Ocean Engineering 138 (2017) 170–178
Contents lists available at ScienceDirect
Ocean Engineering journal homepage: www.elsevier.com/locate/oceaneng
Comprehensive correlation of ocean ambient noise with sea surface parameters
MARK
⁎
Piyush Asolkara, , Arnab Dasb, Suhas Gajrea, Yashwant Joshia a b
Department of Electronics and Telecommunication Engineering, Shri Guru Gobind Singhji Institute of Engineering and Technology, Nanded, India Maritime Research Center, Indian Maritime Foundation, Pune, India
A R T I C L E I N F O
A BS T RAC T
Keywords: Ocean ambient noise Wind Noise Wind speed Wave height Sea surface temperature Correlation of ambient noise
Ambient noise variability is a critical aspect for the sub-optimal performance of sonar systems. The sea surface parameters (SSP) like wind speed, sea surface temperature and wave height are known to be dominant factors responsible for ambient noise levels and their variability is a major challenge in designing noise mitigation algorithms. Variability across three regions namely tropical, temperate and polar region is significant and quantification of these random fluctuations could potentially facilitate signal processing algorithms to enhance signal to noise ratio. This work presents comprehensive statistical analysis of the spatio-temporal variations of these SSP across three regions and correlates their impact on ambient noise in the specific region. Fluctuations in the tropical region are significant and do justify the challenges for effective sonar design in the region. Analysis has been done on SSP data available in open source collected by ocean observatories deployed at designated sites. Real experimental ambient noise data recordings in tropical littoral waters of the west coast of India and open source acoustic data from other regions have been used to validate the proposed analysis. Understanding of the statistical variability of SSP and its correlation with ambient noise may improve modeling efforts and design generalized mitigation strategy.
1. Introduction The sea surface parameters change drastically from the poles to the equator (Nasa, 2017), which has a significant impact on underwater acoustics and the ambient noise in the region (Nicholas et al., 2002; Ainsley, 2010). Ambient noise variability due to spatiotemporal variations of sea surface parameter is a challenge for ensuring optimum performance of underwater systems at the deployment locations (Knudsen et al., 1948). These site-specific fluctuations of ocean parameters in the tropical, temperate and polar ocean restrict the possibility of generalized strategy to mitigate the ambient noise impact. The tropical shallow waters present even higher ambient noise variability due to significantly high diurnal, seasonal and site-specific fluctuations of Sea Surface Parameters (SSP), namely Sea Surface Temperature (SST), wind speed and wave height. Since the pioneering work by Knudsen et al. (1948) various studies have been proposed representing ambient noise variation triggered by natural physical processes, marine life and human activities (Wenz, 1971; Urick, 1983; Harland et al., 2005; Harland and Richards, 2006). Beyond this several inverse sensing studies have been proposed presenting estimation of environmental conditions such as wind speed,
⁎
Corresponding author. E-mail addresses:
[email protected],
[email protected] (P. Asolkar).
http://dx.doi.org/10.1016/j.oceaneng.2017.04.033 Received 30 June 2016; Received in revised form 21 February 2017; Accepted 20 April 2017 Available online 27 April 2017 0029-8018/ © 2017 Elsevier Ltd. All rights reserved.
rain, etc., based on underwater acoustic recordings, indicating strong interaction between surface parameters and ambient noise (Nystuen et al., 2015; Pensieri et al., 2015). In the absence of sound from human activities and marine life, ambient noise levels are mainly generated due to sea surface activities related to wind and wave (Harland et al., 2005, Harland and Richards, 2006). Studies show that wind noise is dominant in the spectral band of 500 Hz to 25 kHz and noise level increases with increasing wind speed and wave height. Detailed studies show, diverse environmental conditions in different regions of the earth that result in deviation of ambient noise level and corresponding spectral characteristics (Harland et al., 2005, Harland and Richards, 2006; Das, 2011; Asolkar et al., 2016; Ramji et al., 2008; Najeem et al., 2015). Hence, it becomes necessary to analyze region specific comparison of ambient noise and SSP, which is critical for any ambient noise estimation and subsequent SNR enhancement efforts. The acoustic data used for analysis had variable sampling frequencies for different data sets with minimum sampling rate of 16384 Hz. Thus, the ambient noise analysis in this work has been limited to 8 kHz. Various studies available based on open source satellite and moored buoy data present distribution of SSP for different regions (Etter, 2009; Asolkar et al., 2016; Wallace et al., 1989; O'Neill et al., 2010; Chu,
Ocean Engineering 138 (2017) 170–178
P. Asolkar et al.
Fig. 1. Data measurement locations. (Star represents Acoustic data location and pentagon represents SSP buoy location.). Note: Map not to scale, for representative purpose only.
et al., 2016). Regional data pertaining to the each of the sea surface parameters recorded at each of the three buoys has been analyzed for distribution fitting to understand the characteristic differences in the three regions. The wind speed and the wave height follows a Weibull distribution with corresponding scale and shape parameters and the SST follows a normal distribution with corresponding mean and standard deviation.
1989; INCOISE, 2017; NOAA, 2017). Analysis of SST shows prominent three bands with maximum temperature in the tropical region and minimum in the polar region. Tropical regions show higher SST time series anomaly which translates to corresponding higher wind speed and wave height anomaly as compare to temperate and polar regions. Various studies have been presented in the literature elaborating the role of wind speed and wave height in noise generation (Marsh, 1963; Piggott, 1964). The SST shows a linear relationship with wind speed (Wallace et al., 1989; O'Neill et al., 2010; Chu, 1989) and hence SST can be analyzed for the early estimation of wind noise. However, the delay in ambient noise and SST varies with regions that need to be accounted for (Asolkar et al., 2016). Wind speed and Wave height anomaly is observed to be directly related to the variation of the ambient noise characteristics and needs to be analyzed regionwise. This work attempts to compare the spatial distribution of ambient noise due to sea surface activities in the three regions to understand the diversity in the global ambient noise characteristics. Spatiotemporal mapping of ambient noise due to the corresponding variation of SSP is a critical challenge which can be achieved based on time series analysis along with statistical comparison based on 3D plots of SSP at different buoy locations. Different descriptive statistics like time series analysis and probability density function (pdf) have been used to present the characteristic differences between SSP for the three regions. The impact of the SSP on ambient noise has been discussed in order to understand the ambient noise generation process and the possibility of precise estimation using measurable surface parameters. Real ambient noise recordings from the three regions have been used to analyze the ambient noise variations in the three regions. Descriptive statistics based on box plot and pdf analysis has been used to present the variations in the ambient noise spectral characteristics for the three regions. Breaking waves generated by wind and wave interaction are reported to be the dominant mechanism for the generation of ambient noise, although some researchers do admit that this mechanism is not fully understood (Harland and Richards, 2006). The analysis of in-situ SSP and real ambient noise data show a wide variation in SSP and ambient noise statistics of the three regions. The study shows distinct ambient noise characteristics in the tropical region with relatively high variation of standard deviation and kurtosis corresponding to higher SST, wind speed, and wave height anomaly. The ambient noise generation process, right from the SSP, if modeled accurately can facilitate precise and early estimation of the ambient noise and subsequent enhancement of sonar performance (Etter, 2009; Asolkar
2. Measurement locations and methodology Various studies have presented the impact of different SSP on ambient noise (Wenz, 1971; Urick, 1983). In this study, we have focused on wind speed, wave height, and SST to study the correlation of SSP and ambient noise. Knudsen and Urick have presented the impact of wind speed and wave height on the ambient noise level in the shallow waters of the Pacific with the majority of observations in the temperate region (Wenz, 1971; Urick, 1983). Local, regional and geostrophic SSP have a great impact on ambient noise and hence shows variations in ambient noise characteristics. In this work, we have used open source sea surface buoy data from three sites in each region for the analysis of the distribution of sea surface environment. Underwater ambient noise data from three regions has been used for the analysis of regional ambient noise characteristics. The SSP buoys are chosen such that it represents diverse characteristics of the regional environment and represents minimum distance between the acoustic measurement site and the SSP measurement location. The measurement sites are shown in Fig. 1 where, the pentagon represents SSP data location whereas star represents acoustic data location. Hourly data of sea surface parameters, like wind speed, wave height, and SST from each region have been obtained from ocean observatories of the corresponding regions. The SSP data of three buoys from each region for the duration 2010–2013 has been considered for the analysis. The buoys from tropical and temperate region provide hourly data for a whole year whereas polar buoys provide data for three months of a year. The location, data length and inference of the data from each buoy has been presented in Table 1. The missing data samples and recordings with poor quality flags have been interpolated at the time of analysis (INCOISE, 2017; NOAA, 2017; EOL, 2017). Acoustic data used for ambient noise analysis in the tropical region was collected in the Arabian Sea off the coast of Goa. Four ITC 8264 omnidirectional hydrophones have been placed horizontally along the east-west direction with 100 m spacing between the sensors. The
171
Ocean Engineering 138 (2017) 170–178
P. Asolkar et al.
Table 1 Data statistics ($represents data length). SSP
Statistical Measure
Tropical
Temperate
Polar
AD 07 (14°N, 68.88°E) $ 35296
AD 09 (8.24°N, 73.30°E) $ 24534
BD 14 (18.18°N, 85.54°E) $ 34583
44066 (39.57°N, 72.59°W) $ 34984
44011 (41.10°N, 66.62°W) $ 23136
46002 (42.61°N, 130.49°W) $ 27615
48021 (70.35°N, 143.00°W) $ 2571
48211 (70.37°N, 146.04°W) $ 2591
48212 (70.87°N, 150.28°W) $ 2346
SST
Mean S.D Skewness Kurtosis
28.85 1.09 0.21 2.42
28.81 0.83 0.93 3.73
29.46 0.15 0.03 3.02
16.34 3.19 0.14 1.42
11.44 2.0411 1.03 5.58
12.66 6.05 −0.20 2.10
10.09 6.21 0.36 2.69
0.62 1.97 1.70 6.25
6.98 3.58 0.71 4.32
Wind speed
Mean S.D Skewness Kurtosis
5.01 2.87 0.31 2.24
5.92 2.75 0.24 2.62
6.32 3.04 0.10 2.25
9.20 3.54 −0.07 2.77
6.96 3.78 0.60 3.16
5.53 2.61 0.41 2.75
5.03 2.44 0.71 3.91
4.32 2.41 0.51 2.90
5.14 2.63 0.46 2.97
Wave height
Mean S.D Skewness Kurtosis
1.78 1.21 1.39 4.1620
1.38 1.25 1.15 3.52
1.55 0.94 1.08 3.94
2.63 1.31 0.90 3.65
2.38 1.25 1.20 5.19
2.19 0.94 1.22 4.86
1.65 1.55 2.45 11.42
1.63 0.32 2.46 10.56
1.87 1.52 2.43 12.43
p(f )= 94H 6/ sf −2 Table 2 Specification of Sensors (Das, 2011). Specification
Description
Hydrophone type Sensitivity Bandwidth Beam pattern
ITC 8264 −175 dB re 1 μPa 10 Hz–100 kHz Horizontal Omni-directional, +2 dB Vertical Omni-directional, +2 dB for upper hemisphere
Gain
−6 dB to +90 dB in remotely controlled steps of 6 dB
Anti-Aliase Pass Band Sample rate Resolution
0–100 kHz 256 kHz 16 bits
= 2.9 H 6/ sf −5/3
, 1 ≤ f ≤ 13.5 Hz , f ≥ 13.5 Hz
(1)
where, f is frequency and s is salinity of sea water. Further studies show a better correlation of ambient noise with wind speed as compared to wave height. Piggott (1964) has presented the log relationship between wind speed (v) and ambient noise level (NL) as in Eq. (2).
NL = B(f ) + 20*n*log10(v )
(2)
where, B(f) is noise level at frequency f for wind speed of 1 knot and n is an empirical coefficient. 3. Data analysis The analysis has been carried out to study the correlation of ambient noise with SSP which will be essential for designing generalized ambient noise mitigation strategies across all the three regions. In this study, we have used univariate descriptive statistics for the characterization of sea surface parameters and ambient noise (Goos and Meintrup, 2015). Time series analysis is used to compare the seasonal variation and basic statistics of SSP for different regions. Parametric methods have been used to study mean, standard deviation, skewness, and kurtosis of the signal for the fitted distribution. Mean describes the average value whereas deviation gives heterogeneity of the data. Skewness is used to represent tailedness of the distribution and kurtosis represents peakedness which gives information about outliers. Mean and standard deviation can be analyzed along with skewness or kurtosis to represent overall characteristics of the data. A parametric method along with box plot representation is used for the characterization of ambient noise in the three regions. Box plot is a graphical representation of data through their quartiles. It gives us spread, the mean, median and interquartile range of the data across the three regions (Goos and Meintrup, 2015). The time series data has been used to analyze seasonal variations and anomaly of the SSPs. Fig. 2, presents the three SSP data across the year and the regions with the x-axis showing the day of the year and yaxis showing the amplitude of the specific parameter. The seasonal variations across the year can be picked up from the plot based on the amplitude variation. Fig. 2(a) represents the time series plot of SST for the three regions. Three regions represent completely distinct characteristics for the variation of SST. SST is highest in the tropical region (shown in red) and least in polar region (shown in black). The diurnal
specifications of the sensors used for measurement are given in Table 2. Measurements were done at a depth of 30 m hourly from 09:30 to 17:30 (GMT +5:30) over a period of two months (Das, 2011; Asolkar et al., 2016). Some of the acoustic data records show dominant shipping noise and such records have been discarded for the analysis. The acoustic data used for analysis in the temperate region is passive acoustic recordings from Icy Bay, Alaska and is obtained from the Earth Observing Laboratory (EOL) (EOL, 2017). It was collected from three deployment sites at a depth of 90 m for first and second sites and 75 m for the third deployment site. The acoustic data used for the polar region is a passive acoustic data from the Alaskan Beaufort Sea provided by EOL (EOL, 2017). SST shows distinct characteristics in the three regions. The tropical region shows higher SST in the range of 20–35 °C, whereas temperate region shows moderate SST in the range of 10–20 °C and is minimum in the polar region which is less than 10 °C. Such distinct characteristics are not observed for wind speed and wave height. Wind speed shows a wide range in the temperate region from 0.5 to 20 m/s, whereas its moderate in polar region in the range of 0.5–15 m/s and is minimum in the temperate region which is less than 12 m/s. Wave height ranges are similar in the temperate and polar region representing maximum wave height up to 8 m, whereas tropical region shows maximum wave height up to 6 m. Out of these SSP, wind speed and wave height have a direct impact on ambient noise characteristics as seen in Eqs. (1) and (2) Whereas, SST has an indirect effect. Marsh (1963) has presented the relationship between noise pressure (p) and wave height (H) given by
172
Ocean Engineering 138 (2017) 170–178
P. Asolkar et al.
Fig. 2. Time series plots of sea surface parameters.
height is highest for the temperate region but time series anomaly is higher for the tropical region. Wind speed and wave height shows higher seasonal variation and can be clearly seen from the corresponding seasonal peaks in the time series plot. The statistical analysis shows the fitted distribution to SSP and ambient noise level, which gives details like spread and mode of SSPs. Fig. 3(a) represents the normal distribution fitted to SST for the three regions. Tropical SST pdf shows highest mean and least standard deviation, however, polar SST pdf shows least mean and highest standard deviation. This study based on moored buoy data
time series anomaly is observed to be maximum for the tropical region, whereas seasonal anomaly is maximum for the temperate region. Fig. 2(b) represents time series variations of wind speed. Average wind speed for the temperate region (shown in blue) is higher as compared to the tropical and polar region. The wind speed anomalies are highest for the tropical region and least for the polar region, represented by rapid amplitude fluctuation across days of the year. This is mainly due to increase in the variance of atmospheric pressure and Coriolis force from the equator to poles. Wave height is highly correlated to wind speed and shows similar variations as shown in Fig. 2(c). Average wave
173
Ocean Engineering 138 (2017) 170–178
P. Asolkar et al.
Fig. 3. Normalized Histogram (NH) and corresponding pdf fit plots for SSP.
174
Ocean Engineering 138 (2017) 170–178
P. Asolkar et al.
Fig. 4. Box plot for the Power Spectral Density (PSD) of the ambient noise.
the wind speed pdf characteristics. The temperate region shows higher skewness and standard deviation as compare to tropical and polar regions. Power Spectral Density (PSD) of 90 real ambient noise samples from each of the three regions is estimated based on the Welch method of moving average periodogram (Welch, 1967). To compare ambient noise characteristics of the three regions, ambient noise levels are analyzed at a depth of 30 m. Roger's model (Rogers, 1981) has been
follows the analysis by Zhang and Zhou (2010). Fig. 3(b) represents the Weibull distribution fitted to wind speed for three regions. The temperate region shows the higher mean and standard deviation of wind speed with minimum kurtosis. Tropical and polar wind speed pdf is observed to be close to each other. This analysis from moored buoy data follows the analysis by Chu (1989), Monahan (2006). Fig. 3(c) represents Weibull distribution fitted for wave height. The tropical and polar region shows similar wave height pdf and follows
175
Ocean Engineering 138 (2017) 170–178
P. Asolkar et al.
Fig. 5. Normalized Histogram (NH) and pdf fit plots for ambient noise level.
The 3D plot has been used to represent the characteristic difference between the three regions. Bulging for the data in a region represents the low regional spatial variation of the SSP and vice versa. Three different regional groupings in the plot represent distinct characteristic features of the region and can be further used for analyzing the regional noise characteristics. Mean and standard deviation along with either kurtosis or skewness represents the complete statistical characteristics including details of the outliers. Fig. 6(a) and (b) represents 3D plot for SST parameter; it shows completely different groupings for the three regions. SST shows major differences in mean and standard deviation where as kurtosis and skewness shows minimum variations for the three regions. Time series anomaly and standard deviation of SST are important parameters affecting wind speed characteristics of the region. Time series anomaly has a direct impact on wind speed where as the deviation of SST impacts standard deviation of wind speed. Fig. 6(c) and (d) represents the 3D plot for wind speed and it shows relatively wide spread in the standard deviation for the tropical region representing uncertainties for the wind speed variations in the region. Kurtosis and skewness of SSP for the temperate region show higher variation as compare to other regions. Wave height follows wind speed characteristics and clearly represents different characteristics for the three regions as shown in Fig. 6(e) and (f). The SSP translates these variations to the ambient noise characteristics which can be further analyzed based on skewness, kurtosis and standard deviation of ambient noise. Fig. 6(g) and (h) represent the 3D plot for ambient noise. These plots show good bulging for the temperate and polar region; however tropical region shows spreading due to the wide variation of standard deviation. The PSD analysis from Fig. 4(a) represents a wide spread of noise levels for the frequencies in the band of 100 Hz–8 kHz, which is mainly due to the wind noise and sea surface agitation. Hence, this higher variation in ambient noise in that frequency band can be attributed to the variation of SSP. From the analysis of Fig. 6, the spreading of ambient noise in the tropical region makes it distinct from other regions attributed to the time series anomaly for the wind speed and wave height rather than the amplitude values of the SSP. The wide variation of standard deviation and kurtosis is the critical challenge for noise estimation and mitigation in the tropical region. This indicates the need for correlation analysis of the multiple SSP with the ambient noise and validation of the underwater acoustic models and techniques in the tropical region.
used to compensate the effect of depth in temperate and polar acoustic data and no other processing has been done. Noise levels at each frequency are presented with box plot representation in Fig. 4. Box plot representation at each frequency enables us to understand the spread, quantiles and median of noise levels considering all the recordings (OSE, 2017). Fig. 4(a) represents the box plot for the PSD of the tropical region. It is evident from the plot that box plot spread at a frequency is significantly high ranging up to 40 dB. Fig. 4(b), presents the box plot representation of PSD in the temperate region representing relatively higher noise levels. However, the spread is not significant. The polar region shows both low noise levels and very low spread as presented in Fig. 4(c). Polar data was measured in the area with rich biodiversity and hence represents traces of biological noise for frequencies lower than 200 Hz. Mean PSD of the noise level is observed to be maximum for the temperate region and minimum for polar region. Fig. 5 represents the generalized extreme value distribution fitted for the ambient noise in three regions. The tropical region shows relatively higher standard deviation with the moderate mean noise level. Polar noise pdf has sharp peak as compare to other regions due to transient noise of biological sources. 4. Results and discussion The known process of typical noise generation as a result of wind activity is attributed to surface waves due to the interaction of the wind with the sea surface. It generates continuous sound between 500 Hz to 25 kHz and can be observed in the Knudsen spectra (Knudsen et al., 1948). Effect of wind is local and earlier studies show that noise level depends on wind speed in the vicinity of the receiver (Wenz, 1971; Urick, 1983). Low wind speed produces bubbles that generate noise in the range of 15 Hz–300 kHz. As wind speed increases, wave height increases and breaking of waves becomes the dominant factor for noise generation (Harland et al., 2005, Harland and Richards, 2006). Wave height is mainly affected by wind speed, fetch of wind and depth of water (Zhang and Zhou, 2010; Monahan, 2006). Hence wind speed and wave height are highly correlated and are collectively represented by Buford sea state. Knudsen and Wenz have effectively analyzed ambient noise with Buford levels (Knudsen et al., 1948; Wenz, 1971; Urick, 1983). Regional spatial variation along with correlation of SSP and ambient noise can be studied based on distributions and statistical characteristics of SSP as shown in Fig. 6.
176
Ocean Engineering 138 (2017) 170–178
P. Asolkar et al.
Fig. 6. SSP and ambient noise characteristics of the three region.
5. Conclusions
optimum performance of the underwater acoustic systems and also limit the possibility of generalized mitigation strategies. Various studies have been proposed representing the relationship of ambient noise with various SSPs like wind speed, wave height, etc. Earlier studies
Spatial and temporal variation of sea surface parameters result in a variation of ambient noise and became a major challenge in ensuring
177
Ocean Engineering 138 (2017) 170–178
P. Asolkar et al.
Goos, P., Meintrup, D., 2015. Statistics with JMP: Graphs Descriptive Statistics and Probability. Wiley. Harland, E.J., Jones, S.A., Clarke T., 2005. SEA 6 technical report: underwater ambient noise. Report by QINETIQ. pp. 1–48. Harland, E.J., Richards, S.D., 2006. SEA 7 technical report: underwater ambient noise. Report by QINETIQ. pp. 1–76. Knudsen, V.O., Alford, R.S., Emling, J.W., 1948. Underwater ambient noise. J. Mar. Res. 7 (410), 429. Marsh, H.W., 1963. Origin of the Knudsen spectra. J. Acoust. Soc. Am. 35, 409–410. Monahan, A.H., 2006. The probability distribution of sea surface wind speeds part I: theory and sea winds observations. J. Clim. 19, 497–520. Najeem, S., Sanjana, M.C., Latha, G., Durai, P., 2015. Wind-induced ambient noise modeling and comparison with field measurements in the Arabian sea. Appl. Acoust. 89, 101–106. Nicholas, G., Pace, Finn, B., 2002. Impact of Littoral Environmental Variability on Acoustic Predictions and Sonar Performance. Springer, Netherlands. Nystuen, J.A., Anagnostou, M.N., Anagnostou, E.N., Papadopoulos, A., 2015. Monitoring Greek seas using passive underwater acoustics. J. Atmos. Ocean. Technol. 32 (2), 334–349. O'Neill, N.W., Chelton, D.B., Esbensen, S.K., 2010. The effects of SST-induced surface wind speed and direction gradients on mid-latitude surface vorticity and divergence. J. Clim. 23, 255–281. Pensieri, S., Bozzano, R., Nystuen, J.A., Anagnostou, E.N., Anagnostou, M.N., Bechini, R., 2015. Underwater acoustic measurements to estimate wind and rainfall in the Mediterranean Sea. Adv. Meteorol., 2015. Piggott, C.L., 1964. Ambient sea noise at low frequencies in shallow water of the Scotian Shelf. J. Acoust. Soc. Am. 36 (11), 2152–2163. Ramji, S., Latha, G., Rajendran, V., Ramakrishnan, S., 2008. Wind dependence of ambient noise in shallow water of Bay of Bengal. Appl. Acoust. 69, 1294–1298. Rogers, P.H., 1981. Onboard prediction of propagation loss in shallow water. Nav. Res. Lab., 1–24. Urick, R.J., 1983. Principles of Underwater Sound 3rd ed.. McGraw-Hill, New York. Wallace, J.M., Mitchell, T.P., Deser, C., 1989. The influence of sea-surface temperature on surface wind in the eastern equatorial pacific: seasonal and inter annual variability. J. Clim. 2 (1492), 1499. Welch, P.D., 1967. The use of fast fourier transform for the estimation of power spectra: a method based on time averaging over short, modified periodograms. IEEE Trans. Audio Electroacoust. 15, 70–73. Wenz, G.M., 1971. Acoustic ambient noise in ocean: spectra and sources. J. Acoust. Soc. Am. 51, 1010–1024. Zhang, W., Zhou, G., 2010. Spatial distribution of global annual average SST variance. In: IITA Proceedings of the International Conference on Geoscience and Remote Sensing. pp. 499–502.
show that wind speed and wave height are directly contributing to wind noise, whereas SST affects the wind and is indirectly correlated to ambient noise. However, these studies lack the comprehensive comparison of SSP and ambient noise and their correlation across the three regions that could significantly enhance our understanding of the region wise variation of ambient noise. The proposed analysis of SSP and ambient noise across the three regions can potentially aid in the effective deployment of underwater systems for operational requirements and possibly generalize mitigation strategies across the regions. The statistical distribution of sea surface parameters like wind speed, wave height, and SST has been analyzed for the three regions which show unique characteristics of the SSP for each region. Time series analysis of SST data shows that SST of the tropical region is higher with minimum deviations; however, the temperate region has a moderate temperature with higher deviation. Wind speed and wave height time series anomaly are observed to be maximum in the tropical region. Distribution analysis shows the higher deviation of SST, wind speed and wave height for the temperate region. 3D plot shows widespread of standard deviation, kurtosis and skewness for wind speed and wave height representing higher variation of SSP in tropical region which can be correlated with corresponding ambient noise variations. Spectral and distribution analysis of real ambient noise data shows the higher deviation of ambient noise levels for the tropical region. The 3D plot shows bulging with wide spread, due to the higher variation of standard deviation, kurtosis and skewness of ambient noise for the tropical region. This explains the relatively higher variation of ambient noise in the region and hence no generalized pattern could be drawn for tropical ambient noise. Acknowledgements The authors thank the Indian National Centre for Ocean Information Services (INCOISE) and National Institute of Ocean Technology (NIOT) Chennai, for providing moored buoy data in Indian Ocean region.
References to a website
EOL, 2017. Earth Observing Laboratory deployments. 〈https://www.eol.ucar.edu/allfield-projects-and-deployments〉. Last access on January 2017.
References
INCOISE, 2017. Indian National Centre for Ocean Information Services data info. 〈http://www.incois.gov.in/portal/datainfo/datainfohome.jsp〉. Last access on January 2017.
Ainsley, M., 2010. Principles of Sonar Performance Modeling. Springer-Verlag Berlin Heidelberg. Asolkar, P.M., Das, A., Gajre, S., Joshi, Y., 2016a. Study of variation in ambient noise with fluctuations of surface parameters for the Indian Ocean region. In: Proceedings of the ICDECT. Springer Singapore. pp. 111–119. Asolkar, P.M., Das, A., Gajre, S., Joshi, Y., 2016. Tropical littoral ambient noise probability density function model based on sea surface temperature. J. Acoust. Soc. Am. 140 (5), 452–457. Chu, P.C., 1989. Relationship between thermally forced wind and SST. PAGEOPH 130, 31–45. Das, A., 2011. Shallow ambient noise variability due to distant shipping noise and tide. Appl. Acoust. 72, 660–664. Etter, P.C., 2009. Review of ocean-acoustic models. In: Proceedings of OCEANS’ 09. Biloxi. pp. 1–6.
Nasa, 2017. Sea surface Climate Variability. 〈http://science.nasa.gov/earth-science/ oceanography/ocean-earth-system/climate-variability/〉. Last access on January 2017. NOAA, 2017. NOAA sea surface data. 〈http://www.ndbc.noaa.gov〉. Last access on January 2017. OSE, 2017. Online Statistics Education: An Interactive Multimedia Course of Study 〈http://onlinestatbook.com/Online_Statistics_Education.pdf〉. Last access on January 2017.
178