Total biomass/numbers checks

mike-levine commented 2 months ago

Hey @abigailatsea ,

I am going to start using this package in the winter GOA reports, so figured I'd do some basic checks.

I basically just summed up three sources for the pre-selectivity corrected era:

Your new package
The dataset that is included in the current 2024 cruise report
The tables from the 2023 SAFE report.

I put the results in: G:\MikeLevine\testing\test_goadata_package\compare_sources.xlsx There's a tab for 'by length' comparisons, and a tab for 'by age'. Things look really good overall! A few notes/questions

Comparisons are % change from the report or the SAFE to the new package values. I highlighted them in green if <1%, yellow if a few %, red, if a lot %. The 'red' cases are all pretty clearly mistakes in the dataset that is in the report (the SAFE and the package data agree). So it will be good to clear those out and replace with better historical values! I think the package data is 'correct' and we should trust it, and it is really similar to the SAFE values.
And a few requests, maybe:
1. Is it possible to get the Shumagins 1994/1995/1996 values? These exist in the report dataset but not the package dataset.
2. In WinterGOAHistorical::PreSelectivityWinterGOAages, the regions are a little funky- any reason that 1995, 1996, and 1997 are 'other'?
3. When you queried these values, is transect width something that is easily added? I don't really know the structure of macebasehistoric, or how the queries are written. If its as easy as adding a word to the query, that would be great. If you'd have to do some complicated thing to assign widths based on vessel log, etc., it isn't worth it.

Thanks, this seems really close, and really helpful!

abigailatsea commented 2 weeks ago

I had to put this dev aside to focus on the EBS, working to address all of the issues now.

abigailatsea commented 2 weeks ago

I updated the script to fetch historical ages so it matched with the one to get historical lengths, which fixed the region == 'other' issue. The Shumagins surveys only go back to 2001 (!!!) in MB2. We briefly discussed this, and IIRC decided to not try to resurrect the pre-2001 Shumagins surveys because they exist just as csvs in old cruise folders or on MB historic.

Looking into Q3.

shelikof_surveys_1995_2007

mike-levine commented 3 days ago

Ok sorry for the delay, just getting back to this part of the cruise report.

In the version of the package I downloaded just before swapping to a private repo, could you check what data you've got? For some reason I only get two surveys now (199504 and 200102). So for example in the old data package for the Shelikof I got:

and in the update:

Thanks for checking @abigailatsea! It could just be wackiness related to the swap to private? I'm not sure.

abigailatsea commented 3 days ago

Yes, I saw that, realized my mistake (I compiled the data in test mode with just two surveys), and the full data set is in the new package. I'm going to make you an admin on the updated version and see if that helps with the install issue.

mike-levine commented 3 days ago

Cool- I think this might argue for going back to public mode! You probably don't want to have to control access to each person in the group? That's sure been an unanticipated side effect of the move.

abigailatsea commented 3 days ago

You know what? Let me try making it public again and seeing if it works properly!

abigailatsea commented 3 days ago

@mike-levine Can you try now? I made it public and re-published the website.

mike-levine commented 3 days ago

Yep I could install from that version no problem!

abigailatsea commented 3 days ago

Yay! Does it look ok?

mike-levine commented 3 days ago

I'll check things out next week more and get back to you on that, thanks for sorting out the install!

noaa-afsc-mace / WinterGOAHistorical

Total biomass/numbers checks #2