502 
Fishery Bulletin 11 5(4) 
Bottom classification results of the best generalized linear model for 238 acoustic 
data segments that were collected in the Gulf of Alaska and used in the building 
and validation of this model. The segments, each represented by a symbol on the 
map at its collection area, are divided into 4 categories: area correctly classified 
as trawlable (green circles), area correctly classified as untrawlable (red circles), 
untrawlable area incorrectly predicted as trawlable (green x’s), and trawlable area 
incorrectly predicted as untrawlable (red x’s). 
effort for trawlable ground in sampling cells that ul¬ 
timately turn out to be untrawlable. Much of this inef¬ 
ficiency can be eliminated by minimizing the number 
of untrawlable sampling cells that make up the unclas¬ 
sified category of the sample of survey stations. This 
can be achieved by applying our model to the unclas¬ 
sified sampling cells so that a tentative trawlability 
status can be assigned to them. However, because of 
the lower confidence in the trawlability status of these 
sampling cells, it would not be appropriate to assign 
them binary sampling probabilities (0 or 1) as is done 
with sampling cells whose trawlability status is deter¬ 
mined by traditional means. On the other hand, it is 
not necessary to assign equal sampling probabilities 
to these model-classified cells as is currently the case 
with the unclassified sampling cells. Instead, higher 
sampling probabilities would be assigned to sampling 
cells that our model predicts to be trawlable than to 
those that our model predicts to be untrawlable. The 
ratio of these sampling probabilities is a function of the 
prediction accuracy of the model. 
Acknowledgments 
S. Kotwicki, S. Stienessen, W. Palsson, J. Napp, J. 
Thorson, and 2 anonymous reviewers provided valu¬ 
able comments which resulted in substantial improve¬ 
ments to the initial manuscript. J. Benson provided 
technical assistance with ArcMap in generating the 
figures. Funding was provided by the Office of Science 
and Technology, National Marine Fisheries Service, via 
the Improvement of Stock Assessment initiative. We 
thank all individuals involved in collecting the acoustic 
data on the FV Sea Storm during the 2013 GOA bottom 
trawl survey. 
Literature cited 
Anderson, J. T., R. S. Gregory, and W. T. Collins. 
2002. Acoustic classification of marine habitats in coastal 
Newfoundland. ICES J. Mar. Sci. 59:156-167. 
Arlot, S., and A. Celisse. 
2010. A survey of cross-validation procedures for model 
selection. Stat. Surv. 4:40-79. 
Bartholoma, A. 
2006. Acoustic bottom detection and seabed classification 
in the German Bight, southern North Sea. Geo-Mar. 
Lett. 26:177-184. 
Busby, M. S., K. L. Mier, and R. D. Brodeur. 
2005. Habitat associations of demersal fishes and crabs 
in the Pribilof Islands region of the Bering Sea. Fish. 
Res. 75:15-28. 
