add text and figures comparing north vs south

[iantaylor-NOAA]

We can't help but have folks asking for information on the relative scale of the two models, so I think providing some discussion on that topic is a good idea. Preparing for this would also provide a useful diagnostic as we make any further refinements to the two models in the days ahead.

As discussed in #14, habitat area north vs south was identified as a major uncertainty in the 2017 STAR. On that issue I suggested trying to get a calculation of habitat area from Curt but dropped the ball and I think it's too late to ask for a new analysis of lingcod-specific habitat at this stage.

However, a crude approximation could be made via total area of the two regions with the depth range 55m and 300m

data.frame(
  name = c("North","South"),
  Latitude_dd.2 = c(49, 40.166667),
  Latitude_dd.1 = c(40.166667, 34.5),
  Depth_m.1 = c(55, 55),
  Depth_m.2 = c(300, 300)
) %>%
  nwfscSurvey::StrataAreas.fn()

   name Latitude_dd.2 Latitude_dd.1 Depth_m.1 Depth_m.2     area
1 North      49.00000      40.16667        55       300 28171.09
2 South      40.16667      34.50000        55       300 12287.28

This suggests that north model area is about double that for the south (although there's lots of good Lingcod habitat in the south). However, our current models have very similar scale across the two regions:

The scale of the south model is very uncertain, and the profile over R0 shows little difference in likelihood between a model with slightly lower scale and much higher scale than the estimate in 2021.s.009.001: .

The age data are the primary reason for R0 estimate not being lower as shown in the profile:

However, it's not clear to me from looking at the good fits to the CAAL data and the trends in mean age associated with the Francis diagnostic plots what element of the age data is leading this way. Perhaps creating a full set of r4ss plots for a model with a lower R0 would be useful.

Recent assessments for two skates and dogfish have put priors on catchability of the WCGBTS, but lingcod are often associated with rocky areas which aren't as well sampled by bottom trawls, and the estimated selectivity of that survey is very different in the south than the north, so interpreting catchability is hard. For what it's worth, it's estimated at 0.78 in the north and 0.23 in the south.

I still think these two models are fine for the internal review deadline, but I think acknowledging these scale questions somehow will serve us better than having the internal review bring it up.

[iantaylor-NOAA]

Relative scale is more reasonable in latest models, but differences in estimated M between regions make comparisons of spawning biomass less useful than comparisons of summary (age 3+) biomass. I'm part-way through a fix to the r4ss::SSplotComparisons() function to address this (see https://github.com/r4ss/r4ss/issues/524).

Growth comparison plots would be useful too.

Shifting deadline to Friday.

[iantaylor-NOAA]

I'm working on adding some figures and then writing some text for the "Comparison of north and south models" section. Growth curve added in 846ffde, more to be added soon.

[iantaylor-NOAA]

Two more figures added in b54a84fa837a9416dc28f1fe94df72c34bee4b7a. Captions and info for all three figs are in figures/figures_compare_north_vs_south.csv.

I have not yet added these to 53figures.Rmd. @kellijohnson-NOAA can you do that wherever you think they would belong best.

There's more to do here, most importantly adding text, but first I'll tackle the forecast issue #73.

[iantaylor-NOAA]

Additional table has been added in e10ddd6 with same format as sensitivity table. This also includes values from 2017 model as that comparison isn't really present much elsewhere. @kellijohnson-NOAA, I'm happy to make changes, but if it looks OK, can you add it to the doc?