search

aodn / nrmn-application

A web application for collation, validation, and storage of all data obtained during surveys conducted by the NRMN
GNU General Public License v3.0
4 stars 3 forks source link

Update length-weight biomass coefficient with values from Fishbase #1233

Closed bpasquer closed 1 year ago

bpasquer commented 1 year ago

Follow up on https://github.com/aodn/backlog/issues/4644

Dump of data retrieved from github : https://github.com/ropensci/rfishbase/tree/master/parquet/fb_parquet_2021-06

After checking multiple values, it appears that there are some discrepancies in the value of A's and B's as compared to what is displayed on the Fishbase website, which suggest the dump is not be the most recent version of the data

The resulting file of Fishbase LW data mapped to NRMN observable item : Mapped_FishBase_LW.csv

@atcooper1 @LizziOh Please inspect the data and let us know if we can use this material to run a bulk update .

atcooper1 commented 1 year ago

Thanks @bpasquer - have circulated to RSS, GE and NB for comment/confirmation and will advise whether they are happy to move forward with FB values as soon as i hear back.

atcooper1 commented 1 year ago

Hi @bpasquer, thanks for progressing with these! The A's and B's are quite different as you say, but this might be ok due to the way biomass is calculated. We were wondering if it would be possible at all to run these new Fishbase a's and b's against a tropical and temperate location (i.e. Encounter and GBR- North) so I can compare these against our current values and see how these will affect the biomass calculations? I think everyone will agree to move forward with the Fishbase values, but it would be good to check this on a small sub-set to ensure we aren't creating more issues.

bpasquer commented 1 year ago

Yes sure @atcooper1, understanding the impact of the change is a very reasonable request.

atcooper1 commented 1 year ago

Hi @bpasquer I've taken a look at the biomass values and the FB values generally look better, but there are a few funky things I've noticed (e.g. the leatherjacket values seem out). Can I confirm whether these biomass calculations include the diver bias calculation that is applied to the existing biomass values?

bpasquer commented 1 year ago

yes, the biomass calculations are done using the same code as the current calculation, and include use the diver bias rule

bpasquer commented 1 year ago

The table below compares a and b values before(in prod replica) and after(in verification) the Fishbase update Some value do change a fair bit with biomass calculation result that can almost double

obsrvable_item_id observable_item_name common_name New a (in verification) Old a (in prod replica) New b (in verification) Old b (in prod replica)
2440 Cantherhines fronticinctus Spectacled Leatherjacket 0.021379599347711 0.0561 2.92000007629395 2.653
2441 Cantherhines dumerilii Barred Leatherjacket 0.025118900462985 0.0561 2.92000007629395 2.653
2442 Cantheschenia grandisquamis Largescale Leatherjacket 0.021379599347711 0.0256 2.92000007629395 2.7
2443 Cantheschenia longipinnis Smoothspine Leatherjacket 0.009999999776483 0.0684 3.03999996185303 2.563
2444 Paraluteres prionurus Blacksaddle Filefish 0.021379599347711 0.0256 2.92000007629395 2.7
2445 Pervagor alternans Yelloweye Leatherjacket 0.021379599347711 0.025 2.92000007629395 2.946
2446 Pervagor janthinosoma Gillblotch Leatherjacket 0.02089300006628 0.025 2.92000007629395 2.946
4423 Pervagor aspricaudus Orangetail Leatherjacket 0.021379599347711 0.025 2.92000007629395 2.946
4910 Pervagor melanocephalus Blackhead Leatherjacket 0.021379599347711 0.025 2.92000007629395 2.946