Sensitivity ideas

[brianlangseth-NOAA]

DATA

• Remove commercial comps. There are sampled in certain areas at certain times and show a strong trend in larger sizes. What happens if remove these comps.
• Alternative values for some data decisions (e.g. for 1991 commercial value)

STRUCTURE

• Combine by female and male (if separate), or separate male and female (if combined)

SELECTIVITY

RECRUITMENT
The estimates of recruitment are uncertain. This is not unexpected given that they growth to Linf quickly and live long. However, recuritment estimates are important and influence status. Explore sensitivities around this.

• Look at effect of data weighting on recruitment estimates. With only rec and comm fleets, and only length, think about what down - weighting lengths implies for things like recruitment.
• Consider relaxing assumption on recdevs having to sum to one

INDICES
If an absolute abundance values is to be used as a sensitivity there are a number of ways this could be incorporated

• Fit the survey as a relative value while estimating q. Then fix q at various set values and see how treating the survey as an absolute measure affects the model estimates. This is Brian's preferred approach for exploring effect of absolute assumption
• Fix steepness and M and growth at their base model values. Apply a multiplier to the catches so the biomass estimates from the assessment match the absolute survey estimate and provide a measure of the degree of underreporting needed to match the scale (assuming productivity parameters were correct)
• Using the base model catch stream, determine what estimates of steepness and natural mortality and growth produce the absolute survey estimate. You can do this quickly by turning off all data sources except the index, fix the index q at 1, and force the model to fit the index perfectly. Can check whether the productivity estimates are plausible
• Look to the Oregon Black rockfish stock assessment which explored consideration of a single absolute abundance index. Ultimately they used it to inform decision tables.