DataStructure

[philippemiron]

The code is working fine now, but it could be made simpler.

• How to store the xgcm.grid objects (velocity fields, and trajectory grids)?

Could have a class xlcs that contains:

• parameters of the trajectory (x,y,dx,dy,T)
• build the grids automatically
• then you can call flowmap() operation without having to pass the grid object

I think that creating a class might be overkill and would make it harder to use the method independently. Have to think about this more carefully...

[philippemiron]

Create independent functions for ftle, lavd, and elliptic vortices.

Each of those functions should take a parameter dictionary and the velocity field store in a xr.Dataset. In one line, we should be able to obtain the scalar across the domain and/or the extracted contours. Fill parameter with default value if not present.

FTLE

• add the trajectory grid, flowmap, eigvalues, eigenvectors to the initial dataset.

LAVD:

• add trajectory grid, flowmap, lavd
• even if flowmap was previously calculated we have to sample the vorticity
• contour as a separate Dataset? Might be better to return as an different object since it has nonuniform length

Elliptic LCS

• same as FTLE
• extracted contours

This seems like a good structure and the function that are call can be general enough to be reused between the 3 use cases..