class: center, middle, inverse, title-slide .title[ # FIMS Case Studies ] .subtitle[ ## Review of
Fisheries Integrated
Modeling System ] .author[ ### Ian Taylor
NOAA Fisheries
Northwest Fisheries Science Center
ian.taylor@noaa.gov
] .date[ ### 2026/04/22 ] --- layout: true .footnote[U.S. Department of Commerce | National Oceanic and Atmospheric Administration | National Marine Fisheries Service] <!-- Start of slides --> --- # Outline - Background on case studies - Results of three case studies (switch to the case studies .hyperlink-style[[website](https://noaa-fims.github.io/case-studies)]) - .hyperlink-style[[Southern New England-Mid Atlantic Yellowtail Flounder](https://noaa-fims.github.io/case-studies/content/NEFSC-yellowtail.html)] original model in ASAP, converted to FIMS by Chris Legault - .hyperlink-style[[Gulf of Alaska Walleye Pollock](https://noaa-fims.github.io/case-studies/content/AFSC-GOA-pollock.html)] original model is a bespoke TMB model, converted to FIMS by Cole Monnahan - .hyperlink-style[[Pacific Hake](https://noaa-fims.github.io/case-studies/content/pacific-hake.html)] original model in SS3, converted to FIMS by Kelli Johnson - Lessons learned (back to this presentation) - Next steps --- # Background on case studies .left-68[ - Initial goal developed in 2022 was to have at least one case study for each NOAA Fisheries Science Center - Case studies repository created in December 2023, initially with Southern New England-Mid Atlantic Yellowtail Flounder thanks to Chris Legault - Multiple goals for the case studies, including: - testing FIMS features and identifying bugs - providing example models for users to learn from and build upon - comparing results between FIMS and existing models to build confidence in FIMS results - prioritizing features to add to FIMS based on differences between FIMS and existing models ] .right-32[
] --- # A success story: finding bugs / undocumented features in FIMS The first case study (Yellowtail Flounder in ASAP → FIMS) revealed several issues (all quickly resolved): - additional recruitment parameter was being estimated - had to fill missing survey data because NA was not allowed - sex ratio was not included in spawning biomass calculations - fixed sex ratio = 0.5 now correctly applied, and flexibility to include sex structure is a priority for 2026 (see .hyperlink-style[[https://github.com/ioos/gsoc/issues/114](https://github.com/ioos/gsoc/issues/114)]) Real world applications are messier than simulated data in terms of data availability and complexity and thus provide a valuable testing ground for FIMS development and debugging. --- # Results of three case studies switch to case studies website: .hyperlink-style[[https://noaa-fims.github.io/case-studies/](https://noaa-fims.github.io/case-studies/)] --- # Lessons learned - When comparing models with features not yet available in FIMS, it can be helpful to revise the original model to be closer to the available options within FIMS - isolates unknown differences in implementation from known differences in model structure - helps identify which features are most important to add to FIMS in the future to bring the FIMS case study closer to the original model - Keeping the case studies up to date with the changes in FIMS is challenging but valuable - provides a reminder for developers of needed changes to the user interface - provides example models for users to learn from and build upon - helps identify and fix bugs in FIMS --- # Next steps - More case studies are currently in development: - Pacific Sardine (SWFSC, SS3, Alex Jensen and Peter Kuriyama) - Opakapaka (PIFSC, SS3, Meg Oshima) - Petrale Sole (NWFSC, SS3, Ian Taylor) - South Atlantic Scamp (SEFSC, BAM, formerly led by Kyle Shertzer) - Add additional case studies - a WHAM model (thanks to Emily Liljestrand) - an AMAK model (thanks to Jim Ianelli) --- # Next steps (continued) .left-68[ - Generalize code used to convert data for case studies into conversion functions that can be used by others - e.g., `ss3_data_to_fims()` converter now available in the {r4ss} package - standardize language across models (e.g., more alignment between column names between FIMS and RCEATTLE to ease the conversion process) - Continue to use case studies as a testing ground for new features in FIMS, including growth estimation and time-varying selectivity - As FIMS feature set expands, move away from simplifying the original models ] .right-32[
] --- # Next steps (continued) - Standardize plots of model output, diagnostics, and comparisons for the case studies by moving to the {.hyperlink-style[[stockplotr](https://nmfs-ost.github.io/stockplotr/)]} package - Refine the selection of case studies once FIMS starts being applied in production assessments (while also maintaining a list of those applications) - Point for discussion: what would be the best collection(s) of case studies to maintain for the long term? - curated examples spanning range of features and complexity? - comprehensive U.S. repository of production stock assessments (easier)? - global repository (how to get folks to contribute)? - models used in research projects? --- <div style="text-align: center;"> <img src="static/FIMS_hexlogo.png" style="width: 100%; max-width: 300px;" /> </div> 🌐 .hyperlink-style[[https://noaa-fims.github.io/case-studies/](https://noaa-fims.github.io/case-studies/)] 📩 .hyperlink-style[[ian.taylor@noaa.gov](mailto:ian.taylor@noaa.gov)]