quota.prop share of catch vs harvest rate

[vbartolino]

I run projections from a multifleet model as below.
Looking at the outcomes it seems that the parameter quota.prop sets the share of the harvest rate among the fleets rather than the share of the catch (quota) as I would have expected.
Any suggestion on how to set the proportion of the catch shared instead?

quota <- 
  gadget3:::g3_quota(
  gadget3:::g3_quota_hockeyfleet(
    predstocks = fleets,
    preystocks = list(mat_stock),
    btrigger = g3_parameterized("ven.hcr.btrigger"),
    harvest_rate = g3_parameterized("ven.hcr.harvest_rate"),
    stddev = g3_parameterized("ven.hcr.stddev")
  ),
  year_length = 1L, # quota is calculated every year
  start_step = 4L, # fishing starts at step4
  run_revstep = -2L # calc quota at step2
  )

comven12 %>% 
        g3a_predate(...,
                          catchability_f =
                              gadget3:::g3a_predate_catchability_project(
                                            quota,
                                            quota_prop = g3_parameterized("quota.prop", by_predator = TRUE, by_stock='species', value = 1),
                                            cons_step = g3_parameterized("quota.step", by_predator = TRUE, by_step = TRUE, by_stock='species',
                                                                          value = quote( step_lengths / 12.0 )),
                                  g3_timeareadata('comven12.catch.SWE',
                                                  comven12.catch.SWE %>% ...,
                                                  value_field = 'total_weight'))),

Running test projections like this

params.proj <- params.out %>%
  g3_init_val('project_years', 12) %>%
  ...
  g3_init_val('*.com_all_swe.quota.prop', value=0.2, lower=0, upper=2, optimise=FALSE) %>%
  g3_init_val('*.com_oth_fin.quota.prop', value=0.4, lower=0, upper=2, optimise=FALSE) %>%
  g3_init_val('*.com_pgt_fin.quota.prop', value=0.4, lower=0, upper=2, optimise=FALSE) %>%
  g3_init_val('*.hcr.harvest_rate',
              value=0.4, lower=0.01, upper=1e5, optimise=FALSE)

harvest rates (eg 2024)

>...
   year fleet        harv.rate
  <int> <chr>            <dbl>
1  2024 com_all_swe 0.0800    
2  2024 com_oth_fin 0.160     
3  2024 com_pgt_fin 0.160     
4  2024 survey_aco  0.00000167

catches (eg 2024)

> ...
# A tibble: 4 × 3
# Groups:   year [1]
   year predator    biomass_consumed
  <int> <chr>                  <dbl>
1  2024 com_all_swe        266368.  
2  2024 com_oth_fin         64736.  
3  2024 com_pgt_fin        214078.  
4  2024 survey_aco              3.99

[lentinj]

That's an oversight, but should be relatively easy to fix.

However, do we think that quota.prop should always be a proportion of the biomass, or should this be configurable? I'm assuming the former.

[vbartolino]

However, do we think that quota.prop should always be a proportion of the biomass, or should this be configurable? I'm assuming the former.

what do you mean by configurable...?
The case of a share on the biomass (alternative to the present share on the harvest rate which is useful to keep) is the main I can think about.

[lentinj]

I mean being able to return to the previous behaviour, as you say. I'll have a look in a bit and let you know!

[bthe]

However, do we think that quota.prop should always be a proportion of the biomass, or should this be configurable? I'm assuming the former.

I'm not sure how you would make this configurable, the problem here is that when fleet harvest rates are fixed individually the overall harvest rates will change due to a different state of the stock so you could end up with a simulation with neither the overall harvest rate nor caught biomass add up the "expected" number.

[vbartolino]

the problem here is that when fleet harvest rates are fixed individually the overall harvest rates will change due to a different state of the stock so you could end up with a simulation with neither the overall harvest rate nor caught biomass add up the "expected" number.

I suspected that difficulty as the target of the HCR here is the harvest rate...

[lentinj]

I think I've finally caught up too. We could convert the (overall) harvest rate into biomass first then apply the ratio, but the biomasses are going to be fleet-specific, so the resultant ratios won't be internally consistent.

The HCR being in biomass to start off with would make the problem go away, but presumably that's not a solution here.