186 
Fishery Bulletin 119(2-3) 
model types (generalized additive models [GAMs], boosted 
regression trees [BRTs], and random forests [RFs]) in pre- 
dicting seafloor trawlability in the GOA, using the same 3 
seafloor characteristics used in the GLM. Finally, on the 
basis of the performances of the 4 models, we identified 
which of the 3 seafloor characteristics determined with 
multibeam sonar data are the most consistently useful to 
discriminate between trawlable and untrawlable seafloor. 
Materials and methods 
Field methods 
Large-scale acoustic trawl surveys that assess the stock 
of walleye pollock (Gadus chalcogrammus) in the GOA 
are conducted biennially aboard the NOAA Ship Oscar 
Dyson during the summer (e.g., Jones et al.”**). Fine- 
scale multibeam surveys were conducted with a Simrad 
ME70 opportunistically during the evening hours in 2011, 
2013, and 2015. The multibeam surveys followed paral- 
lel transects spaced 2 km (1 nmi) apart, within nearby 
25-km? grid cells. Only one fine-scale multibeam survey 
was conducted within each grid cell. Immediately after 
the multibeam survey, video data were also collected at 
1—5 locations (i.e., camera stations) to characterize the 
extent of trawlable and untrawlable seafloor at areas 
covered during each fine-scale multibeam survey (Fig. 1). 
Selection of the locations of camera stations was often 
dependent on wind and current direction, sea state, and 
placement of other camera stations within the same grid. 
Fine-scale multibeam surveys were conducted in similar 
numbers of trawlable and untrawlable 25-km? grid cells. 
However, the extent of actual trawlable and untrawlable 
areas within a grid cell was unknown a priori, ultimately 
resulting in an unbalanced number of trawlable versus 
untrawlable camera stations. 
Multibeam surveys The Simrad ME70 was configured 
with 31 symmetrical split beams, with the middle beam 
vertically oriented (i.e., steered at 0°). The beams in this 
configuration ranged from the spherical 2.8° nadir beam 
(0°) operating at 117 kHz to the 2 ellipsoidal 4.5°-along- 
ship-by-11.0°-athwartship beams steered from —66° to 66° 
2 Jones, D. T., P. H. Ressler, S. C. Stienessen, A. L. McCarthy, and 
K. A. Simonsen. 2014. Results of the acoustic-trawl survey of 
walleye pollock (Gadus chalcogrammus) in the Gulf of Alaska, 
June—August 2013 (DY2013-07). AFSC Process. Rep. 2014-16, 
95 p. Alaska Fish. Sci. Cent., Natl. Mar. Fish. Serv., Seattle, WA. 
[Available from website.] 
3 Jones, D. T., S. Stienessen, K. A. Simonsen, and M. A. Guttorm- 
sen. 2015. Results of the acoustic-trawl survey of walleye pollock 
(Gadus chalcogrammus) in the western/central Gulf of Alaska, 
June—August 2011 (DY2011-03). AFSC Process. Rep. 2015-04, 
74 p. Alaska Fish. Sci. Cent., Natl. Mar. Fish. Serv., Seattle, WA. 
[Available from website.] 
4 Jones, D.T.,S. Stienessen, and N. Lauffenburger. 2017. Results of 
the acoustic-trawl survey of walleye pollock (Gadus chalcogram- 
mus) in the Gulf of Alaska, June-August 2015 (DY2015-06). 
AFSC Process. Rep. 2017-03, 102 p. Alaska Fish. Sci. Cent., Natl. 
Mar. Fish. Serv., Seattle, WA. [Available from website.] 
and operating at 75 kHz (Weber et al., 2013; Stienessen 
et al., 2019). A pulse duration of 1.5-ms was used for each 
beam. The sampling rate was synchronized with that of 
the ship’s downward-looking Simrad EK60 scientific echo 
sounder (Kongsberg Maritime AS) to eliminate interfer- 
ence between the 2 instruments, resulting in an effective 
sampling interval of 1.35 s. A 25-mm tungsten carbide 
sphere was used to calibrate each beam by using the stan- 
dard sphere calibration method (Demer et al.”). 
A POS MV V4 system (Applanix, Richmond Hill, Can- 
ada) was used to compensate beam steering for pitch and 
roll of the vessel by exporting dynamic motion and position 
data directly into the Simrad ME70. It was also used to 
georeference the multibeam sonar data. A C-Nav MBX-4 
system (Oceaneering International Inc., Houston, TX) was 
used to apply differential correction data to the POS MV 
to improve position accuracy. Expendable bathythermo- 
graph probes and casts of a conductivity, temperature, and 
depth instrument (SBE 911plus CTD, Sea-Bird Scientific, 
Bellevue, WA) were used to collect water temperature 
and salinity profile data at selected locations throughout 
the study area (Jones et al.”**). Conductivity and water 
temperature data at transducer depth (5 m) were continu- 
ously transmitted to the Simrad ME70 sonar system from 
the ship’s sensors. 
Camera stations Camera stations were sampled during 
nighttime hours for all years, with one camera deployment 
conducted at each station. During camera deployments, 
the ship and camera were allowed to drift over the seafloor 
at a speed of approximately 1 kt (0.53 m/s). The depths of 
camera stations ranged between 75 and 300 m. A camera 
station is defined as the entire area over which a camera 
drifted during a particular deployment, and the location of 
a camera station is the location of the initial deployment. 
Selection of the locations of camera stations was often lim- 
ited by current and wind speed and direction. 
Video images were collected at camera stations by using 
either a single digital camera or a stereo digital camera 
(SDC) system in 2011 and by using only an SDC system in 
2013 and 2015. The digital camera was equipped with one 
digital video recorder and 2 lights placed above the cam- 
era housing (Pirtle et al., 2015). The version of the SDC 
system used in 2011 had 2 Sony TRD-900 progressive scan 
camcorders (Sony Corp., Tokyo, Japan), both with a resolu- 
tion of 1280 by 720 pixels, and 2 lights placed above the 
camera housing (Williams et al., 2010). Both video cam- 
eras resided in an aluminum cage. The version of the SDC 
system used in 2013 and 2015 is described in Rooper et al. 
(2016). It comprised paired machine-vision cameras that 
were spaced approximately 30 cm apart in underwater 
housing and were used to collect synchronized still images 
at a rate of 1 Hz. Lighting for this camera system was pro- 
vided by 4 LED strobe lights. Each deployment of the 
> Demer, D. A., L. Berger, M. Bernasconi, E. Bethke, K. Boswell, 
D. Chu, R. Domokos, A. Dunford, S. Fassler, S. Gauthier, et al. 
2015. Calibration of acoustic instruments. ICES Coop. Res. Rep. 
326, 133 p. [Available from website.] 
