split-beam echo sounders is a necessity to accurately calculate fish sizes from
acoustic target strengths. Additionally, backscatter intensity of SSS swaths vary
within surveys, necessitating distinct seabed features present throughout the
survey area that can be used as a reference to calibrate grayscale images.
Variable beam geometry also adds complexity to MBES systems. Given such
wide latitude of calibration scenarios, inter-comparability of acoustic studies
awaits the development of universal calibration standards.
• Characterization of the acoustic signal: Corroboration of acoustic classification
requires specific knowledge of how the acoustic signals are processed. Within
ASCs, the multi-echo classification scheme (i.e., E1 and E2) used by RoxAnn is
relatively straightforward compared to the clustering of proprietary PCAreduced acoustic parameters used by QTC. The different beam geometries of
various MBES systems affect feature extraction and are likewise difficult to
replicate. As with calibration and a need for reference areas, inter-comparability
requires greater uniformity in signal processing.
• Single versus multiple frequencies: Integrating multiple frequencies into single
datasets provides greater scope for classifying seabeds or biological characteristics, as both scattering and volume scattering vary with frequency. Anderson
et al. (2008) suggest incorporating multi-frequency ASC with single frequency
MBES during surveys may be a cost-effective way of improving seabed classification. For mapping habitat use by fishes, species identification will remain a
challenge. Future development of multi-frequency and broadband signal processing may hold promise in classifying species groups or higher levels of
biological organization; however, more work is needed in this area.
• Survey design: The current practice of conducting surveys along systematic line
transects needs to be reconsidered in favor of a strategy that adds nested randomized lines. Conducting an SBES or ASC survey along parallel line transects
introduces bias and error into continuous surfaces obtained from spatial interpolation. Similarly, conducting an MBES survey along bathymetric contours to
minimize variability of backscatter intensity may fail to detect small-scale
spatial variability of seabed features. Survey design is particularly important
when conducting surveys of habitat use by fishes. Interpretation of spatial patterns of fish distributions need to account for a variety of fish behaviors,
including feeding migrations and diurnal/nocturnal activity patterns.
• Design in national habitat programs: Advances described in this and preceding
chapters may suggest that the field of ASC is well matured. On the contrary,
significant improvements can still be made in survey design, efficiency and
accuracy, as well as research and development in sensor technologies. For this
reason, we echo the recommendation by Anderson et al. (2008) that formal
mechanisms be established to integrate acoustic remote sensing research and
development into national classification and mapping programs.
• Defining fish habitats and habitat use: Many SBES and some MBES can
simultaneously acquire water column data for detecting and enumerating fishes.
These synoptic fish and habitat data are elucidating important spatial and
temporal variation in habitat use, and showing that not all habitats (even those
248
G. Foster et al.
acoustic target strengths. Additionally, backscatter intensity of SSS swaths vary
within surveys, necessitating distinct seabed features present throughout the
survey area that can be used as a reference to calibrate grayscale images.
Variable beam geometry also adds complexity to MBES systems. Given such
wide latitude of calibration scenarios, inter-comparability of acoustic studies
awaits the development of universal calibration standards.
• Characterization of the acoustic signal: Corroboration of acoustic classification
requires specific knowledge of how the acoustic signals are processed. Within
ASCs, the multi-echo classification scheme (i.e., E1 and E2) used by RoxAnn is
relatively straightforward compared to the clustering of proprietary PCAreduced acoustic parameters used by QTC. The different beam geometries of
various MBES systems affect feature extraction and are likewise difficult to
replicate. As with calibration and a need for reference areas, inter-comparability
requires greater uniformity in signal processing.
• Single versus multiple frequencies: Integrating multiple frequencies into single
datasets provides greater scope for classifying seabeds or biological characteristics, as both scattering and volume scattering vary with frequency. Anderson
et al. (2008) suggest incorporating multi-frequency ASC with single frequency
MBES during surveys may be a cost-effective way of improving seabed classification. For mapping habitat use by fishes, species identification will remain a
challenge. Future development of multi-frequency and broadband signal processing may hold promise in classifying species groups or higher levels of
biological organization; however, more work is needed in this area.
• Survey design: The current practice of conducting surveys along systematic line
transects needs to be reconsidered in favor of a strategy that adds nested randomized lines. Conducting an SBES or ASC survey along parallel line transects
introduces bias and error into continuous surfaces obtained from spatial interpolation. Similarly, conducting an MBES survey along bathymetric contours to
minimize variability of backscatter intensity may fail to detect small-scale
spatial variability of seabed features. Survey design is particularly important
when conducting surveys of habitat use by fishes. Interpretation of spatial patterns of fish distributions need to account for a variety of fish behaviors,
including feeding migrations and diurnal/nocturnal activity patterns.
• Design in national habitat programs: Advances described in this and preceding
chapters may suggest that the field of ASC is well matured. On the contrary,
significant improvements can still be made in survey design, efficiency and
accuracy, as well as research and development in sensor technologies. For this
reason, we echo the recommendation by Anderson et al. (2008) that formal
mechanisms be established to integrate acoustic remote sensing research and
development into national classification and mapping programs.
• Defining fish habitats and habitat use: Many SBES and some MBES can
simultaneously acquire water column data for detecting and enumerating fishes.
These synoptic fish and habitat data are elucidating important spatial and
temporal variation in habitat use, and showing that not all habitats (even those
248
G. Foster et al.