Regridding Tests

[jatkinson1000]

This is a sub-issue of #64 specifically focussing on the regridding tests.

A plan of action is to:

• [x] Review current tests, how they work, and what passes and fails.
• [ ] Define what tests we want to write:
  • Interpolate between identical grids
  • Interpolate from a fine to a coarse grid
  • Interpolate from a coarse to a fine grid
  • Handling when one grid is 'higher' than the origin grid
• [ ] What do we compare
  • For conservative regridding we want to compare the integral in the column
• [ ] How do we test multiple columns
• [ ] How do we setup and define test arrays - this could probably be improved from what is currently there.

[jatkinson1000]

The test test_interp_to_cam_match currently is a specialised test where the CAM and SAM grids are identical AND we use density of 1.0 across the entire column. We then test that all interpolated cells have a density of 1.0.

We could write a more generic test that takes cells with variable density, but checks that total mass is conserved between columns (by summing var_cam and var_sam across the height). This approach could be used in other subsequent tests as well.

[jatkinson1000]

At the moment we always assume the CAM interface pressure is equal to the bottom pressure of the column. Can we test when this is not the case in a separate test perhaps?

[jatkinson1000]

We currently have interpolation tests for matching SAM and CAM grids, and for when CAM is finer than SAM.

We need to add a test for when CAM is coarser than SAM (and extends higher?) as this is the realistic case we are dealing with.

[jatkinson1000]

Some variables could be made module variables and better documented:

• 4 could be replaced by num_cols and applied to all tests as a module variable.

[jatkinson1000]

The setup of the coarse grids in some tests using a reshape is very unintuitive. It would be much better to do so using a loop over column levels and setting the values at each level. e.g. https://github.com/m2lines/convection-parameterization-in-CAM/blob/1adf4ce02eeda6d78c48577211fcfa1d1c783962/tests/test_CAM_interface/test_cam_interface.F90#L78C1-L78C124

[AdelekeBankole]

Some variables could be made module variables and better documented:

• 4 could be replaced by num_cols and applied to all tests as a module variable.

This was addressed in the earlier commit.

[AdelekeBankole]

The setup of the coarse grids in some tests using a reshape is very unintuitive. It would be much better to do so using a loop over column levels and setting the values at each level. e.g. https://github.com/m2lines/convection-parameterization-in-CAM/blob/1adf4ce02eeda6d78c48577211fcfa1d1c783962/tests/test_CAM_interface/test_cam_interface.F90#L78C1-L78C124

This was also addressed.