This is a list of possible things that could be done. Will not get to everything.
[ ] Explore ECE survey indices and compare to non ECE indices. Explore in model
Not going to happen. sdmTMB versions do not converge and give incorrect answers. VAST options are not available as a choice of distribution. Older saptialtemporal deltaGLMM is deprecated. Exploration of sensitivity to survey results explored using differences between gamma and lognormal
[ ] Use previous values for triennial index to be able to account for ECE
This would require keeping the triennial split and not being able to mirror q. That is less than ideal so not exploring this. Should reviewer wish to we can do
[ ] Spatial model
Early exploration indicate some instability when unmirroring and expanding selectivity. Needed to focus on something so preferentially went with coastwide model. Later fixes may have resolved instability
[ ] Rec indices
Not focused on because of concern with how a coastwide model would handle multiple regional rec indices
[ ] Explore alternative M setup by using length based asymptotic selectivity but age-based descending limb selectivity.
[ ] Could also potential explore platoons as way to account for variable ages however this didn't seem viable since growth is sex dpeendent so have some of that implicit already
[ ] Consider adding fishery based conditional age at lengths
[ ] Modeling retention within the model. Would need to split out discards and landings, and then have specific discard compositions.
[ ] Use expanded age and length comps from the at sea hake fisehry
[x] The sd for the late triennial q is big. Makes me think that there is something going on with the negative fleet assignment. Its mirrored so should show up as a parameter but it is. After talking with others, because that fleet is mirrored the q can be anything and not affect the likelihood hence why sd is so large
This is a list of possible things that could be done. Will not get to everything.
Not going to happen. sdmTMB versions do not converge and give incorrect answers. VAST options are not available as a choice of distribution. Older saptialtemporal deltaGLMM is deprecated. Exploration of sensitivity to survey results explored using differences between gamma and lognormal
This would require keeping the triennial split and not being able to mirror q. That is less than ideal so not exploring this. Should reviewer wish to we can do
Early exploration indicate some instability when unmirroring and expanding selectivity. Needed to focus on something so preferentially went with coastwide model. Later fixes may have resolved instability
Not focused on because of concern with how a coastwide model would handle multiple regional rec indices
[ ] Explore alternative M setup by using length based asymptotic selectivity but age-based descending limb selectivity.
[ ] Could also potential explore platoons as way to account for variable ages however this didn't seem viable since growth is sex dpeendent so have some of that implicit already
[ ] Consider adding fishery based conditional age at lengths
[ ] Modeling retention within the model. Would need to split out discards and landings, and then have specific discard compositions.
[ ] Use expanded age and length comps from the at sea hake fisehry
[x] The sd for the late triennial q is big. Makes me think that there is something going on with the negative fleet assignment. Its mirrored so should show up as a parameter but it is. After talking with others, because that fleet is mirrored the q can be anything and not affect the likelihood hence why sd is so large