von Szalay and Somerton: A method for predicting trawlability in the Gulf of Alaska 
497 
i-—--————i----1----—~i— 
170°W 160°W 150°W 140°W 
Figure 1 
Map of the survey area for the bottom trawl survey conducted biennially in the 
Gulf of Alaska by the National Marine Fisheries Service since 1999, confined by the 
lOOQ-m isobath (dotted line). The area confined by the 300-m isobath (solid line) 
represents the area where acoustic data within a grid of cells were used to predict 
trawlability. Green circles represent acoustic data segments from trawlable areas, 
and red circles represent acoustic segments from untrawlable areas. 
lated only to areas that are trawlable, and some new 
sampling methods (e.g., the use of acoustic, longline, 
or camera gear) would be applied to the untrawlable 
areas (Rooper et ah, 2010; Williams et al., 2010; Jones 
et. al., 2012; Thorson et al., 2013). To implement this 
sampling design, the relative proportions of trawlable 
and untrawlable areas must be known, proportions 
that require a technique for identifying and quantify¬ 
ing trawlable and untrawlable habitats (Cordue, 2007). 
Previously, the approach to identify the trawlability 
status of an area was based on the historical perfor¬ 
mance of the survey within individual cells of a grid of 
5x5 km cells superimposed on the survey area. These 
cells, which number -14,000, are the potential sam¬ 
pling units of the survey. To be considered trawlable, 
a sampling cell must have been successfully trawled 
(with a Poly-Nor’eastern 4-seam survey trawl in stan¬ 
dard fishing configuration and without sustaining any 
damage throughout the tow) by a survey vessel dur¬ 
ing a prior GOA bottom trawl survey. To be considered 
untrawlable, a sampling cell must be judged so on the 
basis of the vessel captain’s assessment of the echo- 
gram, which is a real-time image of the backscatter 
data collected from a hull-mounted single-beam echo- 
sounder on a survey vessel. Any seabed feature that 
a vessel captain deems likely to result in moderate to 
severe damage to fishing gear or prevents the fishing 
gear from maintaining proper configuration and bottom 
contact throughout a tow provides a valid reason for 
declaring a sampling cell untrawlable. Nine categories 
of untrawlable seabed features have been identified: 1) 
hard+rocky substrate, 2) steep slopes, 3) rolling sea¬ 
bed, 4) pinnacles, 5) unnavigable areas, 6) snags, 7) 
ledges, 8) presence of underwater cable, and 9) pres¬ 
ence of fixed fishing gear. 
Captains of survey vessels have learned to recognize 
bottom features visible on echosounder traces and that 
could damage a survey trawl (Poly-Nor’eastern 4-seam 
bottom trawl with roller gear; Stauffer, 2004). There¬ 
fore, a sampling cell is classified as untrawlable if the 
captain fails to find a suitable bottom, at least 1 km 
in length, after systematically searching a sampling 
cell for 2 h, a duration that is considered sufficient to 
cover a sampling cell. Although consistent with the op¬ 
erational survey procedures of the bottom-trawl sur¬ 
vey, this approach provides a qualitative and relatively 
slow assessment of seafloor character (only 42% of the 
survey area classified by this method to date). 
We are exploring the use of acoustics to detect bot¬ 
tom features associated with trawlability to increase 
