Trouble fitting length distributions of the survey.

[Guillermo Caballero Luján]

Hi, I am doing an assessment for the meditarranean sardine with SS3 and I am having trouble fitting the length distribution of the survey. As you can see in the attached pictures It always seems to underestimate the peaks of the distributions, I do not know why.

I tried several things, first:

- I tried adding time blocks in the biological length parameters and selectivity parameters.

- I tried changing the Nsamp for the "bad" years. 

- I changed several times the CVs parameters to see if this has any effect.

- I remove some years of the data file to see is the fit without "bad" information was better (It was not).

- I did also try to adjust the adjustment factors for length composition in the controlfile. 

This last attempt was the most succesful one, but when I did get a good fit in the survey, the other fleets get worse and the retrospective patter is pretty bad.

Fleet 1 is commercial, fleet 2 is the survey I am having the most trouble with and fleet 3 (or survey_2) is another survey.

Any help or comments on how to improve this and how to get a good retrospective analysis in general is really appreciated.

[Ian Taylor - NOAA Federal]

Hi Guillermo,

Good question. 

First, I see that you have many more samples for fleet 1 than the surveys in fleets 2 and 3 so it makes sense that the fit would be better. However, the next step I would take to improve the fit to the survey length comps would be alternative functional forms for selectivity for the surveys. Without knowing how it's set up now, it's hard to make specific recommendations, but I would consider things like a stronger dome shape for the surveys. 

Also, the survey comps shown in your attachment comp_lenfit_flt2mkt0.png seem to indicate some cohorts moving through the population (e.g. the mode increasing from 2016 to 2019). Either recruitment deviations aren't being estimated, making it really hard to fit those comps, or the deviations are being estimated and there's some mismatch in growth that prevents those modes from being fit well.

-Ian

--
You received this message because you are subscribed to the Google Groups "SS3 - Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ss3-forum+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ss3-forum/6c3eab23-a0df-4cbc-a449-bec455537c1en%40googlegroups.com.

[Mark Maunder]

Could be growth is wrong (possibly CV of length at age) or you are combining fisheries/surveys that are operating on two different components of the stock.

--

[Rick Methot]

Also consider the seasonality of the sampling for fishery length comp vs survey length comp.  Sardines are fast growing, so the position of the length mode for young fish is moving month-by-month within the year.

To view this discussion visit https://groups.google.com/d/msgid/ss3-forum/SN6PR02MB5648C38B6095D372899B44B8CCC9A%40SN6PR02MB5648.namprd02.prod.outlook.com.