South Pacific ALB 2024 CPUE Analysis

Two analytical steps

1. Fit model
2. Calculate predictions

The predictions are calculated with the get_index() function, weighted by area. Some 5x5 rectangles are larger than others, depending on the latitude. The prediction uses a regular grid called new.data.

Spatial resolution

The observed data and model fit are based on 1x1 degree rectangles.

Model prediction is based on 5x5 degrees, used for calculating the weighting of length data and for calculating the relative biomass indices.

Overview of scripts

VAST analysis

For the ALB 2024 CPUE analysis, Thom first reran the 2021 VAST analysis that Tiffany had done. As documented in Step_by_step_instructions_for_ALB_CPUE_standardization.R, these are the scripts involved in the VAST analysis:

• Run prep.ops.data.r to prepare the raw data
• Run plot.spatial.summaries.r to plot summaries of data
• Run alb_filter_data.r to filter the data for albacore
• Run target_cluster.r to estimate targeting cluster (target species)
• Run alb_cpue_run.r to perform cpue analysis in VAST

Note: there is some data preparation (sub-sampling) that happens in alb_cpue_run.r

• Get results using get_indices.r

sdmTMB analysis

Next, Thom developed scripts to conduct sdmTMB analysis of ALB 2024 CPUE, loosely based on YFT 2023 CPUE analysis.

One difference between VAST and sdmTMB is that sdmTMB uses a mesh that goes well beyond the geographic extent of the data.

These are the scripts to run the sdmTMB analysis:

• Run fit.sdmTMB.spawn.R to fit and predict the spawning region indices
• Run fit.sdmTMB.global.R to fit and predict the global indices
• Run calc.prop.knots.sampled.r to get spatial coverage plot
• Run plot_mesh.r

Encounter probability is used when a delta model is fitted and the data includes zero-catch observations.

Hooks between floats: check with Thom.

Other scripts

Script	Purpose
alb_add_regions.R	VAST code needed somewhere
alb_cpue_prep.R	VAST code needed somewhere
compare.catch.quarterly.R	Plot catch and effort time series by region
fit.sdmTMB.vessel.R	sdmTMB using vessel ID as fixed effect, not used
fit.sdmTMB.vessel.RE.R	sdmTMB using vessel ID as random effect, not used
select_core_vessels.R	Experiment with removing some 5x5, not used
get_ENSO.R	Get ENSO data from web (might want to keep month)
get_indices.R	Compare VAST and sdmTMB results