STAR Major Uncertainty #3: Value of surveys

[brianlangseth-NOAA]

From the 2007 report

"The triennial survey is inefficient for canary; hence, this assessment is really predicated on catches, low natural mortality, and the assumed value of steepness. The survey data and compositional information may not permit reliable estimation of stock status."

Additional future research was suggested as well: -"the possibility of a seasonal effect on q from the triennial survey should be evaluated." -"Investigate the importance of calendar date and other covariates on catch rates from the triennial survey and propose adjustments to account for seasonal and other variation in selectivity/availability."

From the 2015 report

It is evident that the model's reliance on abundance trends from indices is small. Model results would be considerably improved if a reliable abundance index for older fish was available. The WCGBTS survey is currently the best available source of abundance information for older fish in recent years as a direct measure of the extent of rebuilding, although it has been recognized that the survey infrequently encounters canary rockfish, and occasional large catches occur when canary aggregations are encountered.

Update (2/1) based on pre-assessment workshop Apply a sensitivity with triennial combined (one time series) to see if split is really needed. Can apply sensitivity of removing it as well.

[brianlangseth-NOAA]

Much of the issues with the triennial survey are, in my opinion, less relevant given that in 2007 that was really the only survey. Now in 2023, we have the WCGBTS.

In the 2015 assessment, they included Julian date as a covariate in the the standardization for the triennial. See that report for their response. We can do similarly.

~- [ ] Include julian date within the standardization of the triennial survey.~ UPDATE (2/1) I dont see this as happening so crossing out.

[kellijohnson-NOAA]

It would be my preference to not include Julian date at the onset of building an index standardization model for the Triennial or any other data set. I am guessing that using a spatio-temporal field will be sufficient given that the previous model used stratified areas rather than random fields. But, I completely agree with @brianlangseth-NOAA that the 2007 issues are less relevant now that WCGBTS is longer. I think the bigger ticket item here is what distribution will accommodate extreme catch events the best, which I think @chantelwetzel-noaa has much more experience with than I do.

[okenk]

From 2015 assessment research and data needs: "we highly encourage a coast-wide pilot study for an alternative sampling method (e.g., hook-and-line sampling), as well as its calibration against the existing bottom trawl survey via paired sampling methods"

Unsurprisingly, this did not happen.

[chantelwetzel-noaa]

Spoiler: modeling will not "fix" how terrible the the triennial survey is for many rockfish species. This survey was not initially designed for rockfish species, plus the survey designed changed every year through 1995. This survey was so bad in the Pacific ocean perch 2017 assessment, the STAR panel recommended and the STAT agreed to remove this time-series. This decision led to a whole new can of worms with the SSC and I would not recommend removal given this. I would advocate not spending too much time on the triennial time-series but rather focusing that time and effort on the WCGBT survey.

[brianlangseth-NOAA]

• [x] There is a groundfish subcommittee meeting this week to look at hook at line surveys in Oregon and Washington as well as the ROV survey in Oregon. These may or may not be relevant to canary, but something to at least monitor. UPDATE (2/1) Issue #39 includes relevant surveys from that workshop.

[brianlangseth-NOAA]

• [x] Can talk to John Field about juvenile rockfish survey, could this be an actual survey data source or more qualitative, does it have sufficient taxonomic resolution, is definitely more spatially constrained than the assessment (~Central California as I recall)