Skip to contents

Use decamod as the operating model for compositional MSE

Source: R/mse-decamod.R
om_decamod.Rd

This first adapter preserves the single-region, two-season CRA2 toy loop. It uses prepare_mpe() with the complete configured posterior pool before selecting iteration IDs, so changing a small test's iteration count does not silently change the posterior/recruitment pairing. Multiregion operation is explicitly rejected until supported, rather than implicitly collapsed.

Usage

om_decamod(
  models,
  rec_devs,
  initial_tacc,
  catch_rec,
  region,
  reference_years,
  posterior_pool = 2000L,
  recruitment_overlap = 4L,
  tacc_history = numeric(),
  robustness = mse_robustness(),
  outputs = NULL
)

Arguments

models

Named character vector of fitted-model RDA files. Each contains data, parameters, map, and mcmc.

rec_devs

Named character vector of corresponding RDA files with Rdev_yr.

initial_tacc

First projection-year TACC, in tonnes.

catch_rec

Fixed recreational catch in AW and SS, in tonnes.

region

Exact region name in catch arrays.

reference_years

Historical OM years used for evaluation of true status.

posterior_pool

Size of the posterior pool used by the saved experiment.

recruitment_overlap

Number of historical years replaced by supplied recruitment projections (four in the existing CRA2 experiment).

tacc_history

Optional actual TACC history preceding the projection, needed when using latent-year rules.

robustness

A mse_robustness() configuration shared by all scenarios, or a named list with exactly one configuration per scenario.

outputs

Optional mse_outputs() specification for reference levels, bin-aligned abundance indicators, and full numbers-at-size retention.

Value

An mse_om() adapter. Stock-specific files/settings are not bundled into the package.

Details

Catch implementation retains the previous year's commercial seasonal split, adds fixed seasonal recreational catch to fishery 1, and leaves other catch components at their projected OM values. The GP legal observation uses the fishery-1 catch; SPM total observations sum all modelled fisheries. There is no TACC implementation error by default. Explicit mse_robustness() settings can change prospective observations, recruitment, and removals.

Exploitation-fraction OMs record diagnose_exploitation() checks under om_exploitation_* truth columns, including whether a projection trajectory was previously or ever guarded. These flags survive annual checkpoints. They neither discard trajectories nor change catch handling, observations, EM acceptance, or HCR advice. Historical (pre-projection) events are not included in the cumulative projection flag.