runAllPelfossPaper1: Funciton for running all specified surveys of paper 1 of the...

In Sea2Data/pelfoss: Simulate acosutic-trawl surveys and produce survey estimated using Rstox

Description

Funciton for running all specified surveys of paper 1 of the PELFOSS project.

Usage

runAllPelfossPaper1(
  dir,
  type = "Parallel",
  run = type,
  survey = "Herring_IESNS",
  speedFact = 1,
  condPar = NULL,
  LcmMean = NULL,
  b = NULL,
  year = c(2010, 2012),
  res = c(4, 10),
  dateshift = c(-30, 0, 30),
  reversed = c(FALSE, TRUE),
  seed = 1,
  nboot = 100,
  radius = 10,
  subset = "all",
  cores = 1,
  ...
)

Arguments

dir	The directory of the PELFOSS folder structure (see getPelfossSkeleton).
type	The survey design type (see surveyPlanner).
run	The name of the simulation run (which is three timings (-30, 0, +30 days), two directions (normal, reversed), and two resolutions (low, high), in total 12 individual runs).
survey	The name(s) of the survey(s) to run. Must be one or more of "Herring_IESNS", "Herring_NASSHS", "Herring_NASSHS_fishery", "Mackerel_IESSNS", "Mackerel_IESSNS_sept" and "Mackerel_IESSNS_sept_fishery", which are the surveys used in the Paper 1 of the PELFOSS project.
speedFact	A factor to speed the survey up or down by. Values larger than 1 speed up and values smaller than 1 slow down.
condPar	A vector of two elements giving the parameters a and b in the length-weight relationship Wg = a * Lcm^b, where Wg is the weight in grams and Lcm is the length in cm. Typical values are e.g., a = 0.01 and b = 3. If not set, this will be estimated from the super individuals, possibly inside a radius given as the second elements of radius.
LcmMean	The typical length of the fish, representing the mean acoustic backscatter which refects the square of fish length. This can typically be estimated as sqrt(mean(L^2)). If not set, this will be estimated from the super individuals, possibly inside a radius given as the second elements of radius.
year, res, dateshift, reversed	These parameters are linked to the folder structure specifide by getPelfossSkeleton, and specifies the year/resolusion, which are linked in the PELFOSS folder structure, the shift in date, which is a vector of integer days to shift the survey by, and a logical stating whether to reverse the direction of the survey. See the default values.
seed	A single integer, or a list of seeds used in the funciton, including the following seeds: ('transect') for drawing transects using surveyPlanner(), ('trawl') for drawing trawl stations along the transects with probability as a funciton of the NASC (see probfact), and ('bootstrap') for getting the final estimate using runBootstrap() (see nboot).
nboot	Number of bootstrap replicates.
radius	A vector of one or two elements giving the radius inside which to sample trawl stations, and the radius inside which to estimate the parameters a, b and L in the length-weight relationship. If only one element is given, the length-weight relationship is estimated from all adult super individuals.
subset	Either "all" (the default) or an integer vector giving the subset of the simulations to run.
cores	The number of cores to use for writing XML files and for the bootstrapping in the StoX project, given either as a named list such as list(biotic=1, acoustic=1, bootstrap=1), og a single numeric setting the cores for all.
...	Additional inputs overriding the defaults returned by pelfossDefaults().

Examples

## Not run: 
# Script for running experiments for Paper 1 in the PELFOSS project:

# The experiment requires the Rstox and pelfoss package to be installed (see https://github.com/Sea2Data/Rstox and https://github.com/Sea2Data/pelfoss for install instructions):
library(pelfoss)
library(Rstox)

# Define the directory in which to run the simulation experiments:
dir <- "~/Projects/PELFOSS"
# Download the simulation experiment folder to the experiment directory:
expermientURL <- ""
temp <- download.file(expermientURL, dir)
unzip(temp)

# Specify to use parallel transect lines:
type <- "Parallel"
# Specify the number of bootstrap replicates used in the variance estimation of the survey estimates:
nboot <- 100
# Define the year and resolution, and the shift by date and direction, which will constitute a grid in the function runAllPelfossPaper1():
year <- 2012
res <- 10
dateshift <- c(-30, 0, 30)
reversed <- c(FALSE, TRUE)
# Define the radius inside which the trawl stations are drawn:
radius <- 10
# Define the surveys to run:
survey = c(
	"Herring_IESNS", 
	"Herring_NASSHS", 
	"Herring_NASSHS_fishery", 
	"Mackerel_IESSNS", 
	"Mackerel_IESSNS_sept", 
	"Mackerel_IESSNS_sept_fishery"
)
# Define the names of the 
runs <- c(
	"Parallel", 
	"Parallel_Fixed_abL", 
	"Parallel_Fixed_abL_SpeedTimes50"
)
# Run for seeds 1 and 2:
seeds <- c(1, 2)
# Write xml files and run boostrapping on multiple cores:
cores <- 8

################
##### Runs #####
################
############################################################
##### 1. Parallel, with with estimation of W0 and L0: ######
############################################################
for(seed %in% seeds){
	Parallel_Fixed_abL <- runAllPelfossPaper1(
		type=type, run=runs[1], survey=survey, dir=dir, # Paths and names
	year=year, res=res, dateshift=dateshift, reversed=reversed, # spec
	seed=seed, nboot=nboot, radius=radius, subset="all", cores=cores
	)
}
############################################################

############################################################
##### 2. Parallel, with 'condPar' and 'LcmMean' fixed: #####
############################################################
for(seed %in% seeds){
	Parallel_Fixed_abL <- runAllPelfossPaper1(
		type=type, run=runs[2], survey=survey, dir=dir, # Paths and names
		condPar=c(0.01, 3), LcmMean=30, 
		year=year, res=res, dateshift=dateshift, reversed=reversed, # spec
		seed=seed, nboot=nboot, radius=radius, subset="all", cores=cores
	)
}
############################################################

############################################################
##### 3. Parallel, with 'condPar' and 'LcmMean' fixed ######
############# and survey conducted on one day: #############
############################################################
for(seed %in% seeds){
	Parallel_Fixed_abL <- runAllPelfossPaper1(
		type=type, run=runs[3], survey=survey, dir=dir, # Paths and names
		speedFact=50, condPar=c(0.01, 3), LcmMean=30, 
		year=year, res=res, dateshift=dateshift, reversed=reversed, # spec
		seed=seed, nboot=nboot, radius=radius, subset="all", cores=cores
	)
}
############################################################
	
###################
##### Reports #####
###################
# Get all reports (set copy = FALSE to speed up if rerunning this function):
reports <- lapply(runs, reportAllPelfoss, dir=dir, copy=TRUE)
names(reports) <- runs

# Show the parameters:
headPar <- lapply(runs, function(this) head(reports[[this]]$par))
names(headPar) <- runs
headPar

#################
##### Plots #####
#################
#  Plots with ylim up ro 2.15:
temp <- lapply(runs, plotPelfossReports, dir = dir, by = "InsideAll", ylim = c(0, 2.15), subdir = "allPlots_max2.15_cv_allSeeds", case_adj = c(0.5, 0.95), add.cv = TRUE)

# Read the reports back in from the experiment referred to in the PELFOSS Paper 1:
run <- runs[2]
r <- readPelfossReports(run, dir=dir)

# Write a table for use in supplementary materials:
columnsToKeep <- c(
	"Survey", 
	"Resolution", 
	"Year", 
	"Direction", 
	"Timing", 
	"Seed", 
	"TSB_meanByInsideAll", 
	"TSB_5percentByInsideAll", 
	"TSB_95percentByInsideAll", 
	"Ab.Sum.cv"
)
newNames <- c(
	"Survey", 
	"Resolution", 
	"Year", 
	"Direction", 
	"Timing", 
	"Seed", 
	"TSB_mean", 
	"TSB_5percent", 
	"TSB_95percent", 
	"TSB_CV"
)
rsmall <- r[columnsToKeep]
names(rsmall) <- newNames
head(rsmall)
outfile <- file.path(dir, "runs", run, "reports", "AllSurveysReportsSmall.txt")
write.table(rsmall, file=outfile, sep="\t", dec=".", row.names=FALSE, fileEncoding="UTF-8")

#######################
##### Diagnostics #####
#######################
##### Absolute relative change from seed 1 to seed 2:
summary(abs(dd$TSB_mean[dd$Seed == 2] - dd$TSB_mean[dd$Seed == 1]) /  dd$TSB_mean[dd$Seed == 1])

##### CVs for the 6 surveys:
surveys <- unique(dd$Survey)
CV <- sapply(surveys, function(x) summary(dd$TSB_CV[dd$Survey == x]))
colnames(CV) <- surveys
round(CV, digits=2)
#         Herring_IESNS Herring_NASSHS Herring_NASSHS_fishery Mackerel_IESSNS Mackerel_IESSNS_sept Mackerel_IESSNS_sept_fishery
# Min.             0.12           0.02                   0.01            0.08                 0.08                         0.05
# 1st Qu.          0.13           0.06                   0.02            0.13                 0.10                         0.11
# Median           0.18           0.08                   0.03            0.16                 0.12                         0.13
# Mean             0.18           0.08                   0.04            0.15                 0.22                         0.18
# 3rd Qu.          0.21           0.09                   0.04            0.18                 0.38                         0.29
# Max.             0.32           0.14                   0.08            0.19                 0.49                         0.35

##### Average relative estimates for the Herring_NASSHS shifted by +30 days in the normal and reversed direction:
mean(dd$TSB_mean[dd$Survey == "Herring_NASSHS" & dd$Direction == "Normal" & dd$Timing == "30days"])
mean(dd$TSB_mean[dd$Survey == "Herring_NASSHS" & dd$Direction == "Reversed" & dd$Timing == "30days"])

##### The range of the relative estimates for Herring_NASSHS_fishery versus the other herring surveys:
# Herring_NASSHS:
valid <- dd$Survey == "Herring_NASSHS"
range_Herring_NASSHS <- range(dd$TSB_mean[valid])
range_Herring_NASSHS
diff(range_Herring_NASSHS) / mean(range_Herring_NASSHS)

# Herring_NASSHS_fishery:
valid <- dd$Survey == "Herring_NASSHS_fishery" & dd$Timing != "-30days"
range_Herring_NASSHS_fishery <- range(dd$TSB_mean[valid])
range_Herring_NASSHS_fishery
diff(range_Herring_NASSHS_fishery) / mean(range_Herring_NASSHS_fishery)

## End(Not run)

Sea2Data/pelfoss documentation built on Dec. 7, 2020, 12:31 a.m.