SEISMIC SURVEY GREENLAND 2016 Underwater sound propagation for North East Greenland offshore seismic survey. Mark Mikaelsen

Transcription

1 TGS February 2016

2 SEISMIC SURVEY GREENLAND 2016 Underwater sound propagation for North East Greenland offshore seismic survey Mark Mikaelsen

3 PROJECT Seismic Survey Greenland 2016 Underwater sound propagation Project No Document No Version 1 Prepared by MAM Checked by JEK Summary This report presents underwater sound propagation modelling results for seismic survey activities in the offshore north east Greenland in fall 2016, proposed by TGS. Seismic survey activities are performed using airgun arrays that send high pressure waves toward the seabed, with the purpose of determining the geological properties of the different layers of the seabed. These high pressure waves cause sound pressure levels which may injure or disturb nearby marine mammals and fish. Sound propagation modelling was performed to determine the sound pressure levels (SPL) and sound exposure levels (SEL) for a representative 24 hours survey period. Modelling documented in this report, was performed using dbsea by Marshall Day Acoustics, using the dbsearay algorithm. This implementation is range-, depth- and frequency-dependent, and takes the actual bathymetry, sound speed profiles, source directivity and acoustic properties of seabed sediment into account when calculating the sound propagation. All these parameters were included in the modelling. Parameters that had been measured during noise monitoring work completed during seismic survey in the region during 2015, using the same airgun array, were also utilised. These include a sound speed profile, temperature and salinity over depth, as well as general knowledge of the sound propagation in the region. Modelling was performed using available data for all the above mentioned parameters for a representative 24 hour seismic survey line. In line with current regulatory guidance (EAMRA, 2015), results were presented as SPLpp, SPLrms90%, SELSS and SEL24h noise maps as well as maximum and average distances to SEL24h M- weighted noise thresholds relevant to marine species. NIRAS A/S Aaboulevarden Aarhus C, Denmark P: F: E: niras@niras.dk D: M: E: mam@niras.dk

4 CONTENTS 1 Introduction Modelling area Modelling result presentation Modelling distance and frequency considerations Source path (Seismic survey line) Modelling Approach Airgun array Source information Source pressure measurement procedure and implications Underwater sound propagation Modelling methods and implementation Environmental parameters Environmental knowledge/availability of data Background noise Resume chosen modelling parameters Results Distances to thresholds Noise maps Peak-Peak Sound Pressure Level results (SPLpp) RMS90% Sound Pressure level results (SPLrms90%) Sound Exposure level results (SEL) Sound Exposure level results (SEL (Mpw)) Sound Exposure level results (SEL (Mlf)) Discussion References Appendix A: Sound metrics Underwater sound propagation for seismic survey in North East

5 1 INTRODUCTION This report documents the underwater sound propagation modelling performed in connection with the EIA for TGS proposed seismic activities in the north east Greenland waters in the fall, One of the requirements for the EIA is to model the extent of underwater sound exposure, as described in EAMRA, (2015), which states: In order to take account of the area actually ensonified by a seismic survey as well as potential other surveys in the same general area, a model of the expected noise propagation must be included in the submitted EIA/EMA. Seismic surveys are performed using an airgun array that create high-pressure sound waves aimed towards the seabed, in order to analyse the geological properties of the seabed and the layers below. This operation causes high sound pressure levels (SPL) in the surrounding waters and due to the continuous nature of the survey, also high cumulative sound exposure levels (SEL24h). Certain frequency components of the source signal can even be measured hundreds to thousands of km from the survey site. There are many factors that influence how sound propagates within the ocean, among which bathymetry, sound speed profile and acoustic properties of the seabed sediment are important factors. Knowledge about the local conditions were implemented in the model to the extent possible, in order to achieve as accurate results as possible. A description of each factor, and how they are taken into consideration in the modelling, is available in Section 3. The purpose of the modelling described in this report is: To use available knowledge about underwater sound propagation and local environmental conditions to estimate the sound propagation near the seismic survey area. 1

6 2 MODELLING AREA The underwater noise propagation modelling covers a representative possible 24 hour seismic survey line within the survey license area. The modelled survey line is shown in Figure 1. This line was selected by CMACS as one of the closest to the coast and, by nature of running parallel to the coast, potentially ensonifying a relatively large area close to/within coastal protection zones relative to other survey lines which are generally located rather further offshore in the main. Figure 1: Map showing the modelled representative 24 hour seismic survey line in the NEG16 seismic survey. Bathymetry map from (Arndt, J.E. et al. (2015)). 2

7 2.1 Modelling result presentation The EAMRA guidelines, EAMRA, (2015), state the following requirements for documenting the modelling of underwater noise emission from planned seismic surveys: The seismic noise propagation model shall result in project sound levels at different ranges and depths from the airgun array (depths relevant for the species in the area). Noise levels to be presented in the model are peak-to-peak sound pressure levels referenced to 1μPa (peak-to-peak), rms sound pressure levels referenced to 1 μpa (rms measured over 90% of pulse duration, as defined by Malme et al. 1986; Blackwell et al. 2004; Madsen 2005) and in sound exposure levels referenced to 1 μpa2s per pulse. For assessment of cumulative effects the total sound exposure level (across all airgun pulses and all concurrent surveys and activities in the area) per 24 hours shall be presented. Modelling should include all biologically relevant parts of the frequency spectrum, which for seismic surveys means frequencies up to at least 48 khz at ranges out to 20 km, and frequencies up to at least 20 khz at ranges beyond 20 km. Furthermore, EAMRA, (2015) specify requirements for the model used to obtain such results: The model should be based on actual bathymetry, knowledge of sediment properties (to the degree available) and realistic assumptions regarding vertical sound speed profiles and ice cover. The more detailed and current the environmental data are, the better the results of the model. Modelling should not be restricted to the surface layer but extend to at least 1000 m depth or to the seabed. Horizontally, the model should extend to cover all areas exposed to levels likely to affect marine mammals. Based on the above requirements, it was chosen to use the range-dependent model dbsearay, as implemented in the program dbsea version Modelling distance and frequency considerations Modelling distance As cited in section 2.1, modelling should extend to cover areas where marine mammals might be affected by the seismic survey. Based on previous measurements in the area from the 2015 seismic survey measurements in the same area, a maximum modelling distance of 100 kilometres was chosen. Modelling was thus performed for the full 24 hour survey out to a radius of 100 km from the survey. Additionally, in order to show detailed near-source sound levels, a 5 km radius modelling was also performed. 3

8 Frequency Range As specified by EAMRA, (2015), modelling should be based on frequencies up to at least 48 khz within the nearest 20 km and up to at least 20 khz for distances beyond this range. In order to accommodate this requirement, modelling was performed using third octave source levels from 31,5 Hz 50 khz on both 100 km and 5 km radius modelling scenarios. 2.3 Source path (Seismic survey line) It was chosen to model one representative possible 24 hour seismic survey line within the survey area, as indicated on Figure 1, page 2. For the 5 km radius modelling, only the first hour of the survey line was shown. Table 1 gives the start and end coordinates of the seismic survey line for both modelling scenarios. Scenario 5 km radius 100 km radius Eastern [Start/End] Northing [Start/End] Coordinate system (CRS) WSG84, NSIDC Sea Ice Polar Stereographic Table 1: Modelled source coordinates. 3 MODELLING APPROACH This chapter provides a description of the airgun array used in the modelling, as well as sound propagation modelling software and algorithm. 3.1 Airgun array The airgun array type and acoustic specifications was supplied by TGS. The source data used in the modelling reflects the supplied information. This section will only specify the parameters relevant for the underwater sound propagation modelling Source information The information, relevant to underwater sound propagation modelling, was supplied by TGS, and is given in Table 2. Airgun array size Operation Pressure [psi] Peak-peak pressure at 1 m distance [bar-m]* 3350 cubic inches Single pulse every 10 sec bar-m Table 2: Source data supplied by TGS. * The method used to obtain pressure data for the airgun arrays measures the far-field pressure levels and then back-calculating to a point source level at 1 m distance. This method is described in further detail in the next section. 4

9 In addition to the data provided in Table 2, the airgun array time-series for the airgun array was supplied by TGS. This is shown in Figure 2. Figure 2: Time-series for the 3350 cubic inch airgun array. From the supplied time series, the frequency spectrum was obtained through FFT-analysis, see Figure 3. Figure 3: Frequency spectrum for the 3350 cubic inch airgun array. Based on the time-series, the metrics given in Table 3 were calculated by NIRAS. A short description of each metric is given in appendix A. 5

10 Source level for 3350 cubic inch airgun array SPLpp at 1 m distance [db re. 1 µpa] SPLpeak at 1 m distance [db re. 1 µpa] SPLrms90% at 1 m distance [db re. 1 µpa rms] Duration of RMS calculation [s] SEL at 1 m distance [db re. 1 µpa 2 s] per pulse Pulse duration [s] db re. 1 1 m db re. 1 1 m 239 db re. 1 µpa 1 m s db re. 1 µpa 2 1 m s Table 3: Airgun array source level information. From the frequency spectrum of the signal in Figure 3, the 1/3 octave band levels were calculated, and are shown in Figure 4. Figure 4: Airgun array source level shown in 1/3-Octave band levels. The cumulative sound flux density [k Joule/m 2 per 1 m] was also calculated for the 3350 cubic inch airgun array. The results are shown in Figure 5. Figure 5: Cumulative energy flux per pulse for 3350 cubic inch airgun array. 6

11 3.1.2 Source pressure measurement procedure and implications In section 3.1.1, the known parameters for the airgun array were presented, and a number of additional source level representations were calculated. The acoustic source level of the airgun array is based on a back-calculation from a far-field sound pressure measurement. This method of estimating the source level has certain limitations in terms of accuracy. This is explained in detail in Hannay et al., (2010) and Caldwell & Dragoset, (2000), and this section is therefore written based on information presented there. The method of source level estimation by far-field measurement is commonly used to characterise the acoustic output level of airgun arrays, however it has certain disadvantages with regard to accuracy, especially for near-field distances. These disadvantages are briefly described in the following, and the interested reader is referred to Caldwell & Dragoset, (2000) and Hannay et al., (2010) for further details. In Hannay et al., (2010), the near-field inaccuracy of the back-calculation method is described as follows: Far-field source levels do not apply in the near field of the array where pressures of the individual airguns do not add coherently; sound levels in the near field are, in fact, lower than would be calculated from far field estimates which assume coherent summation from all array elements Another factor the far-field measurement does not account for, is the directivity of the airguns. As explained by Hannay et al., (2010), and Caldwell & Dragoset, (2000), far-field measurements of source levels, are done in the vertical direction relative to the airgun array, as this is the direction of interest for the seismic survey operations. The airgun pressure waves are focused downwards, and produce the highest pressure towards the seabed. According to Caldwell & Dragoset, (2000), the horizontal pressure can be up to 20 db lower than the vertical pressure. For accurate near-field source level calculations, as well as accurate far-field sound level calculations in the horizontal direction, it is vital to consider parameters such as airgun directivity and individual airgun impulse responses to account for interference between the airguns. 7

12 TGS supplied directivity information for the airgun array, as shown in Figure 6. This information was included in the modelling for more accurate sound propagation modelling. Figure 6: Airgun array directivity for the 3350 cubic inch airgun array. 3.2 Underwater sound propagation This section is written based on Jensen et al., (2011) chapter 1 and chapter 3 as well as Porter, (2011). This chapter will give a brief introduction to sound propagation in oceans, and the interested reader is referred to Jensen et al., (2011) chapter 1, for a more detailed and thorough explanation of underwater sound propagation theory. In the ocean, the sound pressure level generally decreases with increasing distance from the source. However, many parameters influence the propagation and makes it a complex process. The speed of sound in the ocean, and thus the sound propagation, is a function of first and foremost pressure, salinity and temperature, all of which are dependent on depth and the climate above the ocean and as such are very location dependent. The theory behind the sound propagation is not the topic of this report, however it is worth mentioning one aspect of the sound speed profile importance. Snell s law states that: cos (θ) c = constant 8

13 Where θ is the ray angle, and c is the speed of sound [m/s], thus implying that sound bends toward regions of low sound speed (Jensen et al. (2011). The implications for sound in water are, that sound that enters a low velocity layer in the water column can get trapped there. This results in the sound being able to travel far with very low sound transmission loss. The physical properties of the sea surface and the seabed further affect the sound propagation by reflecting, absorbing and scattering the sound waves. Roughness, density and media sound speed are among the surface/seabed properties that define how the sound propagation is affected by the boundaries. The listed parameters are merely the most important ones. Other parameters include volume attenuation of the water column, which is explained further in Jensen et al., (2011) Modelling methods and implementation Based on the requirements for modelling as set forth in EAMRA, (2015), the ray propagation algorithm dbsearay was chosen due to its ability to calculate high frequency sound propagation efficiently and accurately. Alternative sound propagation models, such as parabolic equation are more computationally demanding at high frequencies and would not be possible given the frequency requirements. The underwater noise program dbsea version from Marshall Day Acoustics was used to perform all modelling. Modelling was approached using as many input parameters obtained during the 2015 seismic survey noise measurements as possible in the modelling. The sound speed profiles obtained were examined and the one obtained the closest to the modelling line was chosen. Furthermore, temperature and salinity information obtained during the measurements was used. Detailed source characteristics and directivity for the specific airgun array were also used Environmental parameters As previously described, there are many parameters that influence the sound propagation in the ocean. Guideline requirements for modelling parameters to be included As cited in section 2.2, EAMRA, (2015) states that the model should include the following parameters: Actual bathymetry Realistic assumptions of the sound speed profile Sediment properties (to the degree available) Realistic assumptions of ice cover Frequencies up to 48 khz 9

14 The dbsearay algorithm in dbsea supports all the above, with the exception of modelling the surface boundary as ice. The ability to model the sound propagation based on the above parameters, is discussed in the following Environmental knowledge/availability of data The availability of each required parameter is discussed in the following. Range dependent bathymetry (Seabed profile) An online freely available published bathymetry map of the seismic survey area was found online by (Arndt, J.E., et al. (2015)). It is a 250 m x 250 m grid of the bathymetry in the north eastern Greenlandic waters. The bathymetry is available in the WSG84 NSIDC Sea Ice Polar Stereographic Coordinate Reference System. The individual maps used in the modelling scenarios were cropped to only include the relevant areas around the survey, however no data was altered in this process. Range dependent sound speed profile (SSP) During TGS seismic survey in north eastern Greenlandic waters in 2015, a number of sound level measurements along with SSP measurements were made. As the modelled survey will take place around the same time of year as the previous seismic survey of 2015, it was chosen to use the sound speed profile obtained closest to the modelled seismic survey line, measured approximately 6 km east of the survey line, see Figure 7. Figure 7: Map showing noise modelling line and location where the used sound speed profile was measured during the NEG15 measurements. 10

15 This sound speed profile obtained at this location is shown in Figure 8. Figure 8: Sound Speed Profile used for Sound propagation modelling. Sediment properties To determine the sediment properties for the seabed, the GEUS maps were studied. Unfortunately, most of the seabed in the survey area is labelled Little known basin with thick sedimentary succession and Area underlain by continental crust. TGS added that a significant part of the area near the modelled seismic survey line is believed to consist of relatively thin glacial sediments on top of Tertiary sediments. Additionally TGS informed that: Density of and velocities of pebbly rocks of glacial origin might vary from Mg m -3 and km s -1 Further detail could not be obtained, and it was therefore chosen to use a thin (5 m thick) top layer on top of a harder sediment as given in Table 4. Sediment cp [m/s] rp [kg/m 3 ] αp [db/lp] Thin Glacial Sediment Tertiary sediment Table 4: Sediment acoustic properties used for modelling. Ice cover A major uncertainty is the ice cover. Experience from previous surveys off NE Greenland suggests that a significant part of the survey area could be covered by ice, and the survey could therefore include an icebreaker vessel. Since the ice cover during the survey period is very dependent on weather and ocean conditions, it is however not possible to accurately predict it during a future survey. 11

16 This presents an issue with regard to modelling accuracy. Since the seismic survey is to be carried out in the fall 2015, the shape and extent of the ice cover is unknown. The ice cover underside shape is of utmost importance due to the scattering effect of ice. Assuming worst case, which would be completely smooth ice, the sound will skip along the underside and continue in the direction away from the source. However a smooth underside is not a very realistic assumption, as ice tends to be rough and have spikes of different length and width. When sound hits such spikes, instead of skipping forward, part of the acoustic energy is scattered in multiple directions, and thus result in an attenuation of the outward propagating sound. It is unknown how big a transmission loss occurs from such scattering, since the underside of the ice cover changes from year to year. There are therefore two options for including ice cover in sound propagation modelling. 1. Make a worst case assumption, that the ice cover underside is smooth, thus scattering is neglected and the calculated levels will be higher than the actual situation. 2. Estimate an RMS roughness of the ice cover underside. This will be a more accurate method of approximating the effect of the ice cover, however if the estimated roughness is set higher than the actual situation, the actual sound levels can be underestimated. It was chosen by CMACS, that a worst case approach should be taken in regard to the ice cover. In case of significant ice cover during the survey, the modelled levels are therefore likely to be overestimated. Volume Attenuation in the water column Another parameter that has influence on especially the high frequency transmission loss over distance is the volume attenuation, defined as an absorption coefficient reliant on chemical conditions of the water column. This parameter has been approximated by: α f2 1 + f f f f 2 (db/km) Where f is the frequency of the wave in khz (Jensen et al., 2011) Background noise There will be several sources of noise not included in the underwater sound propagation modelling. These are: 12

17 - Any biological sources, such as shrimps, whales and other marine mammals. - Noise from ships, both those dragging the airgun array, follower ships etc. - The noise originating from ice breaking by the icebreaker. 3.3 Resume chosen modelling parameters The following parameters were chosen for underwater sound propagation modelling. The reader is referred to the respective previous sections for explanations on choice of parameters. Range dependent sound speed profile measured on site during the 2015 seismic survey sound level measurements. Range dependent bathymetry data from Arndt, J.E. et al. (2015), in a 250 m x 250 m resolution. Seabed sediment is represented by a relatively thin glacial sediment on top of Tertiary sediment. Surface was chosen to be smooth, due to insufficient knowledge of the surface shape and extent of ice cover. Modelling the ice cover as a hard reflective surface, presents the worst case scenario, as backscattering loss is not modelled in this situation. The source is modelled as a point source with directivity applied for all angles and 1/3 octave frequency bands. All modelling took place using frequency information in 1/3 octave bands in the range 31.5 Hz 50 khz. 13

18 4 RESULTS The modelling results are presented by maximum-over-depth colour maps showing the sound level in the required metrics; these are further described in appendix A. 4.1 Distances to thresholds Distances in are given in meters from source location. Rmax indicates the maximum distance at which the sound level can be present in any direction from the source. Rmean indicates the average distance from source at which the sound level can be present. If Rmax for e.g. SEL24h(MPW) = 186 db re. 1 µpa is 250 m, it means that sound levels in excess of 186 db will only occur within 250 m of the source, and that beyond that distance, noise levels will be below 186 db. The distances reflect the entire 24 hour survey line. Cumulative SEL modelling results are given in Table 5. Distance to threshold Distance SEL24h(MLF) = 198 db [re. 1 µpa 2 s] SEL24h(MPW) = 186 db [re. 1 µpa 2 s] Rmax (m) 200 m 2050 m Rmean (m) 180 m 1300 m Table 5: Maximum and average distance from seismic survey line to thresholds by Kyhn. 4.2 Noise maps et. al., (2011). As required by the EAMRA guidelines, the results are also presented in color maps showing the maximum over depth sound levels at range from the source. Maps showing Sound Pressure Level (SPL) results both as SPLpp and SPLrms90%, are given for 3 different source locations. Maps showing Sound Exposure Level (SEL), both M-weighted and unweighted, are given for a 10 waypoint representation of the survey line. This means, for the 100 km radius scenario, that each point represents 2.4 hours of survey (864 airgun array shots pr. position). For the 5 km radius scenario the duration is 1.1 hour of survey (approximately 40 airgun array shots pr. position). 14

19 4.2.1 Peak-Peak Sound Pressure Level results (SPL pp) The following three maps show the SPLpp results of the modelling. Note that the color codes indicate SPLpeak, however the sound levels are in SPLpp. 15

20 16

21 17

22 4.2.2 RMS90% Sound Pressure level results (SPL rms90%) The following three maps show the SPLrms90% results of the modelling in accordance with Malme et al. (1986). 18

23 19

24 20

25 4.2.3 Sound Exposure level results (SEL) The following two maps show the SEL unweighted results for the 5 km and 100 km radius scenarios. The first map shows the 5 km radius, SEL1h scenario results, while the second map shows the 100 km radius, SEL24h scenario results. 21

26 22

27 4.2.4 Sound Exposure level results (SEL (M pw)) The following two maps show the SEL(Mpw)-weighted results for pinnipeds in water for the 5 km and 100 km radius scenarios. The first map shows the 5 km radius, SEL1h(Mpw) scenario results, while the second map shows the 100 km radius, SEL24h(Mpw) scenario results. 23

28 24

29 4.2.5 Sound Exposure level results (SEL (M lf)) The following two maps show the SEL(Mlf)-weighted results for low-frequency cetaceans for the 5 km and 100 km radius scenarios. The first map shows the 5 km radius, SEL1h(Mlf) scenario results, while the second map shows the 100 km radius, SEL24h(Mlf) scenario results. 25

30 26

31 5 DISCUSSION The underwater sound propagation modelling was based on certain assumptions regarding seabed sediment type and airgun characteristics. This led to a model assuming a slightly cautious model. Just as bathymetry changes over distance, so can seabed sediment layer composition (type, thickness, porosity of the different layers) and this in turn will affect the sound interaction over distance, resulting in actual sound measurements differing from the modelled levels based on the deviation from the modelled conditions. With the lack of detailed information on local variations in seabed conditions, combined with the required computational demands, such modelling is not realistic. In the modelling, a best estimate of the average conditions was made based on the available information. Historically, ice cover has been present to some extent during the seismic surveys in the autumn, with more extensive cover in relatively inshore areas such as that modelled here. Ice cover should therefore to some extent be considered in modelling for this region. A worst case approach was however taken in the modelling for two reasons. First and foremost, modelling with a rough surface would underestimate noise levels over distance if the survey was to occur under ice free conditions or with smooth ice underside. Choosing a smooth underside prevents loss of acoustic energy due to backscatter from the surface, and thus represents the more precautious approach when estimating sound levels. Another reason is that the modelling software used does not allow for local changes to the surface boundary conditions, and including an ice cover for the whole modelling radius would therefore be misleading. 27

32 REFERENCES Arndt, J.E. et al. (2015): Northeast Greenland - Digital Bathymetric Compilation, with links to DBM in ArcGIS format. doi: /pangaea , Supplement to: Arndt, Jan Erik; Jokat, Wilfried; Dorschel, Boris; Myklebust, Reidun; Dowdeswell, Julian A; Evans, Jeffrey (2015): A new bathymetry of the Northeast Greenland continental shelf: Constraints on glacial and other processes. Geochemistry, Geophysics, Geosystems, 16(10), ,doi: /2015GC Caldwell, J., Dragoset, W. (2000): A brief overview of seismic air-gun arrays, The Leading Edge, Aug. 2000, pages EAMRA, 2015: Offshore Seismic Surveys in Greenland: Guidelines to Best Environmental Practices, Environmental Impact Assessments and Environmental Mitigation Assessments, Danish Centre for Energy and Environment (DCE), Greenland Institute of Natural Resources (GINR), Environmental Agency for the Mineral Resource Activities (EAMRA), Greenland, November Hannay, D., Racca, R., MacGillivray, A., (2010), Model Based Assessment Of Underwater Noise from an Airgun Array Soft-Start Operation, JASCO Applied Sciences, Oct Jensen, F.B., Kuperman, W.A., Porter, M.B., Schmidt, H., (2011), Computational Ocean Acoustics, 2 nd edition, Springer, Kyhn, L.A., Boertmann, D., Tougaard, J., Johansen, K., Mosbech, A., (2011): Guidelines to environmental impact assessment of seismic activities in Greenland Waters, 3 rd revised edition, Danish Center for Environment and Energy, Dec Malme CI, B Würsig JE Bird & Tyack P (1986) (published 1988) Behavioral responses of gray whales to industrial noise: feeding observations and predictive modeling. Outer Continental Shelf Environmental Assessment Program, Final Report. BBN Rep OCS Study MMS Prepared by BBN Labs Inc., Cambridge, MA, for NMFS and MMS, Anchorage, AK. 28

33 APPENDIX A: SOUND METRICS Underwater sound levels are measured in db re. 1 µpa. Different methods of representing the sound level exist to characterize the intensity, exposure level, or even max levels. Depending on the intended use of the results, and the type of source, it can be useful to use one sound level representation over another. For impulsive sound sources, such as an airgun array, the four most commonly used ones are: 1. The sound pressure level peak-peak (SPLpp) and zero-peak (SPLpeak) 2. The root-mean-square sound pressure level (SPL90%rms) 3. The sound exposure level (SEL) 4. The cumulative energy flux These four metrics are briefly explained in the following SPL peak and SPL pp The SPLpeak is the maximum instantaneous sound pressure level of an impulse p(t), given by: SPL peak = 20 log 10 (max p(t) ) The closely related SPLpp is the maximum difference in sound pressure level of an impulse p(t), given by: SPL pp = 20 log 10 (max(p(t)) + min (p(t)) ) which is also the metric used for the modelling in this project. SPL 90%rms The SPL90%rms is the root-mean-square pressure level over a time window, T, containing the impulse p(t): SPL 90% rms = 10 log 10 ( 1 T p2 (t)dt) T The SPL90%rms is defined as the mean value of a pulse with the time window T containing 90% of the pulse energy as described in [Malme et al. 1986]. As a result of dividing by the time window T in the equation, pulses with the energy 29

34 spread out over a long duration will have a lower SPL90%rms than a short duration pulse with the same total energy. It is therefore a useful metric to describe the impulsivity of a source. SEL The SEL, also known as the sound exposure level is defined as the time-integral of the square pressure over a time window T covering the entire pulse duration, and is given by: SEL = 10 log 10 ( p 2 (t)dt) T To accurately model cumulative effects for seismic surveys, it is necessary to use a moving source that approximates the different positions of the airgun arrays throughout a 24 hour modelling period. M-weighted sound exposure level (SEL(MPW), SEL(MLF)) The M-weighted SEL adapts the SEL modelling to reflect the hearing of a certain species or group of species with similar hearing ability. M-weighting functions can be thought of as the waters counterpart to the A-weighting function which is often used to represent the hearing of humans in air. These weighting functions take into account the nonlinear hearing of the species by a set of correction coefficients at each frequency. Thus, the results represent what the species will actually hear when exposed to a certain noise. The M- weighting functions are therefore very useful when determining the behavioural responses of marine mammals to any noise. It was chosen by CMACS Ltd. to use the M-weighted SEL24h metric to present the results, for low-frequency cetaceans, SELC(MLF), and for pinnipeds in water, SELC(MPW). The results will reflect the distances to the thresholds set in Kyhn et al., (2011). Cumulative Energy Flux The cumulative energy flux is a standard measure for airgun arrays. The power spectrum is integrated, and the result is shown with increasing frequency. 30