Defining the master grid based on ROMS grid

andrew-edwards commented 1 year ago

This is probably the way to go, providing saved data sets don't get too big. Angelica's ROMS output gives the centres of each 3 km x 3 km grid cell.

[ ] Patrick's manuscript says: "The BCCM model outputs were interpolated from a curvilinear to a regular 3 km grid using a thin plate spline using the fields (Douglas Nychka et al. 2017) and raster packages (Hijmans 2022)." Not quite sure exactly what that means (though Angelica's was regular 3 km grid, just at an angle). Ask Patrick.
[ ] Patrick said to ask Jessica, and she gave me a link to her code from that above manuscript, which is adapting Angelica's grid.
[ ] Also see code from Lindsay Davidson.

andrew-edwards commented 1 year ago

Well, after useful lunch chat with Charles and Hannah.

[ ] May be better to do a 2x2 km grid (would match groundfish data, though I don't think that's overly important). But I think would enable the grid to be defined in PACea, which fits better with the existing code set up by Joe, than trying to work out a grid from the ROMS outputs. Do check with Angelica about that being done.
[ ] Also ROMS 3x3 km are not exactly 3x3 km cells, only approximations (which may mess up Joe's area based code).

andrew-edwards commented 1 year ago

Travis figured this out, using newer sf package.

pbs-assess / pacea