Weighting of catch at age data in stock assessment models

Trophia is contracted by the New Zealand Ministry of Fisheries (MFish) to develop techniques to improve the stock assessment of harvested fish stocks. Specific objectives include:

To determine the appropriate weighting of catch at age data in stock assessment models

Stock assessment models often use data inputs from a variety of sources, including catch sampling and abundance surveys. The biomass indices obtained from abundance surveys (or CPUE analysis) are assumed to follow a continuous distribution but catch at age data are often assumed to have an underlying discrete distribution (e.g., multinomial).

This combination of assumptions, and the generally greater availability of catch at age data, means that the statistical likelihood is usually dominated by the catch at age component. This can cause the stock assessment results to be primarily determined by the catch at age data, even though such data are generally considered only to contain reliable information about year class strengths, rather than absolute abundance.

The usual approach to deal with this perceived problem is an arbitrary down-weighting of the catch at age data although other approaches have been tried. The aim of this project is to review alternative methodologies for incorporating catch at age data in stock assessment models and to document the interpretation of diagnostic tests currently used in assessment models in New Zealand and elsewhere, a simulation approach may be used to test if diagnostice tests are appropriate.

The final report for this project has been accepted for publication in: ICES Journal of Marine Science. A copy will soon also be available from this site.

This work is funded by MFish project SAM2003-02