Abstract
Deepwater Wind South Fork LLC (DWSF) is proposing to install an offshore wind energy facility, the South Fork Wind Farm (SFWF), in its lease area on the Atlantic Outer Continental Shelf. The SFWF will consist of up to 15 wind turbine generators (WTG) and an offshore substation, each of which will be supported by a monopile foundation with a maximum diameter of 10.97 meter (m) (36 feet [ft]).
Underwater noise will be generated during impact pile driving for installing the monopile foundations. The objective of this modeling study was to generate predictions of the mean number of animals that may be exposed to sound levels resulting in injury to or behavioral disruption of marine mammals and sea turtles in the SFWF project area. Acoustic fields produced during impact pile driving of the monopile foundations were modeled (see Denes et al (2018) for acoustic modeling details). The JASCO Animal Simulation Model Including Noise Exposure (JASMINE) was used to integrate the sound fields with species-typical behavior. JASMINE results provide an estimate of the probability of sound exposure, which can be compared to acoustic thresholds and then scaled to predict the mean number of animals expected to receive sound levels that may cause injury or behavioral disruption.
The acoustic thresholds used in this study represented the best available science. For potential injury (Level A) to marine mammal species, the Technical Guidance issued by NOAA (NMFS 2018) was used. For potential behavioral disruption (Level B) of marine mammals, the threshold values currently considered by NMFS were used along with an approach suggested by Wood et al. (2012) that accounts for the hearing range of the animals. For potential effects of sound on sea turtles, the guidelines established by Popper et al. (2014), representing the consensus efforts of a scientific working group, were used as well as those developed by Blackstock et al. (2017).
Cetacean exposure probabilities were scaled using the Duke University Marine Geospatial Ecological Laboratory density models (Roberts et al. 2016), including an updated unpublished model for the North Atlantic right whale (Roberts et al. 2017, Roberts et al. 2018) that incorporates additional sighting data. Sea turtle densities were obtained from the U.S. Navy Operating Area Density Estimate (NODE) database on the Strategic Environmental Research and Development Program Spatial Decision Support System (SERDP-SDSS) portal (DoN 2007, 2012). These numbers were adjusted by the Sea Mammal Research Unit (SMRU 2013), available in the Ocean Biogeographic Information System Spatial Ecological Analysis of Megavertebrate Populations (OBIS-SEAMAP) (Halpin et al. 2009). The density models for scaling the exposure results for marine mammals and sea turtles represent the best available data for the SFWF project area.
The mean number of animals that may be exposed to sounds exceeding acoustic thresholds were calculated for a Maximum Design scenario, sixteen foundations installed in twenty days (one pile installed each day), and a Most Likely scenario, sixteen foundations installed in thirty days (one pile installed every other day). Estimates were generated assuming one monopile foundation is driven in a day and that no concurrent pile driving would occur. Noise mitigation was considered by reducing the predicted sound fields by six and twelve decibels to evaluate the effects of using noise reduction systems such as bubble curtains.
The exposure estimates for the Maximum Design scenario and the Most Likely scenario were found to be similar, indicating little difference in expected impacts if one monopile foundation is installed each day or every other day. The case where one of the foundation piles is difficult to install generally resulted in a small increase in exposure estimates of less than a few percent. The behavioral response of animals avoiding loud sounds (aversion) produced during pile driving was also investigated for North Atlantic right whales, harbor porpoises, and humpback whales. It was found that aversive behavior could result in substantial decreases in the exposure estimates, particularly for Level A exposures (injury). Aversion is thought to be common in marine mammals (Ellison et al. 2012), so the exposure estimates that do not include aversion are likely conservative.