data processing and analysis: literature...
TRANSCRIPT
Data Processing and Analysis: Literature Update
FISH 538 Emma DeWitt Cotter
Calibration Results
Biological Samples
Known Target Strength Curves Hydrographic
data
Geography of Site
Survey Track
Acoustic Data
Abundance Estimations
Echo data Abundance estimations for a site
Big Changes in the Last 10 Years
• More data! • Longer deployments enabled by ocean observatories • Use of multi-frequency and broadband devices
• Faster computers
• Mean Volume Backscatter (MVBS) • More accurate, objective information about more species using
multiple frequencies • More widely used for classification
Classification of Coexisting Species
Kang M., Furusawa, M., and Miyashita, M. (2002). Effective and accurate use of difference in mean volume backscattering strength to identify fish and plankton. J. Mar. Sci. 59: 794-804.
Jech, J.M. and Michaels W.L. (2006). A multifrequency method to classify and evaluate fisheries acoustics data. Can. J. Fish. Aquat. Sci. 63: 2225-2235.
Classification of Coexisting Species
1 – 18 kHz 2 – 28 kHz 3 – 120 kHz
Classification of Coexisting Species
Sato, M., Horne, J., Parker-Stetter, S., & Keister, J. (2015). Synthetic echograms generated from the relative frequency response. Fisheries Research, (172), 130-136.
Automated Target Classification
• Applies an established statistical method (probabilistic clustering) to acoustic data
• Widespread uses including insurance, biology, and meteorology
• Determine the intrinsic grouping in a set of unlabeled data
Figure courtesy of:: http://home.deib.polimi.it/matteucc/Clustering/tutorial_html/
Anderson, C.I.H., Horne, J.K., and Boyle, J. (2007). Classifying multi-frequency fisheries acoustic data using a robust probabilistic classification technique. J. Acoustical Soc. Am, 121 (6).
Automated Target Classification
• Each sample is assigned probabilities of membership to each cluster group
• Determined optimum number of clusters for two data sets • Gulf of Alaska: low-diversity, well known system • Mid-Atlantic Ridge: high diversity, less well known system
• Cluster number dependent on site and survey goals • Able to capture biological features of data
Automated Target Classification
High Sv Low Sv Fish
Saturation
Bottom + Intense Schools
Background Zooplankton
Large Fish
Saturation
Bottom + Intense Schools
Background
Zooplankton Small Fish
Gulf of Alaska
• Mid-Atlantic Ridge • 2 Clusters: Non noise vs intense noise features • 3 Clusters – saturation, bubbles, non-noise features • 13 Clusters – fish tracks, biota, bubbles, saturation, dropped pings
Automated Target Classification
Spatially and Temporally Resolved Data
• First example – Jiang et al 2007 • Net-based studies do not provide high temporal resolution and are difficult to
execute over long periods of time
• ADCP backscatter data used to observe diel and annual trends in migration (1996-2000)
• Abundance data shown as a function of time of year, time of day, and depth
Jiang S, Dickey TD, Steinberg DK, Madin LP (2007). Temporal variability of zooplankton biomass from ADCP backscatter time series data at the Bermuda Testbed Mooring site. Deep-Sea Res I 54: 608−636
Spatially and Temporally Resolved Data
Jiang S, Dickey TD, Steinberg DK, Madin LP (2007). Temporal variability of zooplankton biomass from ADCP backscatter time series data at the Bermuda Testbed Mooring site. Deep-Sea Res I 54: 608−636
Spatially and Temporally Resolved Data
• Sato et al 2013 • VENUS Cabled observatory in British Columbia • 2 years of echosounder data • Observed diel vertical migration of euphausiids (krill)
Sato, M., Dower, J.F., Kunze, E., Dewey, R. (2013). Second-order seasonal variability in diel vertical migration timing of euphausiids in a coastal inlet. Mar. Ec. Progress Ser. 480: 39-56.
Spatially and Temporally Resolved Data
Sato, M., Dower, J.F., Kunze, E., Dewey, R. (2013). Second-order seasonal variability in diel vertical migration timing of euphausiids in a coastal inlet. Mar. Ec. Progress Ser. 480: 39-56.