- Email: [email protected]
Wind induced ambient noise modelling and comparison with field measurements in Arabian Sea
Applied Acoustics 89 (2015) 101–106
Contents lists available at ScienceDirect
Applied Acoustics journal homepage: www.elsevier.com/locate/apacoust
Wind induced ambient noise modelling and comparison with field measurements in Arabian Sea S. Najeem ⇑, M.C. Sanjana, G. Latha, P. Edwards Durai Ocean Acoustics, National Institute of Ocean Technology, NIOT Campus, Pallikaranai, Chennai 600 100, Tamil Nadu, India
a r t i c l e
i n f o
Article history: Received 25 April 2014 Received in revised form 8 September 2014 Accepted 15 September 2014
Keywords: Ambient noise Coherence Noise level CANARY
a b s t r a c t The present work focuses on modelling of wind induced ambient noise and comparison of noise properties with field measurements. Time series ambient noise data were taken from shallow waters of Arabian Sea, using a noise recording system consisting of a vertical array with twelve hydrophones. From the field measurements vertical coherence and noise spectral level were estimated and compared with the predictions from Harrison’s formulas for coherence and noise level based on ray theory. Considering that the noise generated by the action of wind dominates in mid frequency range, this study concentrates on the band 0.2–5 kHz. Modelled results of both coherence and noise level compare well with field measurements. Moreover the simulated fluctuations in noise level with wind speed are in accordance with measured data. This is an initial attempt made to simulate noise properties in Arabian Sea and the model can be further used for geoacoustic inversion for extracting seabed parameters of the site. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Ambient noise is generally considered as the unwanted background sound present in the ocean at a particular spot, where a measuring hydrophone is located. The information regarding ambient noise is one of the major input parameter required for the performance of both active and passive sonar, since it could diminish the performance. Wind, rain, marine life and ship traffic are the major sources of ambient noise in the Ocean. Above all, background sound induced by wind is considered as a predominant and constant source of noise in both deep and shallow water. Wind generated noise in shallow water sites are generally about 6–8 dB higher than that in deep water due to the environmental difference [1]. Even though ambient noise is considered as a nuisance, it contains information about the environment where it propagates. This makes it suitable for studying the medium through which it propagates as well as its boundaries. Extraction of sea bed geoacoustic properties using ambient noise is an active research field in underwater acoustics and several studies have been reported [2–4]. Earlier studies show that absolute noise level due to wind and rain depends on wind speed and rainfall rate in the vicinity of the receiver. Thus ocean ambient noise measured at a particular site can be used to figure out wind speed and rainfall rate [5,6]. ⇑ Corresponding author. Tel.: +91 44 66783390. E-mail address: [email protected] (S. Najeem). http://dx.doi.org/10.1016/j.apacoust.2014.09.014 0003-682X/Ó 2014 Elsevier Ltd. All rights reserved.
Measurement of ambient noise for long period of time encounters difficulty considering sensor failure as a result of biological activities and natural calamities. Modelling is an alternative approach to characterize ambient noise field at a particular site. Ambient noise modelling is a complicated subject which includes the prediction of its spatial and temporal properties. Studies concentrating on this regard have started since 1960’s and models based on different theories have been developed. An early analytical model of wind induced noise has been proposed by Cron and Sherman [7]. They considered directional noise from the surface without reflections from the seabed. Two seminal papers published in 1980’s by Kuperman–Ingenito and Buckingham based on normal mode method provided the theoretical foundation for shallow water ambient noise modelling [8,9]. A detailed review of available ambient noise models were given by Hamson [10]. Harrison developed a formula for coherence based on ray theory and the model is known as CANARY (Coherence and Ambient Noise for ARraYs). Harrison’s formula for coherence and noise level is applicable for a range independent ocean environment and later he extended it for a range dependent condition [11–13]. For surface distributed source such as wind CANARY assumes a dipole radiation pattern and the resultant coherence function for two receivers can be expressed in a single equation. Despite the fact that ambient noise studies are meager, measurement and characterization of ambient noise due to wind and shipping in shallow waters off the west and east coast of India have been reported as early as 2007 [14–16].
102
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
In this work vertical coherence and noise spectral level were estimated from field measurements and compared with the simulated results from Harrison’s ray based model [12]. Ambient noise data collected off the west coast of India at 30 m water depth during wind prevailing condition were used for comparison. The rest of the paper is organized as: Section 2 briefly introduces data collection methods and mathematical equations governing the ambient noise model. Comparison of modelled results such as vertical coherence and noise spectral level with field measurements is described in Section 3. Good agreement is achieved between the model and measured data in the Arabian Sea. 2. Measurements and processing methodology 2.1. Data collection An automated subsurface noise recording system was developed and deployed at 30 m depth for time series measurements off the Cochin coast in Arabian Sea. Measurement system was deployed from April 2011 to June 2011 and the datasets measured during peak summer were used for analysis and comparison. Location of deployment and schematic of the noise measurement system are shown in Figs. 1a and 1b. Measurement system consists of data acquisition modules and a vertical linear array of 12 hydrophones with uniform sensor separation of 0.15 m. The omnidirectional hydrophones in the array are capable of measuring noise in the frequency range 0.1–8 kHz with a receiving sensitivity of 170 dB. Noise data were sampled at 50 kHz, recorded for 30 s with eight samples in a day at an interval of 3 h. In addition to ambient noise other environmental information’s such as wind speed, sediment samples and sound speed profile from the site were also collected. 2.2. Model description Harrison’s ray based model for coherence has been used for the simulation of noise properties [12]. Model represents the result for cross correlation between two receivers in a single expression. For an array of receivers cross correlation matrix for different sensor combinations can be derived from that formula. Finally ambient noise properties such as coherence, vertical directionality and absolute level can be estimated. The environmental parameters that can be included in the model are sound speed profile, bottom geoacoustic properties in terms of bottom reflection coefficient, surface roughness in terms of surface reflection coefficient and volume attenuation. The closed form solution used in the model for coherence between two receivers for a range independent environment can be written as [12]
Cðd; cÞ ¼ 2p
Z p=2
½1 Rs ðhs ÞRb ðhb Þ expðasc Þ1
0
ðexpðikd sin hr sin cÞ expðasp Þ þ Rb ðhb Þ expðikd sin hr sin cÞ expðaðsc sp ÞÞÞ J0 ðkd cos hr cos cÞNðhs Þ cos hr dhr
ð1Þ
where d and c represents the spacing and orientation between two receivers. hs and hb are surface and bottom ray angles related to the receiver angle hr by Snell’s law. Rs(hs) and Rb(hb) are the plane wave reflection coefficients at surface and bottom respectively. N(hs) is the function representing surface noise source spectral density with dipole radiation pattern. k is the wave number at the receivers and a is volume attenuation. sc and sp are the full and partial path length of the ray from source to receiver and can be estimated as
sc ¼ ð2H hÞ= sinððhb þ hs =2ÞÞ
ð2Þ
sp ¼ h= sinððhr þ hs =2ÞÞ
ð3Þ
where h and H are receiver and water column depths, respectively. Equation for coherence gives absolute ambient noise level for a single hydrophone when the element spacing is set to zero.
Z p=2
½1 Rs ðhs ÞRb ðhb Þ expðasc Þ1 expðasp Þ 0 þ Rb ðhb Þ expðaðsc sp ÞÞ Nðhs Þ cos hr dhr
Cðd ¼ 0Þ ¼ 2p
ð4Þ
As an input to the model certain environmental information must be either known or estimated. Geoacoustic parameters were included in the model in terms of bottom reflection coefficient for a seabed which supports shear [17]. Surface reflection coefficient used in the model is based on Ainslie formula and is related to wind speed and frequency [18]. Volume attenuation is considered to be negligible since it has limited effect at shallow waters and lower frequencies [19]. The surface source spectrum level used in the model for wind induced noise is based on Kuperman–Ferla and Ainslie [20,21]. Based on the above environmental inputs, spatial coherence and ambient noise level were simulated and compared with field measurements.
3. Results and discussions 3.1. Vertical coherence Coherence function represents the similarity between two time series signals and theoretically it is estimated from cross spectral density normalized by the product of individual power spectral densities. Measured coherence function is complex with both real and imaginary components where the presence of imaginary part implies asymmetry in the noise field along vertical. Spatial coherence due to wind induced ambient noise is a stable feature which depends on seabed geoacoustic parameters, sound speed profile and water column depth. The sound speed profile measured using SVP from the site was downward refracting and shown in Fig. 2a. Sediment samples collected from the site were subjected to sieve analysis and found to be composed of silt, clay and sand. Seabed parameters for reflection loss estimation are taken from Hamilton’s values for continental shelf environment as: compressional sound speed = 1579 m/s, shear speed = 55.2 m/s, compressional attenuation = 0.8 dB/c and density = 1596 kg/m3 respectively [17,22]. Shear attenuation is considered as negligible since water column sound speed is comparatively so much higher than that of shear speed. Estimated bottom reflection loss for the environment is shown in Fig. 2b and it is found to be comparatively higher due to soft sediment composition. Sea surface is considered as totally reflecting since noise coherence is more sensitive to seabed geoacoustic properties. Based on the above environmental inputs and receivers at mid water column with separation 0.75 m, vertical coherence is estimated using Eq. (1) for frequencies of 0.2–5 kHz. In order to estimate the vertical coherence from field measurements first and sixth sensor of the array with separation 0.75 m is taken. Three consecutive data sets during wind prevailing conditions were used for the extraction of vertical coherence. Presence of imaginary component in coherence pattern confirms anisotropy in noise field along vertical direction. Comparison of coherence pattern for real and imaginary components between model and data (average) are shown in Fig. 3. Both real and imaginary parts of coherence function from data fits fairly well with the model. There are two dominant factors which contributes vertical asymmetry in noise field at this location. Firstly downward refracting sound speed profile which supports the downward propagation of locally generated noise and secondly soft sediment composition which absorbs most of the sound interacting with it.
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
103
Fig. 1a. Location of deployment site in the Arabian Sea.
3.2. Noise level Ten days of time series data with wind speed varying from 1.5 m/s to 7 m/s were used for the analysis of ambient noise spectrum level. Noise data from sixth element of the hydrophone array has been taken since it is positioned around mid water column. The spectra of the measurements are produced using 4096 point discrete Fourier transforms of 30 s data. Measured data sets were separated based on wind speed and averaged value is used for comparison. Analysis of noise spectrum clearly shows increase in ambient noise level with respect to wind speed when there is no interference from other sources. The major parameter required for prediction of ambient noise level is the surface source strength induced by wind. The composite formula for surface source dipole spectrum level used in the model is applicable from 0.4 kHz and below that it shows a flat spectrum. The estimated Sea surface loss at the maximum observed wind speed of 7 m/s is shown in Fig. 4. The received
intensity at a particular frequency, ocean depth and receiver angle can be obtained from the model based on Eq. (4) and converted into ambient noise level by integrating over all angles. Comparison of modelled ambient noise spectrum level and measured data for different wind speeds are shown in Figs. 5a–5d. Good agreement was observed between modelled and measured noise spectrum level at different wind speeds. When the wind speed is 2.5 m/s, modelled ambient noise level shows close resemblance with measured data as shown in Fig. 5a. Estimated root mean square error between model and data at 2.5 m/s is found to be 1 dB. Comparison of ambient noise spectra level between model and data at 4 m/s and 5.4 m/s are shown in Figs. 5b and 5c. The results show good correlation between data and model at both wind speeds with corresponding root mean square error estimate of 0.75 and 0.55 dB. At the highest observed wind speed of 7 m/s measured noise level shows a significant variation from the model results at frequency below 1 kHz and is shown in Fig. 5d. Noise generated majorly due to splashing sound from the buoy and shipping
104
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
Fig. 3. Comparison of real and imaginary parts of coherence from model and data.
Fig. 4. Surface reflection loss at wind speed 7 m/s in dB.
Fig. 1b. Subsurface system for ambient noise measurement.
Fig. 5a. Comparison of received noise level from model and data at wind speed 2.5 m/s.
Fig. 2a. Measured sound speed profile of the site.
Fig. 5b. Comparison of received noise level from model and data at wind speed 4 m/s.
Fig. 2b. Bottom reflection loss for the site.
causes this variation at low frequencies. Even though non uniform sources such as shipping can be included in the model, this study focuses on background noise generated due to surface distributed source such as wind with dipole directionality pattern. Due to these sources root mean square error estimate between model and data
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
Fig. 5c. Comparison of received noise level from model and data at wind speed 5.4 m/s.
105
source of ambient noise over the frequency range of 0.5–10 kHz. Received ambient noise level during wind prevailing condition solely depends on the locally generated wind speed. In this study Harrison’s formula for coherence has been used to simulate coherence and absolute level and compared with field measurements. Estimated vertical coherence from field measurements for both real and imaginary components fits fairly well with simulated results. Later ambient noise spectral level at different wind speeds were analyzed and compared with simulation. It is observed that the general pattern for noise spectrum level above 1 kHz has got good agreement with model. The overall increase in Noise level predicted by the model at different wind speeds are comparing well with field measurements. Measured ambient noise from a vertical array can be further used for extracting number of layers and geoacoustic properties of the seabed [23,24]. Acknowledgements
Fig. 5d. Comparison of received noise level from model and data at wind speed 7 m/s.
The authors gratefully acknowledge Director, National Institute of Ocean Technology for the facilities and support provided for this study. The authors acknowledge Dr. C.H. Harrison of Centre for Marine Research and Experimentation (CMRE) for clarifying the doubts regarding the formulas for ambient noise level and coherence. The authors would like to thank Mrs. Malarkodi for testing and calibration of hydrophones. The authors would also like to thank ocean acoustics filed team consisting of Mr. A. Thirunavukkarasu, Mr. G. Raguraman, Mr. M.M. Mohanty, Mr. M. Ashokan, Mr. K. Nityanandam and Mr. C. Dhanraj. References
Fig. 6. Noise level variation with wind speed and comparison with field measurements.
at 7 m/s is found to be 2 dB. The variation of ambient noise spectral level above 1 kHz is consistent with the model predictions at all wind speeds. Analysis of noise spectrum level confirmed wind as the major noise generating mechanism for the measurement site. Observed ambient noise level increases with wind speed since it enhances sea surface agitation. Thus increase in noise level with wind speed is simulated and compared with field observation. Simulated noise level with wind speed at 1, 2, 3 and 4 kHz and the corresponding measured values are shown in Fig. 6. Measurements from the field agree well with simulation as the wind speed increases.
4. Conclusions Since the contributions from surface distributed sources create a complex noise field, modelling of ambient noise becomes more challenging. Action of wind at the ocean surface is a dominant
[1] Knobles DP, Joshi SM, Gaul RD, Graber HC, Williams NJ. Analysis of wind driven ambient noise in a shallow water environment with a sandy seabed. J Acoust Soc Am 2008;124:EL157–62. [2] Buckingham MJ, Jones SAS. A new shallow ocean technique for determining the critical angle of the seabed from the vertical directionality of the ambient noise in the water column. J Acoust Soc Am 1980;81:938–46. [3] Carbone NM, Deane GB, Buckingham MJ. Estimating the compressional and shear wave speeds of shallow water seabed from the vertical coherence of ambient noise in the water column. J Acoust Soc Am 1998;103:801–13. [4] Harrison CH, Simons DG. Geoacoustic inversion of ambient noise: a simple method. J Acoust Soc Am 2002;112:1377–89. [5] Vagle S, Large WG, Farmer DM. An evaluation of the WOTAN technique for inferring oceanic wind from underwater sound. J Atmos Ocean Technol 1990;7:576–95. [6] Nystuen JA. Rainfall measurements using underwater ambient noise. J Acoust Soc Am 1986;79:972–82. [7] Cron BF, Sherman CH. Spatial correlation functions for various noise models. J Acoust Soc Am 1962;34:1732–6. [8] Kuperman WA, Ingenito F. Spatial correlation of surface generated noise in a stratified media. J Acoust Soc Am 1980;67:1988–96. [9] Buckingham MJ. A theoretical model of ambient noise in low loss shallow water channel. J Acoust Soc Am 1980;67:1186–992. [10] Hamson RM. The modelling of ambient noise due to shipping and wind sources in complex environments. Appl Acoust 1997;51:251–87. [11] Harrison CH. CANARY: a simple model of ambient noise and coherence. Appl Acoust 1997;51:289–315. [12] Harrison CH. Formulas for ambient noise and coherence. J Acoust Soc Am 1996;99:2055–66. [13] Harrison CH. Noise directionality for surface sources in range-dependent environments. J Acoust Soc Am 1997;102:2655–62. [14] Ramji S, Latha G, Rajendran V, Ramakrishnan S. Wind dependence of ambient noise in shallow waters of Bay of Bengal. Appl Acoust 2008;69:1294–8. [15] Das A. Shallow ambient noise variability due to distant shipping and tide. Appl Acoust 2011;72:660–4. [16] Sanjana MC, Latha G, Mahanty MM. Seabed characteristics from ambient noise at three shallow water sites in Northern Indian Ocean. J Acoust Soc Am 2013;34(4):EL 366–72. [17] Jensen FB, Kuperman WA, Porter MB, Schmidt H. Computational ocean acoustics. 2nd ed. New York: Springer; 2011. [18] Ainslie MA. Effect of wind generated bubbles on fixed range acoustic attenuation in shallow water at 1–4 kHz. J Acoust Soc Am 2005;118:3513–23. [19] Barclay DR, Buckingham MJ. Depth dependence of wind driven broadband ambient noise in the Philippine Sea. J Acoust Soc Am 2013;133:62–71. [20] Kuperman WA, Ferla MC. A shallow water experiment to determine the source spectrum level of wind generated noise. J Acoust Soc Am 1985;77:2067–73.
106
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
[21] Ainslie MA. Principles of sonar performance modelling. London: Springer; 2010. [22] Hamilton ED. Geoacoustic modeling of the seafloor. J Acoust Soc Am 1980;68:1313–40.
[23] Harrison CH. Sub-bottom profiling using ocean ambient noise. J Acoust Soc Am 2004;115:1505–15. [24] Siderius M, Harrison CH. High-frequency geoacoustic inversion of ambient noise data using short arrays. AIP Conf Proc 2004;728:22–31.
Contents lists available at ScienceDirect
Applied Acoustics journal homepage: www.elsevier.com/locate/apacoust
Wind induced ambient noise modelling and comparison with field measurements in Arabian Sea S. Najeem ⇑, M.C. Sanjana, G. Latha, P. Edwards Durai Ocean Acoustics, National Institute of Ocean Technology, NIOT Campus, Pallikaranai, Chennai 600 100, Tamil Nadu, India
a r t i c l e
i n f o
Article history: Received 25 April 2014 Received in revised form 8 September 2014 Accepted 15 September 2014
Keywords: Ambient noise Coherence Noise level CANARY
a b s t r a c t The present work focuses on modelling of wind induced ambient noise and comparison of noise properties with field measurements. Time series ambient noise data were taken from shallow waters of Arabian Sea, using a noise recording system consisting of a vertical array with twelve hydrophones. From the field measurements vertical coherence and noise spectral level were estimated and compared with the predictions from Harrison’s formulas for coherence and noise level based on ray theory. Considering that the noise generated by the action of wind dominates in mid frequency range, this study concentrates on the band 0.2–5 kHz. Modelled results of both coherence and noise level compare well with field measurements. Moreover the simulated fluctuations in noise level with wind speed are in accordance with measured data. This is an initial attempt made to simulate noise properties in Arabian Sea and the model can be further used for geoacoustic inversion for extracting seabed parameters of the site. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Ambient noise is generally considered as the unwanted background sound present in the ocean at a particular spot, where a measuring hydrophone is located. The information regarding ambient noise is one of the major input parameter required for the performance of both active and passive sonar, since it could diminish the performance. Wind, rain, marine life and ship traffic are the major sources of ambient noise in the Ocean. Above all, background sound induced by wind is considered as a predominant and constant source of noise in both deep and shallow water. Wind generated noise in shallow water sites are generally about 6–8 dB higher than that in deep water due to the environmental difference [1]. Even though ambient noise is considered as a nuisance, it contains information about the environment where it propagates. This makes it suitable for studying the medium through which it propagates as well as its boundaries. Extraction of sea bed geoacoustic properties using ambient noise is an active research field in underwater acoustics and several studies have been reported [2–4]. Earlier studies show that absolute noise level due to wind and rain depends on wind speed and rainfall rate in the vicinity of the receiver. Thus ocean ambient noise measured at a particular site can be used to figure out wind speed and rainfall rate [5,6]. ⇑ Corresponding author. Tel.: +91 44 66783390. E-mail address: [email protected] (S. Najeem). http://dx.doi.org/10.1016/j.apacoust.2014.09.014 0003-682X/Ó 2014 Elsevier Ltd. All rights reserved.
Measurement of ambient noise for long period of time encounters difficulty considering sensor failure as a result of biological activities and natural calamities. Modelling is an alternative approach to characterize ambient noise field at a particular site. Ambient noise modelling is a complicated subject which includes the prediction of its spatial and temporal properties. Studies concentrating on this regard have started since 1960’s and models based on different theories have been developed. An early analytical model of wind induced noise has been proposed by Cron and Sherman [7]. They considered directional noise from the surface without reflections from the seabed. Two seminal papers published in 1980’s by Kuperman–Ingenito and Buckingham based on normal mode method provided the theoretical foundation for shallow water ambient noise modelling [8,9]. A detailed review of available ambient noise models were given by Hamson [10]. Harrison developed a formula for coherence based on ray theory and the model is known as CANARY (Coherence and Ambient Noise for ARraYs). Harrison’s formula for coherence and noise level is applicable for a range independent ocean environment and later he extended it for a range dependent condition [11–13]. For surface distributed source such as wind CANARY assumes a dipole radiation pattern and the resultant coherence function for two receivers can be expressed in a single equation. Despite the fact that ambient noise studies are meager, measurement and characterization of ambient noise due to wind and shipping in shallow waters off the west and east coast of India have been reported as early as 2007 [14–16].
102
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
In this work vertical coherence and noise spectral level were estimated from field measurements and compared with the simulated results from Harrison’s ray based model [12]. Ambient noise data collected off the west coast of India at 30 m water depth during wind prevailing condition were used for comparison. The rest of the paper is organized as: Section 2 briefly introduces data collection methods and mathematical equations governing the ambient noise model. Comparison of modelled results such as vertical coherence and noise spectral level with field measurements is described in Section 3. Good agreement is achieved between the model and measured data in the Arabian Sea. 2. Measurements and processing methodology 2.1. Data collection An automated subsurface noise recording system was developed and deployed at 30 m depth for time series measurements off the Cochin coast in Arabian Sea. Measurement system was deployed from April 2011 to June 2011 and the datasets measured during peak summer were used for analysis and comparison. Location of deployment and schematic of the noise measurement system are shown in Figs. 1a and 1b. Measurement system consists of data acquisition modules and a vertical linear array of 12 hydrophones with uniform sensor separation of 0.15 m. The omnidirectional hydrophones in the array are capable of measuring noise in the frequency range 0.1–8 kHz with a receiving sensitivity of 170 dB. Noise data were sampled at 50 kHz, recorded for 30 s with eight samples in a day at an interval of 3 h. In addition to ambient noise other environmental information’s such as wind speed, sediment samples and sound speed profile from the site were also collected. 2.2. Model description Harrison’s ray based model for coherence has been used for the simulation of noise properties [12]. Model represents the result for cross correlation between two receivers in a single expression. For an array of receivers cross correlation matrix for different sensor combinations can be derived from that formula. Finally ambient noise properties such as coherence, vertical directionality and absolute level can be estimated. The environmental parameters that can be included in the model are sound speed profile, bottom geoacoustic properties in terms of bottom reflection coefficient, surface roughness in terms of surface reflection coefficient and volume attenuation. The closed form solution used in the model for coherence between two receivers for a range independent environment can be written as [12]
Cðd; cÞ ¼ 2p
Z p=2
½1 Rs ðhs ÞRb ðhb Þ expðasc Þ1
0
ðexpðikd sin hr sin cÞ expðasp Þ þ Rb ðhb Þ expðikd sin hr sin cÞ expðaðsc sp ÞÞÞ J0 ðkd cos hr cos cÞNðhs Þ cos hr dhr
ð1Þ
where d and c represents the spacing and orientation between two receivers. hs and hb are surface and bottom ray angles related to the receiver angle hr by Snell’s law. Rs(hs) and Rb(hb) are the plane wave reflection coefficients at surface and bottom respectively. N(hs) is the function representing surface noise source spectral density with dipole radiation pattern. k is the wave number at the receivers and a is volume attenuation. sc and sp are the full and partial path length of the ray from source to receiver and can be estimated as
sc ¼ ð2H hÞ= sinððhb þ hs =2ÞÞ
ð2Þ
sp ¼ h= sinððhr þ hs =2ÞÞ
ð3Þ
where h and H are receiver and water column depths, respectively. Equation for coherence gives absolute ambient noise level for a single hydrophone when the element spacing is set to zero.
Z p=2
½1 Rs ðhs ÞRb ðhb Þ expðasc Þ1 expðasp Þ 0 þ Rb ðhb Þ expðaðsc sp ÞÞ Nðhs Þ cos hr dhr
Cðd ¼ 0Þ ¼ 2p
ð4Þ
As an input to the model certain environmental information must be either known or estimated. Geoacoustic parameters were included in the model in terms of bottom reflection coefficient for a seabed which supports shear [17]. Surface reflection coefficient used in the model is based on Ainslie formula and is related to wind speed and frequency [18]. Volume attenuation is considered to be negligible since it has limited effect at shallow waters and lower frequencies [19]. The surface source spectrum level used in the model for wind induced noise is based on Kuperman–Ferla and Ainslie [20,21]. Based on the above environmental inputs, spatial coherence and ambient noise level were simulated and compared with field measurements.
3. Results and discussions 3.1. Vertical coherence Coherence function represents the similarity between two time series signals and theoretically it is estimated from cross spectral density normalized by the product of individual power spectral densities. Measured coherence function is complex with both real and imaginary components where the presence of imaginary part implies asymmetry in the noise field along vertical. Spatial coherence due to wind induced ambient noise is a stable feature which depends on seabed geoacoustic parameters, sound speed profile and water column depth. The sound speed profile measured using SVP from the site was downward refracting and shown in Fig. 2a. Sediment samples collected from the site were subjected to sieve analysis and found to be composed of silt, clay and sand. Seabed parameters for reflection loss estimation are taken from Hamilton’s values for continental shelf environment as: compressional sound speed = 1579 m/s, shear speed = 55.2 m/s, compressional attenuation = 0.8 dB/c and density = 1596 kg/m3 respectively [17,22]. Shear attenuation is considered as negligible since water column sound speed is comparatively so much higher than that of shear speed. Estimated bottom reflection loss for the environment is shown in Fig. 2b and it is found to be comparatively higher due to soft sediment composition. Sea surface is considered as totally reflecting since noise coherence is more sensitive to seabed geoacoustic properties. Based on the above environmental inputs and receivers at mid water column with separation 0.75 m, vertical coherence is estimated using Eq. (1) for frequencies of 0.2–5 kHz. In order to estimate the vertical coherence from field measurements first and sixth sensor of the array with separation 0.75 m is taken. Three consecutive data sets during wind prevailing conditions were used for the extraction of vertical coherence. Presence of imaginary component in coherence pattern confirms anisotropy in noise field along vertical direction. Comparison of coherence pattern for real and imaginary components between model and data (average) are shown in Fig. 3. Both real and imaginary parts of coherence function from data fits fairly well with the model. There are two dominant factors which contributes vertical asymmetry in noise field at this location. Firstly downward refracting sound speed profile which supports the downward propagation of locally generated noise and secondly soft sediment composition which absorbs most of the sound interacting with it.
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
103
Fig. 1a. Location of deployment site in the Arabian Sea.
3.2. Noise level Ten days of time series data with wind speed varying from 1.5 m/s to 7 m/s were used for the analysis of ambient noise spectrum level. Noise data from sixth element of the hydrophone array has been taken since it is positioned around mid water column. The spectra of the measurements are produced using 4096 point discrete Fourier transforms of 30 s data. Measured data sets were separated based on wind speed and averaged value is used for comparison. Analysis of noise spectrum clearly shows increase in ambient noise level with respect to wind speed when there is no interference from other sources. The major parameter required for prediction of ambient noise level is the surface source strength induced by wind. The composite formula for surface source dipole spectrum level used in the model is applicable from 0.4 kHz and below that it shows a flat spectrum. The estimated Sea surface loss at the maximum observed wind speed of 7 m/s is shown in Fig. 4. The received
intensity at a particular frequency, ocean depth and receiver angle can be obtained from the model based on Eq. (4) and converted into ambient noise level by integrating over all angles. Comparison of modelled ambient noise spectrum level and measured data for different wind speeds are shown in Figs. 5a–5d. Good agreement was observed between modelled and measured noise spectrum level at different wind speeds. When the wind speed is 2.5 m/s, modelled ambient noise level shows close resemblance with measured data as shown in Fig. 5a. Estimated root mean square error between model and data at 2.5 m/s is found to be 1 dB. Comparison of ambient noise spectra level between model and data at 4 m/s and 5.4 m/s are shown in Figs. 5b and 5c. The results show good correlation between data and model at both wind speeds with corresponding root mean square error estimate of 0.75 and 0.55 dB. At the highest observed wind speed of 7 m/s measured noise level shows a significant variation from the model results at frequency below 1 kHz and is shown in Fig. 5d. Noise generated majorly due to splashing sound from the buoy and shipping
104
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
Fig. 3. Comparison of real and imaginary parts of coherence from model and data.
Fig. 4. Surface reflection loss at wind speed 7 m/s in dB.
Fig. 1b. Subsurface system for ambient noise measurement.
Fig. 5a. Comparison of received noise level from model and data at wind speed 2.5 m/s.
Fig. 2a. Measured sound speed profile of the site.
Fig. 5b. Comparison of received noise level from model and data at wind speed 4 m/s.
Fig. 2b. Bottom reflection loss for the site.
causes this variation at low frequencies. Even though non uniform sources such as shipping can be included in the model, this study focuses on background noise generated due to surface distributed source such as wind with dipole directionality pattern. Due to these sources root mean square error estimate between model and data
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
Fig. 5c. Comparison of received noise level from model and data at wind speed 5.4 m/s.
105
source of ambient noise over the frequency range of 0.5–10 kHz. Received ambient noise level during wind prevailing condition solely depends on the locally generated wind speed. In this study Harrison’s formula for coherence has been used to simulate coherence and absolute level and compared with field measurements. Estimated vertical coherence from field measurements for both real and imaginary components fits fairly well with simulated results. Later ambient noise spectral level at different wind speeds were analyzed and compared with simulation. It is observed that the general pattern for noise spectrum level above 1 kHz has got good agreement with model. The overall increase in Noise level predicted by the model at different wind speeds are comparing well with field measurements. Measured ambient noise from a vertical array can be further used for extracting number of layers and geoacoustic properties of the seabed [23,24]. Acknowledgements
Fig. 5d. Comparison of received noise level from model and data at wind speed 7 m/s.
The authors gratefully acknowledge Director, National Institute of Ocean Technology for the facilities and support provided for this study. The authors acknowledge Dr. C.H. Harrison of Centre for Marine Research and Experimentation (CMRE) for clarifying the doubts regarding the formulas for ambient noise level and coherence. The authors would like to thank Mrs. Malarkodi for testing and calibration of hydrophones. The authors would also like to thank ocean acoustics filed team consisting of Mr. A. Thirunavukkarasu, Mr. G. Raguraman, Mr. M.M. Mohanty, Mr. M. Ashokan, Mr. K. Nityanandam and Mr. C. Dhanraj. References
Fig. 6. Noise level variation with wind speed and comparison with field measurements.
at 7 m/s is found to be 2 dB. The variation of ambient noise spectral level above 1 kHz is consistent with the model predictions at all wind speeds. Analysis of noise spectrum level confirmed wind as the major noise generating mechanism for the measurement site. Observed ambient noise level increases with wind speed since it enhances sea surface agitation. Thus increase in noise level with wind speed is simulated and compared with field observation. Simulated noise level with wind speed at 1, 2, 3 and 4 kHz and the corresponding measured values are shown in Fig. 6. Measurements from the field agree well with simulation as the wind speed increases.
4. Conclusions Since the contributions from surface distributed sources create a complex noise field, modelling of ambient noise becomes more challenging. Action of wind at the ocean surface is a dominant
[1] Knobles DP, Joshi SM, Gaul RD, Graber HC, Williams NJ. Analysis of wind driven ambient noise in a shallow water environment with a sandy seabed. J Acoust Soc Am 2008;124:EL157–62. [2] Buckingham MJ, Jones SAS. A new shallow ocean technique for determining the critical angle of the seabed from the vertical directionality of the ambient noise in the water column. J Acoust Soc Am 1980;81:938–46. [3] Carbone NM, Deane GB, Buckingham MJ. Estimating the compressional and shear wave speeds of shallow water seabed from the vertical coherence of ambient noise in the water column. J Acoust Soc Am 1998;103:801–13. [4] Harrison CH, Simons DG. Geoacoustic inversion of ambient noise: a simple method. J Acoust Soc Am 2002;112:1377–89. [5] Vagle S, Large WG, Farmer DM. An evaluation of the WOTAN technique for inferring oceanic wind from underwater sound. J Atmos Ocean Technol 1990;7:576–95. [6] Nystuen JA. Rainfall measurements using underwater ambient noise. J Acoust Soc Am 1986;79:972–82. [7] Cron BF, Sherman CH. Spatial correlation functions for various noise models. J Acoust Soc Am 1962;34:1732–6. [8] Kuperman WA, Ingenito F. Spatial correlation of surface generated noise in a stratified media. J Acoust Soc Am 1980;67:1988–96. [9] Buckingham MJ. A theoretical model of ambient noise in low loss shallow water channel. J Acoust Soc Am 1980;67:1186–992. [10] Hamson RM. The modelling of ambient noise due to shipping and wind sources in complex environments. Appl Acoust 1997;51:251–87. [11] Harrison CH. CANARY: a simple model of ambient noise and coherence. Appl Acoust 1997;51:289–315. [12] Harrison CH. Formulas for ambient noise and coherence. J Acoust Soc Am 1996;99:2055–66. [13] Harrison CH. Noise directionality for surface sources in range-dependent environments. J Acoust Soc Am 1997;102:2655–62. [14] Ramji S, Latha G, Rajendran V, Ramakrishnan S. Wind dependence of ambient noise in shallow waters of Bay of Bengal. Appl Acoust 2008;69:1294–8. [15] Das A. Shallow ambient noise variability due to distant shipping and tide. Appl Acoust 2011;72:660–4. [16] Sanjana MC, Latha G, Mahanty MM. Seabed characteristics from ambient noise at three shallow water sites in Northern Indian Ocean. J Acoust Soc Am 2013;34(4):EL 366–72. [17] Jensen FB, Kuperman WA, Porter MB, Schmidt H. Computational ocean acoustics. 2nd ed. New York: Springer; 2011. [18] Ainslie MA. Effect of wind generated bubbles on fixed range acoustic attenuation in shallow water at 1–4 kHz. J Acoust Soc Am 2005;118:3513–23. [19] Barclay DR, Buckingham MJ. Depth dependence of wind driven broadband ambient noise in the Philippine Sea. J Acoust Soc Am 2013;133:62–71. [20] Kuperman WA, Ferla MC. A shallow water experiment to determine the source spectrum level of wind generated noise. J Acoust Soc Am 1985;77:2067–73.
106
S. Najeem et al. / Applied Acoustics 89 (2015) 101–106
[21] Ainslie MA. Principles of sonar performance modelling. London: Springer; 2010. [22] Hamilton ED. Geoacoustic modeling of the seafloor. J Acoust Soc Am 1980;68:1313–40.
[23] Harrison CH. Sub-bottom profiling using ocean ambient noise. J Acoust Soc Am 2004;115:1505–15. [24] Siderius M, Harrison CH. High-frequency geoacoustic inversion of ambient noise data using short arrays. AIP Conf Proc 2004;728:22–31.