期刊
FISHERIES RESEARCH
卷 191, 期 -, 页码 190-199出版社
ELSEVIER SCIENCE BV
DOI: 10.1016/j.fishres.2017.03.013
关键词
Sonar imaging system; Fish abundance; Image analysis; Automated enumerating; Time series analysis
类别
资金
- North Pacific Research Board [1419]
- Ocean Marine Industries, Inc.
Recent developments in sonar imaging provide an efficient way of obtaining near video quality of frees-wimming fish in hard to reach areas, e.g., permanent structures like docks and piers, and hard to see environments, e.g., highly turbid waters. However, processing large volumes of data output by sonar imaging systems remains a major challenge. In the present study, we developed an automated image processing procedure to process footage recorded for 59 consecutive hours using an Adaptive Resolution Imaging Sonar, ARIS Explorer 3000 (Sound Metrics INC) deployed at a fixed location. Our approach successfully counted large free swimming fish at a precision rate >94% and estimated sample volume with manual and automatic calculations being highly correlated (r = 0.96). An auto-regressive time series model (of the sixth or higher order) with a zero-inflated Poisson distribution was used to estimate local abundance. Fish counts were low to zero during the first 31 h of sampling, and a major influx of fish occurred in the last 28 h. The observed pattern was co-incidental with local weather patterns: intermittent thunderstorms in the first 32 h and relatively calm weather in the last 24 h. Though thunderstorms limited our conclusions on fish-pier usage with tidal and diurnal cycles, it is apparent that weather conditions play a role in pier usage by large fish. Overall, the ARIS3000 deployment, automated data processing and statistical analyses used in this study proved successful in studying fish affiliations with piers in shallow habitats. (C) 2017 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据