In BenWilliams-NOAA/rockfishr:

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  eval = FALSE
)

Goal

The goal of rockfishr is to create a clear workflow for pulling and cleaning data for rockfish fishery stock assessments based upon the ADMB rockfish assessment model. It utilizes a "project oriented workflow" via RStudio. You must be able to have a connection to the AFSC & AKFIN (Answers) data servers (e.g., VPN if offsite), and have usernames/passwords setup.

Installation

The rockfishr package can be installed from github with:

# install.packages("devtools")
devtools::install_github("BenWilliams/rockfishr")

Use

There are a suite of "global" items that are used repeatedly in the rockfishr package. Currently the package is setup to deal with northern rockfish "NORK" or dusky rockfish "DUSK", other species can be added as needed.

library(rockfishr)

# globals ----

species = "NORK" # northern rockfish
year = 2020 # assessment year
afsc_user = "your_afsc_username"
afsc_pwd = "your_afsc_password"
akfin_user = "your_akfn_username"
akfin_pwd = "your_akfin_password"

admb_home = "C:/Program Files (x86)/ADMB-12.1" # location admb exists on your computer 
                                               # if is "c:/admb" can leave NULL
TAC = c(3786, 3681, 4528) # last three years of TAC (year-3, year-2, year-1)
rec_age = 2 # recruitment age
plus_age = 45 # plus group age
model_name = "updated_nr" # name of ADMB .tpl file
dat_name = "goa_nr_2020" # name of ADMB .dat file

mcmc = 1e+07 # number of mcmc draws
mcsave = 2000 # number of saved mcmc draws

Create directories

The first function creates a number of directories within your R project folder. The setup is organized by year within a folder. Using the globals above in the modeldir() function would create the following folders and files:

# setup folders -----

modeldir(year)

project
|__2020
    |__data
        |__models
        |    |__ageage
        |    |     AGEAGE.tpl
        |    |__allometric
        |    |     allometric.tpl
        |    |__length_sd
        |    |     lengthSD.tpl
        |    |__VBL
        |    |      VBL.tpl
        |    |__wVBL
        |    |      wVBL.tpl
        |    |      lvb.ctl
        |
        |__output
        |__raw
        |__SARA
        |__user_input

This can then be updated for each year.

# setup folders -----

modeldir("2021")

Which would create a duplication of the folders for the given year

project
|__2020
|   |__data
|       |__models
|       |    |__ageage
|       |    |     AGEAGE.tpl
|       |    |__allometric
|       |    |     allometric.tpl
|       |    |__length_sd
|       |    |     lengthSD.tpl
|       |    |__VBL
|       |    |      VBL.tpl
|       |    |__wVBL
|       |    |      wVBL.tpl
|       |    |      lvb.ctl
|       |
|       |__output
|       |__raw
|       |__SARA
|       |__user_input
|__2021
    |__data
        |__models
        |    |__ageage
        |    |     AGEAGE.tpl
        |    |__allometric
        |    |     allometric.tpl
        |    |__length_sd
        |    |     lengthSD.tpl
        |    |__VBL
        |    |      VBL.tpl
        |    |__wVBL
        |    |      wVBL.tpl
        |    |      lvb.ctl
        |
        |__output
        |__raw
        |__SARA
        |__user_input        

Query raw data

The next step is to query the raw data from the AKFIN and AFSC databases.

# query databases ----
raw_data(species, year, afsc_user, afsc_pwd, akfin_user, akfin_pwd)

The initial output will be raw data pulls that will be placed in the year/data/raw folder. These are kept in the "raw" format with an associated timestamp of when the data were queried. To see the queries being perfomed one can either type raw_data in their console and hit enter, or explore the code on github.

project
|__2020
    |__data
        |__models
        |__output
        |__raw
            data_called.txt
            fishery_age_comp_data.csv
            fishery_catch_data.csv
            fishery_obs_data.csv
            fishery_size_comp_freq.csv
            srv_age_comp.csv
            srv_age_specimens.csv
            srv_biomass.csv
            srv_saa_age.csv
            srv_saa_length.csv
            srv_size_comp.csv
            srv_size_freq.csv

Clean and process data

There are a suite of functions to process the raw data, the results are placed in the output folder. These functions need to be run in order.

clean_catch(year, TAC)
ageage(reader_tester = NULL, species, year, admb_home, region = "GOA", rec_age, plus_age)
fish_age_comp(year, rec_age, plus_age)
survey_age_comp(year, rec_age, plus_age)
fish_size_comp(year, rec_age)
survey_size_comp(year)
size_at_age(year, admb_home, rec_age)
weight_at_age(year, admb_home, rec_age)  
# note: must provide a file for VAST survey biomass estimates otherwise D-B estimates are output
survey_biomass(year) 
concat_dat(year, "db", rec_age, plus_age) # create a .dat file in the "db" folder

# VAST "bridge model"
survey_biomass(year, "VAST_estimate_mesa.csv") 
concat_dat(year, "m18.2", rec_age, plus_age) # create a .dat file in the "m18.2" folder

Alternatively there is a function that will run all of these functions

clean_raw(year, species, TAC, admb_home, rec_age, plus_age, reader_tester = NULL)

Run models

Currently all models are being run through the ADMB command line.
Compile model, run model with -mcmc and -mcsave (if running mcmc...).
Run model with -mceval.

Process results

Define the model (e.g., folder) run the process_results function and the base_plots function.

The process_results function cleans outputs from ADMB and generates a series of .csv files that are placed within a folder titled processed nested in the model folder. The function also creates a figs and tables folder within the model folder.

The plotting function will populate the figs folder with a number of figures from the model output.

model = "db"
process_results(year, model, model_name, data_name, rec_age, plus_age, 
                mcmc = mcmc, mcsave = mcsave)

base_plots(year, model, model_name, rec_age)


model = "m18.2"

process_results(year, model, model_name, data_name, rec_age, plus_age, 
                mcmc = mcmc, mcsave = mcsave, survey = "VAST_estimate_mesa.csv")

base_plots(year, model, model_name, rec_age)

project
|__2020
    |__data
    |__db
    |   |__figs
    |   |__processed
    |   |__tables
    |__m18.2
        |__figs
        |__processed
        |__tables        

Retrospective

For any models that you have run mcmc draws on and would like retrospective examinations the following code can be used to run the retros. The output will be placed in a retro folder in the model directory.

# run retro ----
run_retro(year, model = "m18.2", tpl_name = model_name, n_retro = 10,
          admb_home = admb_home, mcmc = 500000, mcsave = 100)

Retrospective results can then be plotted. These figures will be located in the figs folder.

# plot retros ----
plot_retro(year, model)
plot_retro_survey(year, model)

BenWilliams-NOAA/rockfishr documentation built on March 1, 2021, 11:12 p.m.