Mapping of noise pollution by different
interpolation methods in recovery section of Ghandi telecommunication Cables
Company
Taghizadeh-Mehrjardi R, PhD 1*, Zare
M, PhD 2, Zare S, MSc 3
1- Assistant Prof.,
Faculty of Agriculture and Natural Resources, University of Ardakan, Ardakan,
Iran. 2- PhD student, LUNAM Université, Universitéd’Angers,
Laboratoired’ergonomieetd’épidémiologie en santé au travail (LEEST), Angers,
France. 3- Expert in Dept of Electrical
Engineering, National Iranian Oil Company (NIOC), Yazd, Iran.
Abstract
Received:
September 2013,
Accepted: January 2013
Background: Noise
pollution and workers' noise exposure are common in industrial factories in
Iran. In order to reduce this noise pollution, evaluation and investigation
of noise emission are both necessary. In this study, different noise mapping
methodsare used for determining the distribution of noise. Materials and Methods: In the
present study, for preparing a noise map in a hall of an industrial factory,
sampling was carried out in 6×6 m grid. After data normalization the
variogram was developed. For interpolation of mentioned parameter, kriging
and Inverse Distance Weighting methods were used. The best model for
interpolation was selected by cross validation and error evaluation methods,
such as Route Mean Square Error(RMSE). Results: The
results showed that kriging method is better than other methods for
prediction of noise property. The noise map was prepared, using the best
interpolation method in Geographical Information System environment.
Conclusion: Workers in this industrial hall
were exposed to noise which is mainly induced by noisy machines. Noise maps which were produced in this study showed the distribution of noise and, also revealed that workers suffer from serious noise pollution. |
Key words: Noise Pollution, Interpolation, Industry.
Introduction
Noise pollution is a major problem
and a complaint among workers within industrial environments (1). It is
estimated that most of industries in Iran suffer from problems which are
induced by noise (2). Exposure to noise levels, especially higher than
threshold limits, will lead to hearing loss and other biological disorders such
as cardio vascular effects, sleep disorders and nervous complaints (3). Most of
workers within industries in Iran are suffering from noise exposure, and
Complaints regarding noise level from workers are high, showing that noise
problems are among the important problems in Iran.
In order* to manage
noise problems in industries, noise distribution must be evaluated and
illustrated in work places before proposing any noise control program. Noise
map is the best way to represent the sound level distribution existing in a given district.
In general, noise maps have two main purposes. First, they can be
used for proposing a noise action plans for management and reduction of noise
level. Secondly a noise map is to provide information on noise levels to
estimate how many people are
affected.
Two main stages can be distinguished in mapping process, 1) the sampling
stage, during which measurements are taken from environmental variable at
selected locations; and 2) the prediction stage, during which the observations
are interpolated to a fine grid. The quality of the resulting map is determined
by both stages. Geostatisticians have concentrated most on the second stage, by
applying different types of interpolation methods (4).
Geostatistical methods were developed to create mathematical models of
spatial correlation structures with a variogram as the quantitative measure of
spatial correlation. The variogram is commonly used in geostatistics and the interpolation
technique of kriging, providing the ‘‘best’’,
unbiased, linear estimate of a regionalized variable in alocation where
sampling isn't done, where ‘‘best’’ is defined in a least-squares sense.
The emphasis is set on local accuracy, i.e. closeness
of the estimate to the actual, but unknown value without any regard for
the global statistical properties of the estimates.
Kriging estimation variances are independent from the estimated values and are
related to the spatial arrangement of the sample data and variogrammodel (5).
The accuracy of different spatial interpolation methods for pollution
parameters mapping in soil, air and water have been evaluated by some research recently
(6). Kriging method was used to estimate heavy metals concentration in
groundwater and concluded that it is the best estimator for spatial prediction
of Lead (7). Tsai et al used geographic information systems (GIS) and concluded
that noise maps are a useful way of evaluating noise levels (8). The present
study aimed to evaluate accuracy of two interpolation methods (kriging and IDW
(Inverse Distance Weighting)) for prediction of noise values in Recovery section of Ghandi Telecommunication
Cable Co. Noise map of this department also was provided to be compared for the
noise level distribution with noise regulation standards.
Materials and Methods
This is a case study which was done in the recovery section of Ghandi
Telecommunication Cable Co. in summer 2011. This section has 66 meter length
and 66 meter wide with 7 meter height. The floor and walls were made of
concrete and break. 25 people worked in this department and the most important
noise resources were Prickly Machine, Caterpillar, Halter Table, Peeler Wiring
and Mill. The location of study area and distribution of 106 sampling points
(data was collected based on systematic method) is shown in Figure 1.
In this study for spatial prediction of noise exposure, 106
temporary noise monitoring stations in the Recovery Section were selected to
determine noise exposure status during eight hour working period. Moreover,
during noise measurement (eight hour working period), other factors (i.e.
Temperature and air condition) were mostly stable. To identify noisy areas,
noise survey was conducted within Recovery Section, and it was divided to 6 ×6
m stations. The sample sizes were selected based on the recovery section area. Noise
measurement was done in centre of each station at 2 meter height. The
measurement was done with a sound level meter model TES 52A on slow response
and A-filter. The sound level meter was held at arm's length. The sound level
meter had also been calibrated before and after each use. The calibrator model
was TES-1356, and calibration was done in 1000 Hz frequency and 94 dB
sound level.
After
collection and normalization of data (Logarithmic method), kriging and IDW
methods of interpolation were used. Finally, the best method of interpolation
was selected using cross-validation. The study was proceeded to prepare the map
based on this interpolation and the Geographical Information System. Figure 2 shows
the flowchart of this study.
Kriging: The presence of a spatial
structure where observations are close to each other
are more alike than those that are far apart (spatial autocorrelation) is a
prerequisite to the application of geostatistics (4;9). The experimental
variogram measures the average degree of dissimilarity between un-sampled
values and a nearby data value (10; 11), and thus can depict autocorrelation at
various distances. The value of the experimental variogram for a separation
distance of h (referred to as the lag) is half the average squared difference
between the value at z(xi) and the value at z(xi + h) (9;11).
Where n (h) is the number of data pairs within a given class of distance
and direction. If the values at Z (xi) and Z (xi + h) are auto correlated the
result of Eq. (1) will be small, relative to an uncorrelated pair of points.
From analysis of the experimental variogram, a suitable model (e.g. spherical,
exponential) is then fitted, usually by weighted least squares, and the
parameters (e.g. range, nugget and sill) are then used in the kriging
procedure.
IDW: In interpolation with IDW method, a weight is
attributed to the point to be measured. The amount of this weight is depended
to the distance of the point to another unknown point. These weights are
controlled on the bases of power of ten. With increase of power of ten, the
effect of the points that are farther diminishes. Lesser power distributes the
weights more uniformly between neighbouring points.
Figure1: Distribution of sampling point in Recovery section of
Telecommunication Cable Company
We should keep in mind that in this method the
distance between the points count, so the points of equal distance have equal
weights [12].
Figure 2:
Flowchart of Geostatistic study and selection of the best model for estimation
of variable
In this method
the weight factor is calculated with the use of the following formula:
Finally, RMSE
was used to evaluate model performances in cross-validation mode (13). The
smallest RMSE indicate the most accurate predictions. The RMSE was derived
according to Eqs. (4).
Results
The variable exhibits a non-normal distribution of measured values and
therefore does not initially satisfy the basic assumption of geostatistics of
statistical normality. This restriction is eliminated, by applying a data
amenable transform to the sample values that make them more to analysis and
estimation. The most useful data transform is the log-transform.
Since natural log values can be back transformed to real values, we can
use a semi-variogram model derived from the transformed sample values to
predict the spatial variation of logarithmic values of the noise. A statistical
summary of the measuring noise in various locations near machines is presented
in Table 1. As shown in this table, the parameter had high skeness, therefore
itwas normalized using logarithmic method (5).
Our task now is to fit models to the experimental or sample values
choosing models and fitting them to data remain among the most controversial
topics in geostatistic. There are still controversial who fit models by eye and
who defined their practice with vigor. They may justify their attitude on the
grounds that when kriging the resulting estimates are much the same for all
reasonable models of the variogram. There are others who fit models numerically
and automatically using “black box” software, often without any choice,
judgment or control.
Table 1: statistical analysis of noise value in
various locations at Recovery Section
Measuring location |
LP Min |
LP Max |
LP Mean |
SD |
Kurtosis |
skeness |
Exposure Time |
Near Mill Machine |
89.6 |
91.2 |
90.36 |
0.67 |
- 0.5 |
0.4 |
141 minutes |
Prickly Machine |
79.9 |
81.7 |
80.8 |
0.69 |
- 0.3 |
0.3 |
232 minutes |
Caterpillar |
79.8 |
81.7 |
80.98 |
0.69 |
- 0.3 |
0.4 |
21 minutes |
Halter Table |
79.6 |
81.4 |
80.66 |
0.69 |
- 0.5 |
0.2 |
95 minutes |
Peeler Wiring |
81.5 |
84.1 |
82.86 |
1.05 |
- 0.3 |
0.4 |
220 minutes |
Noise in all of Recovery section |
77.9 |
92.1 |
84.54 |
3.28 |
- 0.4 |
0.3 |
293 minutes |
This tool can have unfortunate consequences. We used a procedure that embodies
both visual inspection and statistical fitting, as follow. First plot the
experimental variogram. Then choose, from the models, one or more with
approximately the right shape and with sufficient detail to the principal
trends in the experimental values. The first step in using of kriging method is
investigating the presence of spatial structure among the data by variogram
analysis. For achievement to this issue, variograms was computed using normal
data. Variogram related to kriging method is presented in Figure 3. The best
model for fitting on experimental variogram was selected based on less RSS
values (Table 2). Therefore, it was recognized that Gaussian model is suitable
for estimation of noise value.
Figure 3: Variogram related to kriging method
Also, table 3 illustrates the parameters of the variogram. The ratio of
nugget variance to sill expressed in percentages can be regarded as a criterion
for classifying the spatial dependence of parameters. If this ratio is less
than 25%, then the variable has strong spatial dependence; if the ratio is
between 25 and 75%, the variable has moderate spatial dependence; and greater
than 75%, the variables shows only weak spatial dependence. Since this ratio
for all of the noise value is less than 25%, it has strong spatial dependence
(14).
Table 2: RSS value of different models of
variogram
Model |
Gaussian |
Spherical |
Exponential |
Linear |
RSS |
3.7e-7 |
5.08e-7 |
6.1e-7 |
4.8e-7 |
Table 3: Best-fitted variogram models and their
parameters
Property |
Model |
Nugget |
Sill |
Range |
Noise |
Gaussian |
1e-16 |
2.69e-16 |
32.5 |
For determination of the most suitable method, among Kriging and IDW,
RMSE was used. Results showed that geostatistical method had more considerable
accuracy than IDW (Table 4). Our results are similar to the findings of
Xiaopeng and Lingqing (15). They also reported that geostatistic methods had
more considerable accuracy than IDW method for preparing maps.
Table 4: Results of the interpolation error
based on RMSE value
IDW |
kriging |
Property |
||||
5 |
4 |
3 |
2 |
1 |
|
|
1.09 |
1.07 |
1.04 |
1.03 |
1.02 |
0.97 |
Noise |
Result of cross validation illustrated in figure 4. As shown in this
graph, accuracy of prediction method is reliable. Finally, a map of noise value
was prepared using kriging which was the best method for interpolation in GIS
environment (Figure 5).
Figure4: Cross-Validation of
Kriging method
Figure5: The map of spatial
distribution of noise value using of kriging
method
Discussion
The noise map (Figure 5) indicates the noise levels for eight hours in Recovery Section, and
it shows distributions of noise levels in all of locations in this Section. As
shown in this Fig., high noise levels between 84-88dB can be observed near
operational machines. In this area, there are a number of noise sources such as
Prickly Machine, Caterpillar, Halter Table, Peeler Wiring, Mill, vacuum pumps,
couch rolls, air and steam valves etc. Each makes a contribution to the overall
sound pressure level (noise level) at a given position. But the highest level
of noise, >88dB, is in the North of the hall where it is not near the major
machine in the field, it is probably related to old fans operating in this
area. Noise mapping in this study simply showed measurements at predetermined
positions identified by applying a 6 m grid to the floor plan which it has high
accuracy than traditional methods.
To indicate the locations with noise level higher than governmental
noise regulations, measuring noise levels monitored in the map were compared
with the Iranian Noise Standards which was produced by Ministry of Health (Figure6). As shown, the noise level exceeds Noise Standards in some locations
where workers were working. Also, it is indicated that noise, especially near
some machines, is higher or within standard regulation which is 85 dB A in
Iran.
Generally, analysis showed that kriging performed
better than IDW technique in characterizing the spatial variability which is in
line with the work done by Rizzo and Mouser (16); Nazari-Zadeh et al (17); and
Ahmad (18). They also revealed that geostatistical methods are the best model
for interpolation, but we must be careful about it.
Figure 6: The map of workers distribution
exposure to various levels of noise
Geostatistic obviously does not offer a statistical
model which is advantageous in every situation. Careful analysis of the
measurement data using common sense can sometimes result in the same
conclusions as those resulting from lengthily and computationally heavy
calculations. In general, as spacing between samples is large compared to the
dimensions of the investigated field, the potential advantageous of a
geostatisticalanalysis becomes less. For spacing beyond the range of spatial
auto-correlation, kriging estimates reduction of the same results as for the
classical random sampling. A geostatistical analysis is not only
computationally heavy, it requires also an impotent number of samples to be
taken and analysed as acute as possible. At least 30 to 50 pairs of
observations are necessary to calculate one point of the experimental
variogram. Since the lag range over which the variogram is calculated, it
should be approximately one fourth to one half of the dimension of the field
studied, the experimental variogram should contain points ranging from very
small to relative large lags.
As a result, geostatistical investigation is mostly
based on hundred, even thousands, of observation. If one observation of the
variable is costly, this requirement may jeopardize a geostatistical analysis.
Summarized, the disadvantageous of geostatistical approach toward the spatial
inventory of soil variables, also called are:
1- In
practice, observations need to be numerical.
2- Large
data sets are required.
3- Storing
information processing power is needed.
The advantages
are:
1- It
is a reproducible procedure which is easy to verify and update.
2- No
classification of data is required. Hence all problems concerning
classification disappear.
3- The
numerical output can serve as an input for further processing in GIS
4- It
yields as conceptually much more realistic inventory than the traditional
groundwater maps.
Conclusion
Workers in this industrial hall are exposed to
noise which is mainly induced by noisy machines. Noise maps which produced in this study showed the distribution of noise and, it revealed that workers suffer from serious noise pollution. Unacceptable noise in which the more percentage of the workers are exposed must be managed by control action urgently. This study showed that a useful way for evaluating and illustrating noise level in industrial halls is noise mapping. Additionally, Results showed that geostatistical method had
more considerable accuracy than IDW method. Using this method can
identify the locations which
immediately requirenoise controls.
Acknowledgements
The authors would like to acknowledge the University of Ardakan and the
recovery section of Ghandi telecommunication Cables Company inYazd province.
Conflict of interests: Non declared.
References
1.
Celik
O, Yalcin S, Ozturk A. Hearing parameters in noise exposed industrial workers.
Auris Nasus Larynx 1998; 25(4):369-75.
2.
Pourabdiyan
S, Ghotbi M, Yousefi HA, Habibi E, Zare
M. The epidemiologic study on hearing standard threshold shift using
audiometric data and noise level among workers of Isfehan metal industry.
Koomesh 2009; 10(4):253-60.
3.
Abedi
K, Zare M, Rahiminezhad M, Valipour E, Barkhordary A, Halvanee GH, et al.
Hearing loss among airport employees, Isfahan-Iran. Journal of Gorgan
University of Medical Sciences 2010; 11(4):57-63.
4.
Goovaerts
P. Geostatistics in soil science: state-of-the-art and perspectives. Geoderma
1999; 89(1-2):1-45.
5.
Webster
R, Oliver MA (2007). Geostatistics for environmental scientists. 2nd
ed. Australia, Brisbane: John Wiley & Sons.
6.
Istok
JD, Cooper RM. Geostatistics applied to groundwater pollution. III: Global
estimates. J Environ Eng 1988; 114(4):915-28.
7.
D'Agostino
V, Greene EA, Passarella G, Vurro M. Spatial and temporal study of nitrate
concentration in groundwater by means of coregionalization. Environ Geol 1998;
36(3-4):285-95.
8.
Tsai
KT, Lin MD, Chen YH. Noise mapping in urban environments: A Taiwan study.
Applied Acoustics 2009; 70(7):964-72.
9.
Robinson
TP, Metternicht G. Testing the performance of spatial interpolation techniques
for mapping soil properties. Computers and Electronics in Agriculture 2006;
50(2):97-108.
10.
Deutsch
CV, Journel AG (1998). GSLIB: Geostatistical software library and user's guide.
2nd ed. New York, Oxford: Oxford University Press. P369.
11.
Lark
RM. Estimating variograms of soil properties by the method-of-moments and
maximum likelihood. Eur J Soil Sci 2000; 51(4):717-28.
12.
Burrough
PA, McDonnell RA. Creating continuous surfaces from point data. In: Principles
of Geographic Information Systems. 2nd ed. Oxford: Oxford University
Press. P 98-131.
13.
Davis
BM. Uses and abuses of cross-validation in geostatistics. Math Geol 1987;
19(3):241-8.
14.
Shi
J, Wang H, Xu J, Wu J, Liu X, Zhu H, et al. Spatial distribution of heavy
metals in soils: a case study of Changxing, China. Environ Geol 2007;
52(1):1-10.
15.
Yang
X, Wang L. Spatial analysis and hazard assessment of mercury in soil around the
coal-fired power plant: a case study from the city of Baoji, China. Environ
Geol 2008; 53(7):1381-8.
16. Nazari-zade
F, Arshadiyan F, Zand-vakily K. Study of spatial variability of Groundwater
quality of Balarood Plain in Khuzestan province. Proceedings of
The first congress of
optimized exploitation from water source of
Karoon and Zayanderood Plain: Shahrekord: Shahrekord University; 2006;
P1236-40.
17. Rizzo DM,
Mouser PJ. Evaluation of Geostatistics for Combined Hydrochemistry and
Microbial Community Fingerprinting at a Waste Disposal Site. Critical
Transitions in Water and Environmental Resources Management; 2000. p1-11.
18. Ahmed S. Groundwater monitoring
network design: Application of Geostatistics with a few Case studies from a
granitic aquifer in a semiarid region. In: Groundwater Hydrology. Sherif MM,
Singh VP, Al-Rashed M, editors. Japan: Tokyo, Balkema; 2002; P37-57.
*Corresponding author: Ruhollah Taghizadeh-Mehrjardi, Faculty of Agriculture and Natural Resources, University of Ardakan, Ardakan, Iran. Email Address: rh_taghizade@yahoo.com