Field gradient calculator about parcels HOT 3 CLOSED

This certainly is an interesting idea; it makes sense when the gradient needs to be calculated everywhere. But if the particles are only in a subregion of the domain, would it be overkill to calculate the gradient in the entire domain?

Anyway, perhaps this is the sort of optimisation that we can do in subsequent stages of development. Just like we can start optimising I/O then.

So for now, feel free to create a .gradient(time) function indeed.

As for the edges between NaN and real values, how are you getting along with landmasks, @Abobie? Would it make sense to make a PR for just the landmasks, so that they can be used here in this function too?

from parcels.

Bounding boxes to limit the calculated gradient could be further arguments in addition to time that default to the whole grid?

I've left out dealing with landmasks for the moment, but I assumed at the edges I'll calculate just the forward or backward differences- as I do at the edges of the grid.

from parcels.

Solved with #62

from parcels.