Specify Movement Using R Shiny App

This vignette includes an example of how to specify movement using the Specify_move() Shiny APP when spatial structure is complex (i.e. the number of stocks/regions are high).

1. Load “WHAM” and “whamMSE”:

library(wham)
library(whamMSE)
library(shiny)
main.dir = here::here()

2. Generate basic information

year_start  <- 1  # starting year in the burn-in period
year_end    <- 20  # end year in the burn-in period
MSE_years   <- 3     # number of years in the feedback loop
# Note: no need to include MSE_years in simulation-estimation 

info <- generate_basic_info(n_stocks   = 4, # Note: now change to 4 stocks
                            n_regions  = 4, # Note: now change to 4 regions
                            n_indices  = 4, 
                            n_fleets   = 4, 
                            n_seasons  = 4, 
                            base.years = year_start:year_end, 
                            n_feedback_years = MSE_years, 
                            life_history  = "medium", 
                            n_ages        = 12, 
                            Fbar_ages     = 12, 
                            recruit_model = 2, 
                            F_info     = list(F.year1 = 0.2, Fhist = "updown",
                                              Fmax = 1, Fmin = 0.2, change_time = 0.5),
                            catch_info = list(catch_cv = 0.1, catch_Neff = 200), 
                            index_info = list(index_cv = 0.1, index_Neff = 200, 
                                              fracyr_indices = 0.625, q = 0.2), 
                            fracyr_spawn = 0.625, 
                            bias.correct.process     = FALSE, 
                            bias.correct.observation = FALSE, 
                            bias.correct.BRPs        = FALSE, 
                            mig_type = 0) 

basic_info = info$basic_info # collect basic information
catch_info = info$catch_info # collect fleet catch information
index_info = info$index_info # collect survey information
F_info = info$F # collect fishing information

# see more details using ?generate_basic_info

3. Specify movement type and movement rate

basic_info <- generate_NAA_where(basic_info = basic_info, move.type = 2) # "bidirectional" movement

move <- generate_move(basic_info = basic_info, move.type = 2, move.rate = c(0.1,0.1,0.1,0.1), move.re = "constant")

4. Mannually specify stock- or region-specific movement rate using Shiny APP

Specify_move()
# Note: for more information please also check the generate_move() function

Figure 1

5. Load defined movement rate array

move <- readRDS("move.rds")

6. Configure selecitvity and natural mortality

n_stocks  <- as.integer(basic_info['n_stocks'])
n_regions <- as.integer(basic_info['n_regions'])
n_fleets  <- as.integer(basic_info['n_fleets'])
n_indices <- as.integer(basic_info['n_indices'])
n_ages    <- as.integer(basic_info['n_ages'])

# Selectivity Configuration
fleet_pars <- c(5,1)
index_pars <- c(2,1)
sel <- list(model=rep("logistic",n_fleets+n_indices),
            initial_pars=c(rep(list(fleet_pars),n_fleets),rep(list(index_pars),n_indices)))

# M Configuration
M <- list(model="constant",initial_means=array(0.2, dim = c(n_stocks,n_regions,n_ages)))

sigma        <- "rec+1"
re_cor       <- "iid"
ini.opt      <- "equilibrium" # option <- c("age-specific-fe", "equilibrium")

NAA_re <- list(N1_model=rep(ini.opt,n_stocks),
               sigma=rep(sigma,n_stocks),
               cor=rep(re_cor,n_stocks),
               recruit_model = 2,  # rec random around the mean
               NAA_where = basic_info$NAA_where) # NAA_where must be specified in basic_info!

7. Generate wham input

Here we use prepare_wham_input() function to generate a wham input using the basic information we set above:

input <- prepare_wham_input(basic_info = basic_info, selectivity = sel, M = M, NAA_re = NAA_re, move = move, catch_info = catch_info, index_info = index_info, F = F_info)

8. Generate operating model

om <- fit_wham(input, do.fit = F, do.brps = T, MakeADFun.silent = TRUE)
# Note: do.fit must be FALSE (no modeling fitting yet)