Currently this is hard-coded for the fin whale example, filenames etc are set in each of the files 0 through 4, which should be run in order. Once these have been run the fig_* files can be run.

• Files:

  • 0_format_aux_data.R - do some additional data manicure.
  • 1_model_and_data.R - setup the data and fit the spatial model.
  • 2_prop_that_var.R - propagate the detection function uncertainty through to the GAM using the Bravington et al. method.
  • 3_big_uncertainty.R - simulate from the model over multiple "days" (more generally time periods) with different environmental covariates.
  • 4_process_uncertainty.R - take the outputted simulated surfaces and calculate summaries (mean/variance per cell etc) overall and per year.
  • fig_Nhat_timeseries.R - generate timeseries for abundance.
  • fig_gam_terms.R - plot the smooth terms in the GAM.
  • fig_maps.R - plot maps of mean and standard error overall and per year.
  • fig_plot_df.R - plot the fitted detection function.
  • fig_plot_raw_data.R - plot the segments with bathymetry background.
  • X_investigate.R - investigate the 2008 high uncertainty.
  • support_scripts/get_covar_bnds.R - find the max/min of the covariates.
  • support_scripts/get_varprop_model.R - cut-down version of dsm::dsm_varprop.
  • support_scripts/obs_exp.R - compare observed vs. expected values for the model vs. data.
  • support_scripts/prepare_preds.R - take ROMS grids and format them for predicting.

• Input/output file folders:

  • RData/ - holds intermediate model outputs as RData files, this includes fitted spatial models, variance propagated models etc. (empty to begin with).
  • data/ - holds the "raw" data including the segment data, the fitted detection function object, average group size table, CCE grid.
  • data/roms_grids - ROMS grids should be placed here, one for each time period that needs to be predicted.
  • figures/ - output figures from fig_*.R scripts go here (empty to begin with).
  • out/ output grids will be stored here. One file per time period, name format is as for the ROMS files. Timeseries are stored in Nhat_ests.csv.

Worth running the following before running anything above!

install.packages(c("mgcv", "Distance", "dsm", "data.table", "lubridate",
                   "patchwork", "ggplot2", "mapdata", "tidyr", "dplyr",
                   "mrds", "RColorBrewer", "cmocean", "mapproj"))