Enable Option for High Res Mesh in Ocean #894

alexolinhager · 2025-02-21T17:43:57Z

This PR creates the option (max_res_in_ocn=True) to maintain the highest resolution used on land ice cells in all ocean cells when generating the GIS MALI mesh. This is done to resolve fjord geometry and coastal bathymetry necessary to translate ocean properties to glacier termini when coupling MALI to MPAS-Ocean (or prescribing ocean forcing to MALI in standalone mode). As this has the potential to drastically increase mesh generation time, so there is an additional option define_bnds_by_geojson=True that performs an initial cull of the BedMachine dataset according to the bounds set by compass/landice/tests/greenland/gis_contShelfExtent_PS.geojson before generating the mesh in Jigsaw.

Checklist

User's Guide has been updated
Developer's Guide has been updated
API documentation in the Developer's Guide (api.rst) has any new or modified class, method and/or functions listed
Documentation has been built locally and changes look as expected
Document (in a comment titled Testing in this PR) any testing that was used to verify the changes

compass/landice/tests/greenland/mesh.py

trhille · 2025-08-04T23:15:48Z

compass/landice/mesh.py

    high_bed = section.getfloat('high_bed')

-    # convert km to m
-    cull_distance = section.getfloat('cull_distance') * 1.e3


We definitely don't want to remove culling by distance entirely.

This call of cull_distance is only used to assign large widths for cells that are to be culled later on (to speed up efficiency). I think Matt and I decided this created more problems than it was actually worth. The actual culling still happens here:
https://github.com/MPAS-Dev/compass/blob/b3d1faba9502e38209b3376b2241c35b9dbffa54/compass/landice/mesh.py#L713C5-L722C63

Okay, thanks for pointing that out. I don't know that I necessarily agree with this choice as a general decision for all meshes going forward. I can imagine a ton of wasted compute time trying to make a high-resolution AIS or GrIS mesh that creates high-resolution cells hundreds of km away from the ice sheet, only to cull them away afterwards. I think we need some demonstration that this doesn't increase cost in general too much.

trhille · 2025-10-27T22:51:34Z

@alexolinhager, this is missing an updated mesh_gen.cfg file.

trhille

@alexolinhager, thanks for this really useful PR. I have a number of questions and thoughts, mostly regarding generalization. I've also pointed out a few logical inconsistencies that should be easy to fix.

trhille · 2025-10-27T22:50:29Z

compass/landice/mesh.py

+        # thermal forcing parameterizations. Otherwize, telescope out
+        # to coarse resolution with distance from the ice front.
+        if section.get('max_res_in_ocn') == 'True':
+            spacing_bed[np.logical_and(bed < 0, thk == 0)] = min_spac


It seems restrictive to impose that the ocean resolution is the same as the minimum ice-sheet resolution. If the ocean model resolution is >10km and we have ≤1 km minimum ice-sheet resolution, then that seems like a lot of wasted cells. What about explicitly setting the desired resolution in the open ocean and then either defining a length-scale over which the resolution varies from the ice edge outwards, or defining an open-ocean mask with a geojson file?

trhille · 2025-10-27T22:54:16Z

compass/landice/mesh.py

+            spacing_bed[np.logical_and(bed < 0, thk == 0)] = min_spac
+            print("Maximizing resolution in ocean. {} ocean cells \
+                  ".format(np.sum(np.logical_and(bed < 0, thk == 0))))
+        else:


This should not be within an else-block, or we will not be able to set maximum resolution in the ocean and have bed topography influence resolution for the ice sheet.

trhille · 2025-10-27T22:56:56Z

compass/landice/mesh.py

    high_bed = section.getfloat('high_bed')

-    # convert km to m
-    cull_distance = section.getfloat('cull_distance') * 1.e3


Okay, thanks for pointing that out. I don't know that I necessarily agree with this choice as a general decision for all meshes going forward. I can imagine a ton of wasted compute time trying to make a high-resolution AIS or GrIS mesh that creates high-resolution cells hundreds of km away from the ice sheet, only to cull them away afterwards. I think we need some demonstration that this doesn't increase cost in general too much.

trhille · 2025-10-27T22:58:34Z

compass/landice/mesh.py

-        bnd = section.get(option)
-        if bnd != 'None':
-            bnds[index] = float(bnd)
+    # If necessary, get bounds defined by user or use bound of gridded dataset


I'm confused about why this needs to be changed.

trhille · 2025-10-27T23:00:45Z

compass/landice/mesh.py


-    return (cell_width.astype('float64'), x1.astype('float64'),
-            y1.astype('float64'), geom_points, geom_edges, flood_mask)
+    if not calc_geom_bnds:


I think it's a bad idea to return a different size tuple based on the input arguments. If it's really necessary to put the if-statement above, it would be better to have an else-statement that just sets geom_points, geom_edges = None, None and keep the return statement simple.

trhille · 2025-10-28T01:43:33Z

compass/landice/tests/greenland/mesh.py

-                self, section_name=section_name,
-                gridded_dataset=source_gridded_dataset_2km,
-                flood_fill_start=[100, 700])
+        if section_gis.get("define_bnds_by_geojson") == 'True':


As noted above, I'd rather that build_cell_width return None, None for geom_points, geom_edges here, and then move the call set_geojson_geom_points_and_edges() after the call to build_cell_width to get those arrays.

trhille · 2025-10-28T01:51:08Z

compass/landice/tests/greenland/mesh.py

        ds.to_netcdf(self.mesh_filename, 'a')
+
+
+def set_geojson_geom_points_and_edges(self):


This should be generalized and moved to the landice.mesh module.

Creates the option to maintain resolution in the ocean. Used to resolve fjords and bathymetry when coupling to MPAS-Ocean.

This commit creates the option to maintain high resolution in all ocean cells (same resolution as what is prescribed at outlet glacier termini). This is necessary to resolve the fjord geometry that thermal forcing parameterizations rely on. To reduce time when generating these higher resolution meshes, this commit also introduces an option to do an initial cull of the bedMachine dataset using the extent of the contintental shelf before generating the GIS mesh.

This commit moves the calculation of geom_points and geom_edges using a geojson of the gis contintental shelf extent to compass/landice/tests/greenland/mesh.py, instead of compass/landice/mesh.py. If this options is not being used, thn geom_points and geom_edges are calculated within build_cell_width, as per usual

Co-authored-by: Trevor Hillebrand <trhille@lanl.gov>

Re-introduces code that sets a large cell width for cells that will be culled by distance from ice edge. Only applied when not using a geojson to pre-cull domain

Ensures the bedTopography constraint on mesh resolution is enforced regardless 'max_res_in_ocn'. Instead, grid resolution in the ocean is maximized following the bedTopgraphy constraint, overwritting it where necessary.

alexolinhager requested a review from matthewhoffman February 21, 2025 17:44

alexolinhager assigned trhille and andrewdnolan Feb 21, 2025

trhille requested changes Aug 4, 2025

View reviewed changes

alexolinhager requested a review from trhille October 27, 2025 20:26

trhille requested changes Oct 28, 2025

View reviewed changes

andrewdnolan and others added 7 commits October 28, 2025 14:46

Enable coarse mesh resolution over the open ocean by using geojson file.

ee007ea

Enable high resolution in ocean

f804202

Creates the option to maintain resolution in the ocean. Used to resolve fjords and bathymetry when coupling to MPAS-Ocean.

Update compass/landice/tests/greenland/mesh.py

df05827

Co-authored-by: Trevor Hillebrand <trhille@lanl.gov>

Keep large culled cell width if not using geojson

fef3755

Re-introduces code that sets a large cell width for cells that will be culled by distance from ice edge. Only applied when not using a geojson to pre-cull domain

Remove bedTopography res. constraint from if/else

688b86e

Ensures the bedTopography constraint on mesh resolution is enforced regardless 'max_res_in_ocn'. Instead, grid resolution in the ocean is maximized following the bedTopgraphy constraint, overwritting it where necessary.

alexolinhager force-pushed the alexolinhager/hiRes_ocean_GIS_mesh branch from 5f0591d to 688b86e Compare October 28, 2025 21:47

		ds.to_netcdf(self.mesh_filename, 'a')


		def set_geojson_geom_points_and_edges(self):

Enable Option for High Res Mesh in Ocean #894

Are you sure you want to change the base?

Enable Option for High Res Mesh in Ocean #894

Uh oh!

Conversation

alexolinhager commented Feb 21, 2025

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

trhille commented Oct 27, 2025

Uh oh!

trhille left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants