FY 1996

Monitoring whales in the eastern north Pacific Ocean using Navy hydrophone arrays: A pilot study

Fox, C., K. Stafford, H. Braham, M. Dahlheim, D. Bain, and S. Moore

In Final Report to Space and Naval Warfare Systems Command, Integrated Underwater Surveillance System (IUSS) Pilot Study, 24 May 1996, 48 pp. + appendices (1996)

Executive Summary

In late August 1991, NMML and PMEL entered into a collaborative study to investigate the utility of the Navy fixed hydrophone arrays for detecting whales by passive reception of their calls in the eastern North Pacific. Data from five SOSUS (SOund Surveillance System) hydrophones were made available to NMML, via PMEL and Whidbey Island NOPF. Due to high data rates, a pseudo-random sampling strategy was adopted to extract portions of data for marine mammal call analysis. Between September 1991-92, a total of 18,720 10-minute tape segments was analyzed, comprised of 3,744 segments, from each of five SOSUS sites. Seven signal types were defined, based on their frequency and temporal characteristics. Although signal origin could not be determined with certainty, four signal types were classified as whale calls based on their similarity to published spectrograms of calls where species identification was certain. Of the four signals classified as whale calls, two were from blue whales, one was from fin whales and one was from sperm whales. One signal was classified as whale-like, but could not be assigned to species. Two signals were classified as possibly of geologic origin because of their similarity to sounds ascribed to volcanic activity in the South Pacific. These signals dominated samples at four of five SOSUS sites during the NMML/PMEL pilot study.

Once signal types were established, they were recognized and counted automatically via an algorithm developed using SIGNAL acoustic analysis software. There were 59,759 signals counted during the 12-month pilot study: 10,095 (17%) attributed to whales, and 49,664 (83%) provisionally classified as "geologic signals." Unfortunately, the automation of signal counts appears to have resulted in some sounds being inadvertently placed in the wrong categories. This seems to have occurred most often between three signal types (Constant Frequency 20, Harmonic 20 and Upsweep 20) and "miscellaneous signals," which exhibited frequency and duration parameters that were sometimes difficult to distinguish from each other. Conversely, signals believed to be from fin whales (Pulse 20) exhibit distinct frequency and duration parameters, so counts of this signal type are likely more reliable. Further refinement of the algorithm that recognizes signal types, or greater manual verification of signals, could resolve this problem. Overall, the integrity of the signal types seems secure, but the specific counts for each type likely contain some error.

There was demonstrable geographic and seasonal variation in the number and types of signals received at the five SOSUS sites. The number of whale calls differed by as much as a factor of three among the five sites. Although calls attributed to blue, fin and sperm whales were received at each SOSUS site, their ratios varied among sites. There was a marked seasonal pattern to whale call reception at each site. Blue whale calls were common at hydrophones 1 and 3, with a bi-modal pattern of occurrence that may be associated with spring and fall migration patterns. Fin whale calls were most prevalent at hydrophone 3, with continuous occurrence from July through February. Fin whale calls were most common at hydrophone 7 during summer, and during autumn at hydrophones 5 and 9. Sperm whale calls were nearly four times more prevalent at hydrophone 9 than at any other hydrophone; sperm whale clicks were received in comparatively low numbers throughout the year at the other hydrophones.

The NMML/PMEL pilot study demonstrated that it is possible to efficiently sample the large quantity of acoustic data produced by SOSUS hydrophone arrays. A balanced sampling strategy was developed that supported the analysis of a large number of signals for determination of variation in signal counts over broad geographic and temporal scales. Several advantages and limitations unique to the use of SOSUS for monitoring whales were identified during the pilot study. Probably the greatest advantage of SOSUS is its broad coverage, both geographically and temporally. Its greatest limitation is its focus on low frequencies, which essentially eliminates its utility for monitoring all but one species (sperm whales) of odontocetes. Potentially, NMML could use SOSUS data over the long term to investigate habitat use, relative abundance of calls and their variability, and temporal patterns of call reception. This would require additional research to better define call repertoires for mysticete whales in the eastern North Pacific and additional efforts to automate and validate the data analysis protocol. Ultimately, SOSUS can become a unique and cost-effective tool for monitoring mysticete whale populations and providing inferences on population size and habitat use patterns.