csa: Prepare an object for a CSA assessment to be used in...

In lbatts/sbar: Stage-based Assessments in R

Description

Create list with a TMB framework, including data, gradients and NLL function for a CSA assessment that can be optimised.

Usage

csa(
  catch_no,
  indices_no,
  indices_att,
  ts,
  selrec = 1,
  start_q = 1e-08,
  start_surveycv = 0.1,
  start_prec0,
  start_rec,
  start_nmort = 0.2,
  start_f_calc = 0.3,
  start_catchcv = 0.1,
  fix_nmort = TRUE,
  fix_prec0 = FALSE,
  fix_surveycv = FALSE,
  fix_catchcv = TRUE
)

Arguments

catch_no	numeric vector of catch numbers over time period of assessment
indices_no	matrix of survey indices (numbers) of dimensions: no. of indices x no.years
indices_att	matrix of survey indices attributes of dimensions: no. of indices x 2. First column defines survey and second column defines survey type (1 = recruit index, 2 post-recruit index, 3 = whole asessed population index). For example the minimum needed to run CSA is one survey split into a recruit index and a post-recruit index, the attribute matrix should look like: survey type 1 1 1 2
ts	numeric. Survey timing parameters
selrec	matrix of selection proportions of the recruit indices (in comparison to the post-recruit index) if known. Dimensions: no. of recruit indices x no. years. Defaults to 1 for all years (i.e. no difference between recruit and post-recruit indices of a survey)
start_q	Starting values for survey catchability parameters. Default is 1e-6
start_surveycv	Starting values for survey cv parameters. Default is 0.1
start_prec0	Starting parameter value for post-recruit numbers at first time step. Default is 4*max(catch.no).
start_rec	Starting parameter values for estimated recruit numbers. Default is 2*max(catch.no).
start_nmort	Starting parameter value for natural mortality. Default is 0.2
start_f_calc	Starting parameter values for estimated fishing mortality. Default is 0.3.
start_catchcv	Starting parameter value for catch cv. Default is 0.1
fix_nmort	logical. Should natural mortality be fixed in the model
fix_prec0	logical. Should post-recruit numbers in the first time step be fixed in the model.
fix_surveycv	logical. Should survey CV be fixed in the model.
fix_catchcv	logical. Should catch CV be fixed in the model.

Details

csa is simply a wrapper function that gives the output from MakeADFun from TMB, i.e. an objective function with derivatives, hessian etc. Otimisation and extraction of values from the assessment models has been kept separate to allow flexibility with optimisation methods as well as easy access to MakeADFun outputs such as the hessian, gradients etc. See http://kaskr.github.io/adcomp/_book/Introduction.html and TMB documentation for details. Users should refer to the sbar vignette by running vignette("intro_to_sbar", "sbar") for details on the csa function.

The table below gives the outputs and description of the values that can be extracted from this assessment model after optimisation and summary(TMB::sdreport(x)) of the csa object.

Output	Description
logitqhat	logit transformed survey catchability
logphat1	log transformed post-rec numbers at first time step
logrhat	log transformed recruitment numbers
logf_calc	log transformed fishing mortality
log_surveycv	log transformed survey CV
log_catchcv	log transformed catch CV
lognmort	log transformed natural mortality
phat	post-recruit numbers
rhat	recruit numbers
bhat	total numbers
lnphat	log transformed post-rec numbers
lnbhat	log transformed total numbers
lnc	log transformed catch numbers
c_calc	catch numbers
logpred_survey	log transformed predicted survey indices
sdsurv	Survey index standard deviation
f_calc	fishing mortality
phat1	post-rec numbers at first time step
nmort	natural mortality
qhat	survey catchability

Value

List with components for optimiser in R. This output is that of the function MakeADFun from TMB

Access (transformed) starting values of parameters to be estimated with x$par (where x is the csa object) to see what parameters are to be estimated.

Examples

## Not run:  obj <- csa(catch_n = catch.no, indices_no = obs,
indices_att = indices_att, ts = #'timing, #'start_nmort = nm)
## End(Not run)

lbatts/sbar documentation built on Feb. 25, 2022, 9:02 a.m.