Dynamic range models improve the near-term forecast for a marine species on the move

This repository is organized as follows:

• raw-data contains any data that were not generated by us, in the form they were made available in (so not necessarily "raw", but not processed in this repository)
• processed-data contains data written out by code in this repository
• functions contains user-defined functions
• src contains source code for Stan models
• results contains results written out by the scripts

Scripts are all in the main project folder.

To run models for summer flounder, follow these steps:

1. Download the 2020 OceanAdapt release from this stable DOI: 10.5281/zenodo.3885625
2. Run 01_import_summer_flounder.R to pull in and filter these data
3. Run 02_get_summer_flounder_params.R to mine parameters from summer flounder stock assessment tables
4. Run 03_prep_stan_data.R to reshape the data for the DRM
5. Run 04_run_drms.R to actually run the models. Be warned they are memory- and storage-intensive, and set up to run on a HPC.
6. Run 05_convergence_checks.R to summarize diagnostics for the models.
7. Run 06_model_evaluation.R to iterate over converged models, create alternative models (SDMs, persistence forecast), and quantify model skill.
8. Run 07_thin_posteriors.R to thin the Stan outputs and make them tractable files. (You can skip this; the outputs of it are on GitHub, so you can still examine the thinned posteriors without running any Stan code or having the full posterior files, which are large.)
9. Run 08_paper_stats_and_figures.R to generate the figures and in-text statistics in the manuscript.
10. Run 09_supplement.Rmd to generate the supplementary materials for the paper.