search

pacific-hake / hake-assessment

:zap: :fish: Build the assessment document using latex and knitr
MIT License
13 stars 6 forks source link

Understanding increase in 1980 recruitment from last year's assessment #1104

Closed andrew-edwards closed 10 months ago

andrew-edwards commented 10 months ago

Not really an issue, but doing here rather than clogging up Slack. So here's two figures for recruitment scaled by 2010 recruitment: 2023 assessment: image

2024 assessment: image

Look at 1980. Much higher (relatively) now than for last year. Here are the numbers:

Billions of fish, median and 95% intervals:
                    1980 recruitment     2010 recruitment:
2023 assessment:    17.1 (10-34          16.9 (11-30)
2024 assessment:    19.7 (12-37)         15.8 (11-27)

So 1980 has gone up and 2010 has gone down a bit, hence the change in relative values (looking at medians, credible intervals in that table still mostly overlap). I think this is caused by the penultimate bridging step, adding in the modelled temporal weight-at-age, see the red to blue step here (green then mostly covers blue), from Fig 15. Slight increase (but on log scale) for 1980, slight decrease for 2010. image

So these slight changes in rec deviations seem to translate to changes in billions of fish (haven't done the math). Just trying to tease out the changes.

aaronmberger-nwfsc commented 10 months ago

Yep, agree with your assessment.

andrew-edwards commented 10 months ago

Great. And such changes aren't "bad", because as Kelli has pointed out, we're moving to better assumptions for weight-at-age.

andrew-edwards commented 10 months ago

Closing. Have included the numbers in Section 3.4.2. Presumably the extra 2 billion 1980 recruits will be further explained by the new time-varying weight-at-age and maturity values.

andrew-edwards commented 9 months ago

Just updating these numbers for the record, as am editing the related text: Billions of fish, median and 95% intervals:

                    1980 recruitment     2010 recruitment:
2023 assessment:    17.1 (10-34)          16.9 (11-30)
2024 pre-SRG:       19.7 (12-37)          15.8 (11-27)
2024 post-SRG:      17.7 (10-35)          16.0 (11-29)

So no longer have a change of 2 billion fish born in 1980.

Consequently the post-SRG scaled recruitment plot shows 1980 again slightly larger to 2010 (moreso than last year's assessment, but not as much as the pre-SRG plot above).

Upshot is I can just delete the explanation in the text, esp with such large credible intervals). ![Uploading image.png…]()