Record Details
Sensitivity of the stock synthesis assessment model : a simulation approach
ScholarsArchive at Oregon State University
| Field | Value |
|---|---|
| Title | Sensitivity of the stock synthesis assessment model : a simulation approach |
| Names |
Yin, Yanshui
(creator) Sampson, David B. (advisor) |
| Date Issued | 2001-09-28 (iso8601) |
| Note | Graduation date: 2002 |
| Abstract | Stock assessments for many U.S. Pacific coast groundfish stocks are developed using the catch-at-age method known as Stock Synthesis. In this work a simulation package was developed and used to evaluate the sensitivity of the Stock Synthesis program. More specifically, the evaluation focused on the impacts of input data errors and stock characteristics on the accuracy and precision of Synthesis estimates. Factors examined included the length of the time series of data, the rate of natural mortality, the shape of the fishery and survey selectivity curves, the trend in the rate of fishing mortality, the recruitment pattern, and errors in the observed data for annual catch, fishing effort, fishery and survey age composition, and survey biomass indices. First, the study evaluated the sensitivity of the Stock Synthesis program applied to populations with simple multinomial age compositions. The length of the data series and sample size were the two most influential factors. Second, the study focused on populations with compound multinomial age composition, in which the age composition data were over-dispersed relative to simple multinomial samples. When the fishery age composition actually followed a compound multinomial distribution, the estimates produced by the Stock Synthesis program, which assumed simple multinomial distributions with maximum sample sizes of 400 fish, were moderately more biased and more variable. When applying Synthesis to populations whose age compositions follow compound multinomial distributions, the results from the experiments indicated that a common configuration, in which age sample sizes in the likelihood specification are limited to 400 fish per sample, probably gives age composition data too much emphasis. The experiments indicated that using 200 as the upper limit provided more accurate results than using 400. Third, the actual stock assessment of yellowfin sole (Limanda Aspera) was taken as a case study and it was found that more accurate assessment results could be achieved from a better balance in the amount of sampling effort allocated to age composition data versus survey biomass estimates. |
| Genre | Thesis/Dissertation |
| Topic | Fish stock assessment -- Mathematical models |
| Identifier | http://hdl.handle.net/1957/28564 |
