In:
The Journal of the Acoustical Society of America, Acoustical Society of America (ASA), Vol. 144, No. 3_Supplement ( 2018-09-01), p. 1777-1777
Abstract:
Acoustic data collection is a cost-effective approach to evaluating activity patterns of otherwise challenging-to-study offshore cetaceans. However, manual analysis of acoustic data is time-consuming and impractical for large data sets. This study evaluates diel and seasonal patterns in Pacific white-sided dolphin communication through automated analysis of 20 months of continuous acoustic data collected from the Barkley Canyon node of Ocean Networks Canada’s NEPTUNE observatory, offshore of Vancouver Island, British Columbia, Canada. In this study, cetacean signals are manually annotated in a sub-set of the data, and 94 time and frequency features of these and other sounds are extracted and used to train random forest classifiers targeting Pacific white-sided dolphin pulsed calls. The performance of binary and multiclass models with various forest sizes, minimum leaf sizes, and confidence thresholds for Pacific white-sided dolphin classification are compared through nested 10-fold cross-validation to select the best model. Vocalizations are classified with the resultant classifier, manually verified, and examined for seasonal and diel patterns. Pacific white-sided dolphins are shown to be vocally active during dawn and day in spring and summer and at dusk and night year-round with reduced overall activity in fall and winter, suggesting both consistent nocturnal and migratory diurnal populations.
Type of Medium:
Online Resource
ISSN:
0001-4966
,
1520-8524
Language:
English
Publisher:
Acoustical Society of America (ASA)
Publication Date:
2018
detail.hit.zdb_id:
1461063-2
Permalink