al. / Urban Forestry & Urban Greening 14 (2015) 282–285
283
Table 1 The characteristics of the sample plots. Sample number
Area of per plan (m2 /per plan)
Average tree height (m)
Average height under branch (m)
Average breast diameter (cm)
Length (m)
Width (m)
Green belt location
1 2
24.5 16.1
13.7 15.0
4.5 3.9
19.0 19.9
513 180
40 80
Dong Xiaokou forestry park Bajia country park
Fig. 1. Measuring points layout plan of Populus tomentosa green belt in Bajia country park.
loudspeaker frequency response range was 50 Hz–10 kHz, with a distortion rate lower than 5%. Following investigation of a large number of greenbelts in Beijing, we chose as test samples two P. tomentosa greenbelts with similar planting specifications, moderate scale and low environment noise. The samples are described in Table 1. Experimental methods The sound sources were placed in front of the greenbelts, and the measuring points in the greenbelts were at different distances from the sound source (measuring point layouts and cross-sectional views are shown in Figs. 1–3). At every measuring point, the noise attenuation values at different frequencies were measured with a sound level meter (AWA6218B+ noise statistical
analyzer, Hangzhou Aihua Instrument Co. Ltd.; range, 35–130 dB; measurement frequency, 31.5–8000 Hz; each successive measurement time, 8 s; sampling interval of analyzer, 0.01 s). The sound pressure level at the measuring point was the equivalent consecutive sound level A of the analyzer statistical output value. At each measuring point, the sound pressure level values at each frequency were measured 10 times consecutively and averaged. Finally, 259 groups of noise attenuation data were measured for the two greenbelts and the relationship between the frequency of the sound source, the distance between the sound source and the measurement point and the noise attenuation by the greenbelt was calculated. This procedure resulted in a noise-reduction prediction model using P. tomentosa greenbelts, verified by measured data. All measured values were measured in summer weather with no wind.
Fig. 2. Measuring points layout plan of Populus tomentosa green belt in Dong Xiaokou forestry park.
284
J. Zhang et al. / Urban Forestry & Urban Greening 14 (2015) 282–285
Table 2 The prediction values and measured values of prediction model of noise reduction by Populus tomentosa green belt. Measuring points
125 250 500 1000 2000 4000 8000
1
2
3
4
5
6
7
8
9
10
11
12
13
−9.41 −8.49 −3.08 −3.63 −5.65 −1.38 −6.46
−5.12 −4.64 −7.13 −7.83 0.79 −2.86 −8.27
−3.47 −5.45 −2.63 −1.78 3.64 −2.19 1.13
−6.62 −5.12 −1.88 −3.28 0.05 −7.54 −5.22
−4.45 −4.52 −4.51 −6.58 2.28 −4.23 −3.9
0.52 −2.69 −3.31 −7.24 0.78 −4.09 −3.84
−2.46 0.15 −3.52 8.26 4.58 7.41 −1.64
−2.04 0.61 −1.59 2.4 −0.99 4.66 1.78
−0.38 0.2 −0.22 3.17 6.2 4.2 3.86
0.85 2.42 1.31 7.39 6.76 −1.06 2.81
2.57 1.03 0.77 4.54 1.66 −2.34 4.86
1.44 −0.14 −0.69 3.93 −0.54 −2.75 3.32
7.06 2.3 −1 1.2 1.44 −4.33 2.04
250
500
The arrangement of measuring points was as follows: the sound source was located in front of the greenbelt and 3 m from its edge at a height of 1.2 m (GB3096-2008 China, 2008). B0 was located in the greenbelt boundary and the line from B0 to the sound source was perpendicular to the greenbelt edge. This line was extended to the greenbelt, and the other measuring points were placed on the basis of this line. Points C0 , D0 and E0 were placed on the extension line, and three straight lines parallel with the greenbelt edge were derived from these three points. The other measuring points were all placed on the three lines. Points C01 , C02 ,. . .C0P were placed to the right of C0 , and points C10 , C20 ,. . .CP0 were place to the left of C0 . The measuring points on both sides of D0 and E0 have the same arrangement with above. The method for determination of E0 , C0P and CP0 was as follows: there were two points on the straight lines derived from C0 on both sides and another point at the extension line from the sound source and B0 when the sound source was similar to the actual traffic noise sound level (Ye et al., 2009). At these three points, noise was reduced to the noise limit value of city zone five, a value which is clearly defined in the ‘Environmental quality standard for noise’; a minimum value of 40 dB was used in this study. The three measuring points were E0 , C0P and CP0 and the method for determining D0P , DPO , E0P and EPO was the same as that for C0P and CP0 . The other measuring points were randomly placed at heights of 1.2 m. Two-thirds of the test data were randomly selected for the development of a noise-reduction prediction model by P. tomentosa greenbelts and the remaining one-third of the data was used to validate the model. The analysis yielded the relationship between distance, noise frequency and noise reduction by the greenbelt. The statistical analysis software is SPSS PASW Statistics 18.0. The validation data thus obtained were used to test the model. The noise-attenuation values were calculated and the measured values were compared with the predicted values. The deviation of the predicted values from the observed values (error) is expressed as
n
i=1
(xi − yi )2 /n , where xi is the predicted value of the ith sam-
ple, yi is the measured value of the ith sample and n is the total number of samples.
5.5 5
The average value of root-mean-square error
Frequency (Hz)
4.5 4 3.5 3 2.5 125
1000
2000
4000
8000
Noise frequency /Hz Fig. 4. The average value of root-mean-square error of the noise reduction model’s prediction and measured values for certain frequencies.
Results The noise-reduction prediction model using greenbelts that is presented here consists of both distance attenuation and additional attenuation by greenbelts. The following formula describes the final noise-reduction prediction model using P. tomentosa greenbelts: L = 20lg
r n
r1
+ 3.08 × f 0.469 × (rn − r1 ) 0.367 × 0.955(rn −r1 )
× 0.88f (1/3)
(1)
where L is the attenuation value at noise measuring points (dB); rn is the distance between measuring point and sound source (m); r1 is the distance between sound source and the nearest greenbelt edge (m). In the formula, 20 lg rn /r1 is the distance attenuation
of the greenbelt and 3.08 × f 0.469 × (rn − r1 ) 0.367 × 0.955(rn −r1 ) × 0.88f (1/3) is the additional attenuation by the greenbelt. Comparison of the predicted values with the measured values yielded an error of 4.1 dB, approximately 9.8% of the average value of noise attenuation. Table 2 lists the relevant data for model validation, including the different frequencies, and the differences of the predicted values subtracting measured values at the different measuring points. Fig. 4 shows the average error of the predicted and measured values. When the frequency was 500 Hz, the average error of the predicted and measured values was minimum. Thus the accuracy of the model is highest when the traffic noise frequency is 500 Hz. Discussions
Fig. 3. Measuring points cross-sectional view of the Populus tomentosa green belt.
Acoustic wave propagation and attenuation in greenbelts is very complex. It is related to noise source frequencies and the structural characteristics of greenbelts. Acoustic wave reflection, diffraction and scattering in greenbelts are difficult to quantify directly. In this experiment, the complex process of noise reduction by greenbelts is fully taken into account, yielding a nonlinear noise-reduction prediction model using P. tomentosa greenbelts which is more precise.
J. Zhang et al. / Urban Forestry & Urban Greening 14 (2015) 282–285
In an in-depth study and analysis of the noise-reduction mechanism of greenbelts and a noise-reduction model of greenbelts along with their sound propagation characteristics, noise reduction showed a strong relationship with distance and noise frequency. Noise reduction by greenbelts included distance attenuation and additional attenuation. Distance attenuation refers to sound-energy loss caused by sound wave spherical detachment and air friction during motion. Additional attenuation by greenbelts refers to the presence of obstacles between the sound source and receiver, in which sound attenuation is caused by reflection, diffraction, scattering, absorption and other phenomena (Xie, 2003; Tarrero Femandez and Gonzalez, 2002; Herrington, 1976; Harris, 1979; Beranek, 1971).In this model, noise attenuation by greenbelts is composed of distance attenuation and additional attenuation by greenbelts. The number of measuring points in this experiment was greater than those used in previous studies, resulting in large amounts of measurement data. The model established with the measured data reflects the relationship between noise reduction by greenbelts, noise frequency and sound source distance from measuring points. Experiments on noise reduction by greenbelts have high requirements for the test environment. First, the test area should be as quiet as possible to avoid the influence of environmental noise on the tests. Second, the greenbelts should be regular, so that it is convenient to arrange measuring points. However, such greenbelts are difficult to find. A field survey of the green belts in Beijing by the author of this article yielded two P. tomentosa greenbelts. Because different structural characteristics of greenbelts (including planting density, tree height and DBH) have different effects on noise reduction (Margaret et al., 1988; Wang et al., 2012; Rosa et al., 2006), developing the most accurate noise-reduction prediction model using greenbelts requires selection of greenbelts with different structural features. Accordingly, in future research, a large number of tests of different compositions and structural characteristics of greenbelts should be performed, noise-reduction prediction model using different greenbelts should be developed and the relationships between these models should be investigated. A database should be established, and the error range of the models should be calculated. The model error range is taken into account in guiding the construction of road greenbelts. Conclusions A noise-reduction prediction model was established for P. tomentosa, which was composed of both distance attenuation and additional attenuation by greenbelts. The parameters for this model were the distances between the measuring point and the sound source and the frequencies of the sound source. The error analysis is carried out on the model. The model error was 4.1 dB, approximately 9.8% of the average noise attenuation. When the noise frequency was 500 Hz, the average error was minimized. Thus the accuracy of the model is highest when the traffic noise frequency is 500 Hz. Given that the equivalent frequency of road traffic noise is 500 Hz (Yu, 2008; HJ/T90-2004 China, 2004), the model has good applicability. The model not only provides guidance for the construction of P. tomentosa greenbelts on both sides of roads but can also help in the further study of noise-reduction prediction models using greenbelts.
285
Acoustic wave propagation and attenuation in greenbelts is very complex, and in the experiment the tree species were single, the sample areas were less, so the model cannot be used to predict the noise reduction effect of other greenbelts. Accordingly, in future research, a large number of tests of different compositions and structural characteristics of greenbelts should be performed, noise-reduction prediction model using different greenbelts should be developed. A database should be established, and the error range of the models should be calculated. The model error range is taken into account in guiding the construction of road greenbelts. References Arnold, T., Michelle, E.S., 2009. Numerical model to calculate microphysical influences on sound wave propagation in forests. Appl. Acoust. 70 (6), 857–867. Arnold, T., 2010. Numerical experiment to revisit micrometeorology and sound speed calculation in forests. Meteorol. Atmos. Phys. 107 (3/4), 103–108. Beranek, L.L., 1971. Noise and Vibration Control[M]. McGraw-Hill, New York, NY. Ding, Y.C., Zhou, J.X., Li, H., et al., 2004. Investigation of traffic noise attenuation provided by tree belts. Chin. J. Highw. Trans. 12, 204–208. Du, Z.Y., Xing, S.J., Song, Y.M., et al., 2007. Study on traffic noise attenuation by greenbelts along expressway. Int. J. Ecol. Environ. Sci. 16 (1), 31–35. GB3096-2008 China, 2008. Environmental quality standard for noise. In: GB30962008. Guo, X.P., Peng, H.Y., Wang, L., 2009. The effects of traffic noise attenuation by greenbelts. J. Environ. Sci. 29 (12), 2567–2571. Harris, C.M., 1979. Handbook of Noise Control, second ed. McGraw-Hill, New York, NY. Herrington, L.P., 1976. Effect of vegetation on the propagation of noise in the outof-doors. In: USDA Forest Service General Technical Report, US Rocky Mountain Forest Range Experimental Station No. 25., pp. 229–233. HJ/T90-2004 China, 2004. Norm on acoustical design and measurement of noise barriers. In: HJ/T90-2004. Islam, M.N., Rahman, K.S., Bahar, M.M., 2012. Pollution attenuation by roadside greenbelt in and around urban areas. Urban Urban Green 11 (4), 460–464. JTGB03-2006 China, 2006. Specifications for environment impact assessment of highway. In: JTGB03-2006. Li, B.G., Tao, S., 2002. Progress in road traffic noise prediction model study. Res. Environ. Sci. 15 (2), 56–59. Margaret, A.P., Keith, A., Nicholas, W.H., 1988. Sound attenuation through trees: measurements and models. J Acoust. Soc. Am. 84 (5), 1836–1844. Ozer, S., Irmak, M.A., Yilmaz, H., 2008. Determination of roadside noise reduction effectiveness of Pinus sylvestris L. and Populus nigra L. in Erzurum, Turkey. Environ. Monit. Assess. 144 (1/2/3), 191–197. Rosa, M.S., Constanza, R., Luis, M.G.R., Juan, V.S.P., Enrique, A.S.P., Llinares, J., 2006. Control of noise by trees arranged like sonic crystals. J. Sound Vibr. 291 (1–2), 100–106. Tarrero Femandez, A.I., Gonzalez, J., 2002. Validation of an outdoor sound propagation model in presence of trees. In: Forum Acusticum Sevilla. Wang, H., Guo, J.P., Zhang, Y.X., et al., 2012. Study on noise attenuation impact of roadside tree-belts and related influence factors. Int. J. Ecol. Environ. Sci. 19 (6), 1403–1408. Xie, B.L., 2003. The influence of greenbelt on traffic noise. Shanxi Sci. Technol. 4, 31–32. Ye, H.H., Wei, X.W., Wang, Y.Q., et al., 2009. Study on characteristic of urban traffic noise. Noise Vibr. Control S2, 367–370. Yu, W.Z., 2008. Design and noise reduction of elevated road barrier of microperforated panels with linear-change cavity. Environ. Pollut. Control 30 (7), 55–61. Yuan, L., Wang, X.C., Lu, Y.Y., et al., 2009. Calculated method of noise attenuation of tree belts at expressway. Chin. J. Highw. Trans. 22 (3), 107–112. Yuan, L., 2008. Influence of plant tissue structure on the spectrum characteristics of traffic noise attenuation. Noise Vibr. Control 28 (5), 154–167. Yuan, X.X., 2009. Highway of traffic noise attenuation by green belts effect analysis. Highw. Automot. Appl. 2, 114–116. Zhang, W.Q., Liu, J.W., Hu, H.Y., et al., 2009. Effect of the construction of green belts on the attenuation of traffic noise along the urban trunk roads in Xiamen City. Subtrop. Plant Sci. 38 (4), 74–78. Zhang, H.K., 2009. Study on the parameter design of highway greenbelts noise reduction. Highw. Automot. Appl. 4, 156–159.
Contents lists available at ScienceDirect
Urban Forestry & Urban Greening journal homepage: www.elsevier.com/locate/ufug
Nonlinear prediction model of noise reduction by greenbelts Jing Zhang a , Xiaoping Guo a,∗ , Chuangang Zhao b a Key Laboratory for Soil and Water Conservation Desertification Combating of Ministry of Education, College of Soil and Water Conservation, Beijing Forestry University, Beijing, China b College of Information Science and Technology, Beijing Forestry University, Beijing, China
a r t i c l e
i n f o
Keywords: Model of noise reduction Populus tomentosa greenbelt Traffic noise
a b s t r a c t Based on the noise-reduction mechanism of greenbelts, a noise-reduction model of greenbelts is proposed along with their sound-propagation characteristics. Two Populus tomentosa greenbelts suitable for noise experiments were selected. A noise-reduction prediction model was established for P. tomentosa. The parameters for this model were the distances between the measuring point and the sound source and the frequencies of the sound source. The model error was 4.1 dB, approximately 9.8% of the average noise attenuation. When the noise frequency was 500 Hz (frequency equivalent to the road traffic noise), the average error was minimized. The experimental method and the model may be extended to noise reduction studies for other greenbelts and used to provide guidelines for road greenbelt construction. © 2015 Elsevier GmbH. All rights reserved.
Introduction Prevention and control of traffic noise primarily employs three approaches (Ding et al., 2004; Li and Tao, 2002): prevention and control of the sound source, cutting the route of transmission and protection at the point of sound reception. For cutting the transmission route, the principal methods are building sound barriers and planting greenbelts. However, sound barriers are expensive and most of them have monotonous appearances. A greenbelt can reduce noise while providing landscaping. So in the case of general noise reduction target, reducing noise by greenbelts is an effective and practical method (Ozer et al., 2008; Islam et al., 2012). The noise reduction effect of greenbelts is mainly related to noise source frequencies and the structural characteristics of greenbelts. For these reasons, the development of a greenbelt noise-reduction model that may be used to guide greenbelt construction for traffic-noise prevention and control is of great significance. Existing noisereduction prediction models may be divided into univariate and multivariate models (Xie, 2003; JTGB03-2006 China, 2006; Du et al., 2007; Yuan et al., 2009; Yuan, 2009; Zhang, 2009; Zhang et al., 2009; Guo et al., 2009). However, according to research methods, due to background conditions and the complexity of noise attenuation by the greenbelt, greenbelts of different structures have different effects on acoustic waves. The sound transmission losses under
∗ Corresponding author. Tel.: +86 13889886097. E-mail addresses: [email protected] (J. Zhang), [email protected] (X. Guo), [email protected] (C. Zhao). http://dx.doi.org/10.1016/j.ufug.2015.01.007 1618-8667/© 2015 Elsevier GmbH. All rights reserved.
the canopy are associated with ground impedance (Arnold and Michelle, 2009; Arnold, 2010). With the aim of establishing a more precise and practical greenbelt noise-reduction model, separate models should be developed for greenbelts composed of different plants. In this study, the common greenbelt species Populus tomentosa was selected as the research subject. Noise attenuation of various noise frequencies at several measuring points was measured and a noise-reduction prediction model was developed for P. tomentosa greenbelts. Measured data were used for validation with the aim of providing a method and basis for later designing of noise-reduction greenbelts and a guide for the construction of road greenbelts. Materials and methods Selection of sound source and sample area Given the relationship between the noise reduction effect of greenbelts and the noise frequency (Margaret et al., 1988; Tarrero Femandez and Gonzalez, 2002; Yuan, 2008), seven different frequencies of pure noise were selected as experimental sound sources: 125, 250, 500, 1000, 2000, 4000 and 8000 Hz. Pure noise as octave-band centre frequency noise, was generated by the Adobe Audition software (Adobe Audition 1.5, Adobe Audition(C)1992·2004 Adobe System, Incorporated), which is often used in the acoustic field. The selected noise frequencies included the main frequencies of traffic noise (Guo et al., 2009). The experimental sound source was a point sound source transmitted through loudspeakers as pure noise of different frequencies. The
J. Zhang et al. / Urban Forestry & Urban Greening 14 (2015) 282–285
283
Table 1 The characteristics of the sample plots. Sample number
Area of per plan (m2 /per plan)
Average tree height (m)
Average height under branch (m)
Average breast diameter (cm)
Length (m)
Width (m)
Green belt location
1 2
24.5 16.1
13.7 15.0
4.5 3.9
19.0 19.9
513 180
40 80
Dong Xiaokou forestry park Bajia country park
Fig. 1. Measuring points layout plan of Populus tomentosa green belt in Bajia country park.
loudspeaker frequency response range was 50 Hz–10 kHz, with a distortion rate lower than 5%. Following investigation of a large number of greenbelts in Beijing, we chose as test samples two P. tomentosa greenbelts with similar planting specifications, moderate scale and low environment noise. The samples are described in Table 1. Experimental methods The sound sources were placed in front of the greenbelts, and the measuring points in the greenbelts were at different distances from the sound source (measuring point layouts and cross-sectional views are shown in Figs. 1–3). At every measuring point, the noise attenuation values at different frequencies were measured with a sound level meter (AWA6218B+ noise statistical
analyzer, Hangzhou Aihua Instrument Co. Ltd.; range, 35–130 dB; measurement frequency, 31.5–8000 Hz; each successive measurement time, 8 s; sampling interval of analyzer, 0.01 s). The sound pressure level at the measuring point was the equivalent consecutive sound level A of the analyzer statistical output value. At each measuring point, the sound pressure level values at each frequency were measured 10 times consecutively and averaged. Finally, 259 groups of noise attenuation data were measured for the two greenbelts and the relationship between the frequency of the sound source, the distance between the sound source and the measurement point and the noise attenuation by the greenbelt was calculated. This procedure resulted in a noise-reduction prediction model using P. tomentosa greenbelts, verified by measured data. All measured values were measured in summer weather with no wind.
Fig. 2. Measuring points layout plan of Populus tomentosa green belt in Dong Xiaokou forestry park.
284
J. Zhang et al. / Urban Forestry & Urban Greening 14 (2015) 282–285
Table 2 The prediction values and measured values of prediction model of noise reduction by Populus tomentosa green belt. Measuring points
125 250 500 1000 2000 4000 8000
1
2
3
4
5
6
7
8
9
10
11
12
13
−9.41 −8.49 −3.08 −3.63 −5.65 −1.38 −6.46
−5.12 −4.64 −7.13 −7.83 0.79 −2.86 −8.27
−3.47 −5.45 −2.63 −1.78 3.64 −2.19 1.13
−6.62 −5.12 −1.88 −3.28 0.05 −7.54 −5.22
−4.45 −4.52 −4.51 −6.58 2.28 −4.23 −3.9
0.52 −2.69 −3.31 −7.24 0.78 −4.09 −3.84
−2.46 0.15 −3.52 8.26 4.58 7.41 −1.64
−2.04 0.61 −1.59 2.4 −0.99 4.66 1.78
−0.38 0.2 −0.22 3.17 6.2 4.2 3.86
0.85 2.42 1.31 7.39 6.76 −1.06 2.81
2.57 1.03 0.77 4.54 1.66 −2.34 4.86
1.44 −0.14 −0.69 3.93 −0.54 −2.75 3.32
7.06 2.3 −1 1.2 1.44 −4.33 2.04
250
500
The arrangement of measuring points was as follows: the sound source was located in front of the greenbelt and 3 m from its edge at a height of 1.2 m (GB3096-2008 China, 2008). B0 was located in the greenbelt boundary and the line from B0 to the sound source was perpendicular to the greenbelt edge. This line was extended to the greenbelt, and the other measuring points were placed on the basis of this line. Points C0 , D0 and E0 were placed on the extension line, and three straight lines parallel with the greenbelt edge were derived from these three points. The other measuring points were all placed on the three lines. Points C01 , C02 ,. . .C0P were placed to the right of C0 , and points C10 , C20 ,. . .CP0 were place to the left of C0 . The measuring points on both sides of D0 and E0 have the same arrangement with above. The method for determination of E0 , C0P and CP0 was as follows: there were two points on the straight lines derived from C0 on both sides and another point at the extension line from the sound source and B0 when the sound source was similar to the actual traffic noise sound level (Ye et al., 2009). At these three points, noise was reduced to the noise limit value of city zone five, a value which is clearly defined in the ‘Environmental quality standard for noise’; a minimum value of 40 dB was used in this study. The three measuring points were E0 , C0P and CP0 and the method for determining D0P , DPO , E0P and EPO was the same as that for C0P and CP0 . The other measuring points were randomly placed at heights of 1.2 m. Two-thirds of the test data were randomly selected for the development of a noise-reduction prediction model by P. tomentosa greenbelts and the remaining one-third of the data was used to validate the model. The analysis yielded the relationship between distance, noise frequency and noise reduction by the greenbelt. The statistical analysis software is SPSS PASW Statistics 18.0. The validation data thus obtained were used to test the model. The noise-attenuation values were calculated and the measured values were compared with the predicted values. The deviation of the predicted values from the observed values (error) is expressed as
n
i=1
(xi − yi )2 /n , where xi is the predicted value of the ith sam-
ple, yi is the measured value of the ith sample and n is the total number of samples.
5.5 5
The average value of root-mean-square error
Frequency (Hz)
4.5 4 3.5 3 2.5 125
1000
2000
4000
8000
Noise frequency /Hz Fig. 4. The average value of root-mean-square error of the noise reduction model’s prediction and measured values for certain frequencies.
Results The noise-reduction prediction model using greenbelts that is presented here consists of both distance attenuation and additional attenuation by greenbelts. The following formula describes the final noise-reduction prediction model using P. tomentosa greenbelts: L = 20lg
r n
r1
+ 3.08 × f 0.469 × (rn − r1 ) 0.367 × 0.955(rn −r1 )
× 0.88f (1/3)
(1)
where L is the attenuation value at noise measuring points (dB); rn is the distance between measuring point and sound source (m); r1 is the distance between sound source and the nearest greenbelt edge (m). In the formula, 20 lg rn /r1 is the distance attenuation
of the greenbelt and 3.08 × f 0.469 × (rn − r1 ) 0.367 × 0.955(rn −r1 ) × 0.88f (1/3) is the additional attenuation by the greenbelt. Comparison of the predicted values with the measured values yielded an error of 4.1 dB, approximately 9.8% of the average value of noise attenuation. Table 2 lists the relevant data for model validation, including the different frequencies, and the differences of the predicted values subtracting measured values at the different measuring points. Fig. 4 shows the average error of the predicted and measured values. When the frequency was 500 Hz, the average error of the predicted and measured values was minimum. Thus the accuracy of the model is highest when the traffic noise frequency is 500 Hz. Discussions
Fig. 3. Measuring points cross-sectional view of the Populus tomentosa green belt.
Acoustic wave propagation and attenuation in greenbelts is very complex. It is related to noise source frequencies and the structural characteristics of greenbelts. Acoustic wave reflection, diffraction and scattering in greenbelts are difficult to quantify directly. In this experiment, the complex process of noise reduction by greenbelts is fully taken into account, yielding a nonlinear noise-reduction prediction model using P. tomentosa greenbelts which is more precise.
J. Zhang et al. / Urban Forestry & Urban Greening 14 (2015) 282–285
In an in-depth study and analysis of the noise-reduction mechanism of greenbelts and a noise-reduction model of greenbelts along with their sound propagation characteristics, noise reduction showed a strong relationship with distance and noise frequency. Noise reduction by greenbelts included distance attenuation and additional attenuation. Distance attenuation refers to sound-energy loss caused by sound wave spherical detachment and air friction during motion. Additional attenuation by greenbelts refers to the presence of obstacles between the sound source and receiver, in which sound attenuation is caused by reflection, diffraction, scattering, absorption and other phenomena (Xie, 2003; Tarrero Femandez and Gonzalez, 2002; Herrington, 1976; Harris, 1979; Beranek, 1971).In this model, noise attenuation by greenbelts is composed of distance attenuation and additional attenuation by greenbelts. The number of measuring points in this experiment was greater than those used in previous studies, resulting in large amounts of measurement data. The model established with the measured data reflects the relationship between noise reduction by greenbelts, noise frequency and sound source distance from measuring points. Experiments on noise reduction by greenbelts have high requirements for the test environment. First, the test area should be as quiet as possible to avoid the influence of environmental noise on the tests. Second, the greenbelts should be regular, so that it is convenient to arrange measuring points. However, such greenbelts are difficult to find. A field survey of the green belts in Beijing by the author of this article yielded two P. tomentosa greenbelts. Because different structural characteristics of greenbelts (including planting density, tree height and DBH) have different effects on noise reduction (Margaret et al., 1988; Wang et al., 2012; Rosa et al., 2006), developing the most accurate noise-reduction prediction model using greenbelts requires selection of greenbelts with different structural features. Accordingly, in future research, a large number of tests of different compositions and structural characteristics of greenbelts should be performed, noise-reduction prediction model using different greenbelts should be developed and the relationships between these models should be investigated. A database should be established, and the error range of the models should be calculated. The model error range is taken into account in guiding the construction of road greenbelts. Conclusions A noise-reduction prediction model was established for P. tomentosa, which was composed of both distance attenuation and additional attenuation by greenbelts. The parameters for this model were the distances between the measuring point and the sound source and the frequencies of the sound source. The error analysis is carried out on the model. The model error was 4.1 dB, approximately 9.8% of the average noise attenuation. When the noise frequency was 500 Hz, the average error was minimized. Thus the accuracy of the model is highest when the traffic noise frequency is 500 Hz. Given that the equivalent frequency of road traffic noise is 500 Hz (Yu, 2008; HJ/T90-2004 China, 2004), the model has good applicability. The model not only provides guidance for the construction of P. tomentosa greenbelts on both sides of roads but can also help in the further study of noise-reduction prediction models using greenbelts.
285
Acoustic wave propagation and attenuation in greenbelts is very complex, and in the experiment the tree species were single, the sample areas were less, so the model cannot be used to predict the noise reduction effect of other greenbelts. Accordingly, in future research, a large number of tests of different compositions and structural characteristics of greenbelts should be performed, noise-reduction prediction model using different greenbelts should be developed. A database should be established, and the error range of the models should be calculated. The model error range is taken into account in guiding the construction of road greenbelts. References Arnold, T., Michelle, E.S., 2009. Numerical model to calculate microphysical influences on sound wave propagation in forests. Appl. Acoust. 70 (6), 857–867. Arnold, T., 2010. Numerical experiment to revisit micrometeorology and sound speed calculation in forests. Meteorol. Atmos. Phys. 107 (3/4), 103–108. Beranek, L.L., 1971. Noise and Vibration Control[M]. McGraw-Hill, New York, NY. Ding, Y.C., Zhou, J.X., Li, H., et al., 2004. Investigation of traffic noise attenuation provided by tree belts. Chin. J. Highw. Trans. 12, 204–208. Du, Z.Y., Xing, S.J., Song, Y.M., et al., 2007. Study on traffic noise attenuation by greenbelts along expressway. Int. J. Ecol. Environ. Sci. 16 (1), 31–35. GB3096-2008 China, 2008. Environmental quality standard for noise. In: GB30962008. Guo, X.P., Peng, H.Y., Wang, L., 2009. The effects of traffic noise attenuation by greenbelts. J. Environ. Sci. 29 (12), 2567–2571. Harris, C.M., 1979. Handbook of Noise Control, second ed. McGraw-Hill, New York, NY. Herrington, L.P., 1976. Effect of vegetation on the propagation of noise in the outof-doors. In: USDA Forest Service General Technical Report, US Rocky Mountain Forest Range Experimental Station No. 25., pp. 229–233. HJ/T90-2004 China, 2004. Norm on acoustical design and measurement of noise barriers. In: HJ/T90-2004. Islam, M.N., Rahman, K.S., Bahar, M.M., 2012. Pollution attenuation by roadside greenbelt in and around urban areas. Urban Urban Green 11 (4), 460–464. JTGB03-2006 China, 2006. Specifications for environment impact assessment of highway. In: JTGB03-2006. Li, B.G., Tao, S., 2002. Progress in road traffic noise prediction model study. Res. Environ. Sci. 15 (2), 56–59. Margaret, A.P., Keith, A., Nicholas, W.H., 1988. Sound attenuation through trees: measurements and models. J Acoust. Soc. Am. 84 (5), 1836–1844. Ozer, S., Irmak, M.A., Yilmaz, H., 2008. Determination of roadside noise reduction effectiveness of Pinus sylvestris L. and Populus nigra L. in Erzurum, Turkey. Environ. Monit. Assess. 144 (1/2/3), 191–197. Rosa, M.S., Constanza, R., Luis, M.G.R., Juan, V.S.P., Enrique, A.S.P., Llinares, J., 2006. Control of noise by trees arranged like sonic crystals. J. Sound Vibr. 291 (1–2), 100–106. Tarrero Femandez, A.I., Gonzalez, J., 2002. Validation of an outdoor sound propagation model in presence of trees. In: Forum Acusticum Sevilla. Wang, H., Guo, J.P., Zhang, Y.X., et al., 2012. Study on noise attenuation impact of roadside tree-belts and related influence factors. Int. J. Ecol. Environ. Sci. 19 (6), 1403–1408. Xie, B.L., 2003. The influence of greenbelt on traffic noise. Shanxi Sci. Technol. 4, 31–32. Ye, H.H., Wei, X.W., Wang, Y.Q., et al., 2009. Study on characteristic of urban traffic noise. Noise Vibr. Control S2, 367–370. Yu, W.Z., 2008. Design and noise reduction of elevated road barrier of microperforated panels with linear-change cavity. Environ. Pollut. Control 30 (7), 55–61. Yuan, L., Wang, X.C., Lu, Y.Y., et al., 2009. Calculated method of noise attenuation of tree belts at expressway. Chin. J. Highw. Trans. 22 (3), 107–112. Yuan, L., 2008. Influence of plant tissue structure on the spectrum characteristics of traffic noise attenuation. Noise Vibr. Control 28 (5), 154–167. Yuan, X.X., 2009. Highway of traffic noise attenuation by green belts effect analysis. Highw. Automot. Appl. 2, 114–116. Zhang, W.Q., Liu, J.W., Hu, H.Y., et al., 2009. Effect of the construction of green belts on the attenuation of traffic noise along the urban trunk roads in Xiamen City. Subtrop. Plant Sci. 38 (4), 74–78. Zhang, H.K., 2009. Study on the parameter design of highway greenbelts noise reduction. Highw. Automot. Appl. 4, 156–159.