fix: vtec extrapolation calculation in TIEGCM

[hkershaw-brown]

Description:

Fixes problems in model_interpolate for VTEC

• result of 4 corners was not being stored correctly, overwriting corner11
• domain of ZG was incorrect
• units of ZG for need to be m in the vtec calculation

I've taken out the todo "test vtec" 🙃

Fixes issue

fixes #851

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update

Documentation changes needed?

• My change requires a change to the documentation.
  • I have updated the documentation accordingly.

Tests

None yet, just compiled user reported changes.
Derecho is down today.

Checklist for merging

• Updated changelog entry
• Documentation updated
• Update conf.py

Checklist for release

• Merge into main
• Create release from the main branch with appropriate tag
• Delete feature-branch

Testing Datasets

• Dataset needed for testing available upon request
• Dataset download instructions included
• No dataset needed

[mjs2369]

Only one question, a little unrelated to the PR, but concerns val11, 12, 21, 22

DART/models/tiegcm/model_mod.f90

Line 429 in f564745

bogus_level = 1 !HK what should this be? Do only 2D fields have VERTISUNDEF?

Looks like you made this comment 4 years ago and I share the same concern - if 1 is a valid value for level, then is 1 a good value for a bogus_level? It doesn't look like it really matters, since any 2D variables (that therefore have VERTISUNDEF) are not using the value for level at all when it is passed into get_dart_vector_index (note that level becomes kloc)

DART/assimilation_code/modules/io/state_structure_mod.f90

Lines 1626 to 1631 in f564745

	if (ndims == 1) then
	get_dart_vector_index = offset + iloc - 1
	else if(ndims == 2) then
	get_dart_vector_index = offset + iloc + (jloc-1)*dsize(1) - 1
	else if(ndims == 3) then
	get_dart_vector_index = offset + iloc + (jloc-1)*dsize(1) + (kloc-1)*dsize(1)*dsize(2) - 1

But just something to think about. Feel free to just tell me to disregard this - other than this I see no issues with the changes here so I'll go ahead and approve

[hkershaw-brown]

Only one question, a little unrelated to the PR, but concerns val11, 12, 21, 22

DART/models/tiegcm/model_mod.f90

Line 429 in f564745

bogus_level = 1 !HK what should this be? Do only 2D fields have VERTISUNDEF?

Looks like you made this comment 4 years ago and I share the same concern - if 1 is a valid value for level, then is 1 a good value for a bogus_level? It doesn't look like it really matters, since any 2D variables (that therefore have VERTISUNDEF) are not using the value for level at all when it is passed into get_dart_vector_index (note that level becomes kloc)

DART/assimilation_code/modules/io/state_structure_mod.f90

Lines 1626 to 1631 in f564745

	if (ndims == 1) then
	get_dart_vector_index = offset + iloc - 1
	else if(ndims == 2) then
	get_dart_vector_index = offset + iloc + (jloc-1)*dsize(1) - 1
	else if(ndims == 3) then
	get_dart_vector_index = offset + iloc + (jloc-1)*dsize(1) + (kloc-1)*dsize(1)*dsize(2) - 1

But just something to think about. Feel free to just tell me to disregard this - other than this I see no issues with the changes here so I'll go ahead and approve

Its hanging out from this comment:

DART/models/tiegcm/model_mod.f90

Lines 353 to 356 in f564745

	! @todo from Lanai code:
	! FIXME ... is it possible to try to get a pressure with which_vert == undefined
	! At present, vert_interp will simply fail because height is a negative number.
	! @todo HK what are you supposed to do for pressure with VERTISUNDEF? level 1?

I am not sure what the correct thing to do is still, or if there are calls to model_interpolate with VERTISUNDEF.

[mjs2369]

Its hanging out from this comment:

DART/models/tiegcm/model_mod.f90

Lines 353 to 356 in f564745

	! @todo from Lanai code:
	! FIXME ... is it possible to try to get a pressure with which_vert == undefined
	! At present, vert_interp will simply fail because height is a negative number.
	! @todo HK what are you supposed to do for pressure with VERTISUNDEF? level 1?

I am not sure what the correct thing to do is still, or if there are calls to model_interpolate with VERTISUNDEF.

Let's just punt on this for now then, the rest of the PR looks great

[nsc1]

my guess is that the level=1 code was put in before the state structure was used, and all three indices were used to compute the location in the state vector. it may not matter anymore because the state structure code knows about 2d fields.

also, the answer to the question on line 419:
! HK why does it have to be QTY_GEOMETRIC_HEIGHT?

it doesn't - but we had usually used something that was simple/fast to interpolate so the code didn't go through a long computation to find a value that wasn't going to be used other than a yes/no counter.