Open douglatornell opened 2 years ago
Implementation for NEMO datasets will likely use NEMO-generated mesh mask dataset. Specifically, its mbathy
variable, which is the "fortran index of deepest water cell, T-grid".
Extracted dataset will be 3D (time, gridY, gridX)
because depth
value is potential different for every (gridY, gridX)
point.
Need to consider how easy it is to create a depth
variable in the dataset.
If it is not easy, it will be left to the user to provide those values if they need them.
If it is easy, then there is also the possibility of extracting multiple layers upward from the bottom. That would make the extracted dataset 4D (time, depth, gridY, gridX)
with the depth
coordinate having synthetic values like -1 for the deepest, -2 for the layer above that, etc. That also raises the problem of a name collision between depth
as a coordinate, and depth
as a variable. It also raises the problem of how to handle a request for more depth layers than there are water layers in shallow areas of the domain; maybe NaNs?
The 2nd point above seems quite complicated. So, initial implementation is unlikely to include depth
as a variable, nor multiple bottom layers in the extraction.
Feature request from Greig and others.