search

NOAA-EDAB / survdat

https://noaa-edab.github.io/survdat/
https://noaa-edab.github.io/survdat/
Other
2 stars 1 forks source link

Lengthweight #38

Closed slucey closed 3 years ago

slucey commented 3 years ago

Added in the ability to calculate the catch weight at length using the length-weight coefficients in SVDBS. Also created a helper function that will pull those coefficients into a table if needed.

andybeet commented 3 years ago

I am not sure we can use this table of coefficients since it doesn't report the variance of the fits. And we need that to estimate the fitted weights. Since the model is Gaussian on the log scale (log(W) = log(a) + b.log(L)), when we exponentiate back we need to add the term var/2. If you plot all of the lengths against the weights and then the fitted line you'll see it will be an underestimate.

Aside from this: the get_survey_data.r doesn't call the get_length_weight.r function that you created. I think it probably should to reduce duplicating that piece of code. Also the other get functions in the package return a data field, a sql field and a colNames field. I think the get_length_weight.r function should too. These output fields could then be used in get_survey_data. What do you think?

A possible workaround for all of this is to use our own length weight coefficients (ie update the Wigley et al paper). I have been working on this with Laurel and have these coefficients and variances for the 74 species in her paper.