POPSIM-A as an operating model for testing North Pacific chub mackerel stock assessment models

This report was written under contract to assist the North Pacific Fishery Commission (NPFC) and its Chub Mackerel Stock Assessment Working Group (WG) with work testing stock assessment models for chub mackerel ( Scomber japonicus ). Paraphrasing the consultancy agreement, I mean to attend a review meeting, participate in discussions and 2 provide and demonstrate the following products:
a) Preliminary POPSIM-A input files for simulating a hypothetical chub mackerel population and fishery for one demonstration scenario based on a Japanese a Virtual Population Analysis (VPA) provided to me. Done, as described in this report. Both will be distributed at or before the 2020 TWG CMSA03 meeting via Commission staff or as staff direct.
b) Functions in R or POPSIM-A for saving simulated population variables and pseudo-data in one or more databases. Same as a).
c) A database with simulated “true” values and a database with pseudo-data from the demonstration scenario. Same as a). POPSIM-A outputs a data structure with true values automatically. R software generates the database with pseudo-data from POPSIM-A output.
d) R software for extracting POPSIM population variables and pseudo-data from database(s). Same as a). Very little software is required because simple R commands are used. The subset() function, in particular, works well.
e) Model performance results for demonstration scenario(s) and assessment models that can be called directly from POPSIM. Not completed because it is no longer possible to call current versions of the assessment models from POPSIM-A (see this report). Members will have to write their own software for testing the assessment models they are interested in using. Such software will extract pseudo-data from the database building input files, running models,
collecting model estimates and evaluating model performance. These tasks can all be done in R and the R software provided is designed to facilitate the work. Organization of WG efforts and the structure of a database for sharing assessment model test results are important topics to address. In an effort to enhance progress, I demonstrated R code at the review meeting for testing a “silly” assessment mackerel for chub mackerel either fitting the model entirely in R or
by calling an AD-Model Builder program that runs in DOS (Appendices 9-11).
f) A written report and detailed presentation at the March 2020 assessment meeting. The presentation will be interactive to demonstrate the use of software for simulation and testing and how scenarios can be modified. Report same as a). Presentation to be completed at the meeting. I expect to modify the report, software and other materials at the meeting based on discussions. 
h) Complete documentation for running POPSIM, documentation useful in linking POPSIM to new assessment models, description of new software and any changes made to existing software, along with advice about using POPSIM for chub mackerel work. Same as a). The documentation in this report supplements help files distributed with POPSIM-A that members will still need to read.
i) New POPSIM code if any modifications are made. No changes were made to POPSIM-A (see text for advice re future modifications).

Associated files are accessible on the Collaboration website https://collaboration.npfc.int/node/21