NOAA Protocols for Fisheries Acoustics Surveys and Related ...

Vessel position is determined by using Global Positioning System (GPS) fixes. These fixes
serve as the primary geographic reference for all data and events.
In rough seas, survey speed may need to be reduced to maintain data quality and safe
shipboard operations. The field party chief, in consultation with the master of the vessel, must
balance the need to maintain data quality, the need to make progress on completing the survey,
and safety considerations when deciding whether to alter or suspend survey operations.
Error
Uncertainty, random, systematic – The national protocol document notes that
“[t]he survey design (timing and location) should consider potential systematic changes in detection
probability. If systematic changes in detection probability are discovered, either a change in the survey design
is required or analyses should be conducted to determine a correction factor.” (NOAA Protocols for Fisheries
Acoustics Surveys and Related Sampling)
As mentioned previously, a major assumption made in this survey is that the entire stock is
available to the survey effort. Potential bias includes incomplete coverage of the population.
The technique of linear interpolation at each cell area and subsequent summing to desired
area does not allow for propagation of error in the estimates of abundance, meaning the level of
uncertainty in biomass estimate is not known.
The reader should note that this section addresses potential sources of error in the acoustic
survey design and sampling, not in the stock assessment modeling process.
Considerations
Remediation – If it is found that the survey design is in some facet inappropriate (e.g., ill
timed, deficient in geographic coverage, or the acoustic technique used is found not to be
robust across full range of conditions employed) a new survey design must be considered.
However, changes in design must include a strategy for considering the potential impacts on
the complete survey time series as on future surveys. As an example, the survey design by the
Pacific hake survey underwent changes in 1992 and 1995: the survey was expanded offshore
and further northward, and previous data points in the survey time series were back-corrected
for this expansion in the assessment (Dorn et al., 1994; Dorn, 1996; Wilson and Guttormsen,
1997). The revision of the design was done based on an accumulation of new information
about stock distribution (more northerly and offshore) to ensure more complete coverage of the
population.
Understanding the uncertainty associated with the coast wide Pacific hake biomass estimate
is an area of current research. One initial approach that has already been attempted is to apply
the technique of Jolly and Hampton (1990) in a post survey stratification scheme that treats each
transect as a sampling unit (Fleischer et al., in review). In this way, a mean and variance for
biomass in each stratum and for the total biomass was estimated, however the error associated
with the point estimate propagated by this technique did not consider observation errors.
123