Include the BCCM output for its whole domain

[andrew-edwards]

To save storage space, we clipped the BCCM output to only cover the Canadian Exclusive Economic Zone.

Request from Lisa Crozier (@lisa439), arising from our DFO/NOAA meeting, to include the US waters, which should reach down to the Columbia River.

Given Travis figured out how to store the EEZ model outputs, I think the best way to do this would be to just save the full BCCM output in another data repository, such as pacea-data-full-bccm . Keep the structure of everything that's saved the same, and just have the variable names appended by _full so that we can have both sets of results stored locally on our computers. Then, the existing functions and vignettes should still work with variables appended by _full, depending on how automatically maps etc. get resized.

We may want to create a new GitHub organisation, given the email we got about Large File Size and being at 80% of capacity. This wouldn't matter as users don't actually need to interact with the data repository page, since Travis's functions call it directly when first downloading the BCCM results.

This will likely end up being more complex than outlined above, but just getting some ideas here for now.

[andrew-edwards]

Similar request from Eva and Brianna:

Brianna and I are hoping to use the BCCM data from the pacea package for some spatial modelling but our model extent goes beyond the Canadian EEZ (specifically we need to include WA state waters). Is it possible to remove the EEZ mask when calling pacea::bccm_surface_temperature() to get the full geographic extent of Pena et al.’s (2019) model predictions?

[andrew-edwards]

Here's the full extent of the BCCM model:

[andrew-edwards]

@LisaCrozier-NOAA - I am working on this now.

So for the BC waters we made a 2x2 km inshore grid and a 6x6 offshore grid, to save storage space. I just want to check if a similar thing is sufficient for your needs, with the grid extended down to the US? Basically, covering the full domain of Angelica’s model.

[andrew-edwards]

Here's the current 2x2 inshore grid (green) and 6x6 offshore grid (blue) currently used for restricting the BCCM outputs to BC waters. So the idea would be something similar extending south (to save storage space, I think 2x2 everywhere will get unwieldly).

And see the BCCM vignette for what functionality we have built in already: https://htmlpreview.github.io/?https://github.com/pbs-assess/pacea/blob/main/vignettes/bccm.html My aim is to make all this work fine for the full domain results.

[andrew-edwards]

That figure didn't upload it seems. A better one is this, with pink for what the full BCCM grid is:

[andrew-edwards]

Okay, I may just stick with 2x2 throughout as it's a bit simpler to deal with.

Filesizes will be large (but only need to be downloaded once), so I've been investigating ways of hosting them on Zenodo rather than GitHub and adapting some of our existing R functions to download them. Have successfully saved the .rds file for bccm_avg0to40m_temperature_full, which has 161,000 cells for each month from 1993 to 2019 (not usable yet by users). File is 120Mb, and there will be about 20 similar files.

Then all the functionality in Travis's BCCM vignette will work, such as:

plot(bccm_avg0to40m_temperature_full)

[andrew-edwards]

Have locally built all the BCCM files (for different variables) at 2x2 km resolution over the full original domain. Need to upload then download them, but here's an example of what will be easy to look at (primary production for each month in 2018):

plot(bccm_primaryproduction_full, months = c(1:12))

[andrew-edwards]

Done in the above commit. See the new vignette.