Standard deviations and F calculations

Forums

RSS (Opens New Window)

Audric Vigier, modified 5 Years ago.

Standard deviations and F calculations

Youngling Posts: 6 Join Date: 9/6/16 Recent Posts

Hi all,

I am working on a spatially-explicit SS3, including 3 zones. In the Report.sso file, in the DERIVED_QUANTITIES section, there are estimates for annual SPB, F and recruitment, and also a standard deviation for each of these estimates. These estimates are about the whole fishery, and are very interesting to me, but I also want the same thing for each zone.

I can easily get / calculate the SPB, number of recruits and F for each zone, and also movement estimates, however, it doesn't seem to be the case for standard deviations; I haven't found a feature in SS3 or r4ss which computes standard deviations for these estimates, per zone. For now, to get these standard deviations, I am using the delta method described in Seber 1982. It took me a bit of time for movements and spatial repartition of recruitment standard deviations; but I don't want to use this method for F and SPB spatialized estimates, it just would be endless.

Also, I don't get the way F is calculated in the DERIVED_QUANTITIES section . I chose the 4th option of F report value in starter.ss, and chose raw values for F report basis. According to the user guide, the F in the DERIVED_QUANTITIES section are "real annual F calculated as a numbers weighted F for a specified range of ages". How is the weighting done? What is the formula? I tried to find the formula by trying several weightings on F per age and year using abundance in numbers at the beginning of the year, I still can't reproduce the F values in the DERIVED_QUANTITIES section.

Also, It would be more convenient if I could report a mean F for a specified range of ages (especially for the standard deviations calculation), instead of numbers weighted F. For now, SS3 doesn't seem to allow such a F reporting. Could it be possible to allow the user to choose between these F reporting in the future?

Best regards,

Audric

Richard Methot, modified 8 Years ago.

RE: Standard deviations and F calculations

Youngling Posts: 222 Join Date: 11/24/14 Recent Posts

Audric,

You are right to want to see area-specific standard errors on these quantities. We will put that on the wish list. We already have unweighted average F on the wish list.

The calculation of F as a single annual value is a complex concept when there are multiple fleets operating in multiple areas. Even worse if there are multiple seasons. The option you chose is best for such complex situations and it uses a Z-M approach. Say you chose to get the F for age range 9-10. SS adds up all the 9 year olds across all areas at the beginning of one year to get N9 and all the 10 year olds at the beginning of the next year. Then an overall annual Z across all areas is ln(N10/N9). If M was a simple constant, then F=Z-M. But M might also be complex in some way, so SS goes back through the time series with F turned off so that all dynamics are based on M. Now ln(N10/N9) gives the annual M across all areas. Then F=Z-M can be done.

Rick