Welcome to the NOAA Institutional Repository |
Stacks Logo
Advanced Search
Select up to three search categories and corresponding keywords using the fields to the right. Refer to the Help section for more detailed instructions.
 
 
Help
Clear All Simple Search
Advanced Search
Examination of the Statistical Performance of the Mean Basket Weight Protocol in Survey Catch Processing
Filetype[PDF-1.92 MB]


Details:
  • Description:
    The Alaska Fisheries Science Center’s biennial bottom trawl surveys in the Gulf of Alaska (GOA) and Aleutian Islands (AI) are essential for assessing groundfish stock status and for formulating best management practices. When the catch of a species is too large to completely sample, one of several catch estimation procedures is employed. The ‘counted basket’ method is common in which the catch is processed using weighed and unweighed (counted) baskets, and total catch weight is estimated using the estimated mean basket weight and total basket count. Thirty consistently-filled basket weights are recommended to estimate the mean. No analytical foundation exists to demonstrate the accuracy of this approach in estimating large volume catches. We assess the error associated with this estimation procedure, and examine its statistical performance in terms of the representativeness and robustness of the catch estimates. If shown to be robust, then its use would address ergonomic and safety concerns of whole-haul processing large catches. Three data sets were the subject of our analysis. Two were large volume (> 7.0 metric tons [t]) catches from the 2016 AI survey (Atka mackerel [Pleurogrammus monopterygius]) and the 2017 GOA survey (Pacific ocean perch [Sebastes alutus]). While they were completely sampled, the baskets were not consistently filled and, consequently, there was greater variability in these data than if the protocol had been applied. The third data set was a medium volume (~ 3.0+ t) catch of Pacific ocean perch collected from the 2018 AI survey, completely sampled and processed consistent with design specifications. The performance of the counted basket method was simulated via bootstrap analysis over a range of population sizes (~ 1.0 t to 7.0+ t) and sample sizes (n = 1 to 50 baskets). For the two non-standard data sets, the point estimates of catch were accurate and relatively precise over the range of catch volumes and sample sizes.
  • Document Type:
  • Main Document Checksum:
  • Supporting Files:
    No Additional Files
No Related Documents.
You May Also Like: