Wisniewska, D.M., Kyhn, L.A., Tougaard, J., Simon, M., Lin, Y.-T., Newhall, A., Beedholm, K., Lynch, J. & Madsen, P.T. 2014. Propagation of airgun pulses in Baffin Bay 2012. Aarhus University, DCE – Danish Centre for Environment and Energy, 150 pp. Scientific Report from DCE – Danish Centre for Environment and Energy No. 109 http://dce2.au.dk/pub/SR109.pdf
In 2012, four simultaneous seismic surveys and twelve shallow core drillings were planned to be carried out in Baffin Bay, northwest Greenland. In Greenland, guidelines for Environmental Impact Assessments (EIA) of seismic activities require that each participating company models its own expected noise emission, as well as the cumulative noise levels resulting from all concurrent activities in a given area. This is required in order to evaluate the potential effects of hydrocarbon exploration on marine life under the assumption that all planned activities are carried out. Model precision is very important as it is used to predict received noise levels at various ranges from the source. Since these received levels are used to predict potential effects on marine life and ultimately to evaluate whether a given project may be carried out, it is essential to be able to trust the results of predictive modeling. However, one of the important unknown factors in predictive noise modeling is the propagation of airgun pulses. On top of this, the environmental factors influencing propagation, including bottom substrate, bathymetry, salinity and temperature profiles, were poorly known for Baffin Bay. It was therefore decided that a large scale acoustic monitoring study should document the noise levels from the four planned seismic surveys. The measurements were also to be used to 1) verify the validity of the sound propagation modeling studies commissioned by the hydrocarbon companies for the purpose of the EIA, and to 2) obtain environmental data to feed into, and acoustic data to validate, an advanced sound propagation model developed by collaborators at Woods Hole Oceanographic Institution (WHOI). Furthermore, this study substituted the requirement for the companies to document their noise emission during their seismic activities.
The study was carried out by deploying 21 acoustic dataloggers at seven moored stations, each with dataloggers distributed at three different depths. Additionally, each mooring was equipped with a number of sensors, which recorded depth and temperature over the course of the deployment period. CTD measurements were taken at each station at deployment and retrieval to obtain valid environmental data for the post-season modeling. In addition to the moored stations, close-up recordings of one of the seismic airgun arrays were conducted at ranges of 0.5, 1, 2, 4 and 8 nautical miles to characterize the signature of the array at short ranges. Data from this study were supplemented with data collected by JASCO for Shell before and during the seismic program. Shell’s data comprised acoustic recordings collected with moored dataloggers, and CTD data obtained at a number of locations around Baffin Bay throughout the seismic season.
19 acoustic dataloggers were successfully retrieved from Baffin Bay following the 2012 seismic season. The most apparent contribution of the seismic activities to the noise budget in Baffin Bay and Melville Bay were stepwise increases in noise levels, at the onsets of the four seismic surveys. On a minute by minute basis, on several occasions the sound exposure levels (SEL calculated over 1 minute) increased by more than 60 dB in relation to the pre-exposure background noise SEL of about 120 dB re 1 µPa2 s. The SEL was on average approximately 20 dB higher than the pre-exposure level. Cumulative SEL (cSEL) over 24 hours increased from about 153 dB re 1 µPa2s to around 170-180 dB re 1 µPa2s and at times up to 189 dB re 1 µPa2s.
During a seismic survey, the background noise level was constantly elevated, as one airgun pulse would not fade out before the arrival of the next pulse. Furthermore, with several concurrent seismic surveys undertaken in the same area, multiple pulses were constantly apparent at various levels. This general rise in background noise may cause the airgun sounds to mask other sounds in the frequency range of 1-10 kHz, including sounds of importance to marine animals, especially at close ranges to the airgun array.
The airgun signals in the close-up recordings contained significant energy above ambient noise up to, and possibly beyond, 50 kHz at close ranges. This result stresses the importance of including higher frequencies in assessments of potential effects of seismic surveys on marine organisms, especially when considering Odontocetes, which have exceptionally good high-frequency hearing.
The vertical extent of the typical of Arctic waters near-surface low-sound-speed channel was greater than expected. Accordingly, in most recordings, the highest sound levels were consistently recorded on the top dataloggers. These are also the depths at which marine mammals spend a significant amount of time to breathe and socialize. There were exceptions where the highest sound levels were recorded on the middle or bottom datalogger, which may have been caused by shadow and/or convergence zones. This is important to keep in mind when considering zones of impact in relation to mitigation of effects on marine mammals for EIAs.
The measured received sound levels were within the ranges predicted by JASCO in their pre-season modeling included in the EIAs. There were, however some exceptions were the noise levels were higher than predicted: For example the advanced model by WHOI documented a channel of low sound speed running along the slope of a north-south facing deep water area. The low transmission loss herein resulted in higher than expected sound levels to the north of the channel which was confirmed by the acoustic recordings. This channel was not picked up by the predictive modeling, most likely due to the quality of the environmental input data, which generally were inadequate for this region, and the limited number of source locations considered by the model.
Overall, the results of this study however lend weight to the utility of predictive modeling for the purpose of EIAs, and suggest that it is possible, even in some cases with less than ideal input data, to predict noise exposure from multiple seismic surveys with sufficient accuracy to provide a reasonable basis for assessing the potential impact on animals.
The fully 3D transmission loss modeling performed by WHOI demonstrated that the acoustic environment of the north-eastern Baffin Bay and Melville Bay is highly complex. In particular, the near surface low–sound-speed ducting, the bottom geoacoustic properties, and the detailed bathymetry in shallow and high gradient regions, could produce large effects on the transmission loss. Wave and ice conditions were not factored into the models, partly due to lack of good input data, but also because these effects are not yet well represented in parabolic equation models. This stresses the need for collection and dissemination of high-quality data on hydrography, bathymetry and sediment properties as well as statistics for ice coverage and surface roughness (waves) prior to impact assessment procedures.
