Constant latitude intersections using bounding box

[philipc2]

Closes #1018

Overview

• Uses the internally computed bounding box to compute the constant latitude intersections for each face

Usage

uxgrid = ...

# bounding box
uxgrid.get_faces_at_constant_latitude(lat=20.0, method='bounding_box_intersection')

# intersection of each edge
uxgrid.get_faces_at_constant_latitude(lat=20.0, method='edge_intersection')

[philipc2]

@hongyuchen1030 @amberchen122

This implements _get_candidate_faces_at_constant_latitude from #785 using the desired API.

Let me know if method='bounding_box_intersection' is appropriate for this.

[philipc2]

Feel free to merge any tests from #785 that are relevant to the candidate face computation also.

[hongyuchen1030]

@hongyuchen1030 @amberchen122

This implements _get_candidate_faces_at_constant_latitude from #785 using the desired API.

Let me know if method='bounding_box_intersection' is appropriate for this.

Hi @philipc2
Thanks for putting this up together. I am extremely busy this week for my paper and my QE. I will try my best to review all the recent PRs this weekend.

However, I don't think method='bounding_box_intersection' is the proper name.

In fact, this is the only method we mathematically calculate the intersections correctly.

All other "faster" methods didn't guarantee the intersection points and just gave some approximate results that "might work".

So I feel like we should name this one similar to "mathematical intersections" and other methods as similar to "tentative methods".

[philipc2]

@hongyuchen1030 @amberchen122

This implements _get_candidate_faces_at_constant_latitude from #785 using the desired API.

Let me know if method='bounding_box_intersection' is appropriate for this.

Hi @philipc2
Thanks for putting this up together. I am extremely busy this week for my paper and my QE. I will try my best to review all the recent PRs this weekend.

However, I don't think method='bounding_box_intersection' is the proper name.

In fact, this is the only method we mathematically calculate the intersections correctly.

All other "faster" methods didn't guarantee the intersection points and just gave some approximate results that "might work".

So I feel like we should name this one similar to "mathematical intersections" and other methods as similar to "tentative methods".

Thanks for the quick feedback! Good luck with your work this week!

[rajeeja]

Hey @philipc2 thanks for putting this together, would be the first step to determine the faces to perform area correction when edge is around line of constant latitude or great circle arc?

[philipc2]

Hey @philipc2 thanks for putting this together, would be the first step to determine the faces to perform area correction when edge is around line of constant latitude or great circle arc?

I don't believe this would require any area correction, as the bounding box implemented by @hongyuchen1030 already considers the spherical geometry.

Though area calculation is definitely something that down the line is important, as a lot of calculations are area-weighted.

[hongyuchen1030]

Hey @philipc2 thanks for putting this together, would be the first step to determine the faces to perform area correction when edge is around line of constant latitude or great circle arc?

I don't believe this would require any area correction, as the bounding box implemented by @hongyuchen1030 already considers the spherical geometry.

Though area calculation is definitely something that down the line is important, as a lot of calculations are area-weighted.

For non-conservative zonal average, we don't need to consider area calculation.

Area calculation is one of the last two steps of conservative zonal average.