Topography Tasks

The polaris.tasks.e3sm.init.topo module provides tools for working with topography data in Polaris. This includes steps for processing, modifying, and combining topography datasets to create inputs for E3SM components such as MPAS-Ocean. The framework is designed to handle both global and regional datasets, supporting various grid types like lat-lon and cubed-sphere grids.

Combine Steps and Tasks

Global datasets of base elevation (land surface elevation and ocean bathymetry) do not typically include the latest datasets around Antarctica needed for ice-sheet and ice-shelf modeling. For this reason, we typically combine a global topography dataset north of the Southern Ocean with one for Antarctica.

The polaris.tasks.e3sm.init.topo.combine.CombineStep step is a key component of the topography framework. It is responsible for combining global and Antarctic topography datasets into a single dataset suitable for use in E3SM simulations. The step supports blending datasets across specified latitude ranges and remapping them to a target grid.

The polaris.tasks.e3sm.init.topo.combine.CombineTask wraps the CombineStep into a task that can be used to generate and cache combined topography datasets for reuse in other contexts.

The polaris.tasks.e3sm.init.topo.combine.VizCombinedStep step is an optional visualization step that can be added to the workflow to create plots of the combined topography dataset. This step is particularly useful for debugging or analyzing the combined dataset.

High-Resolution and Low-Resolution Versions

There are two versions of the combine steps and task:

1. Standard (High-Resolution) Version: This version maps to a high-resolution (ne3000, ~1 km) cubed-sphere grid by default, producing topogrpahy that is suitable for remapping to standard and high-resolution MPAS meshes (~60 km and finer).

2. Low-Resolution Version: This version uses a coarser ne120 (~25 km) grid for faster remapping to coarse-resolution MPAS meshes (e.g., Icos240). It is designed to reduce computational cost while still providing adequate accuracy for low-resolution simulations used for regression testing rather than science.

The low-resolution version can be selected by setting the low_res parameter to True when creating the CombineStep or CombineTask.

Key Features

• Dataset Support: Supports multiple datasets, including bedmap3, bedmachinev3, and gebco2023.

• Grid Types: Handles both lat-lon and cubed-sphere target grids.

• Blending: Blends global and Antarctic datasets across a configurable latitude range.

• Remapping: Uses tools like mbtempest, ESMF_RegridWeightGen and ncremap for remapping datasets to the target grid.

• Output: Produces combined topography datasets with consistent variables and attributes.

• Visualization: Generates rasterized images of various fields (e.g., base elevation, ice draft) using the datashader library.

Configuration Options

The CombineStep step is configured through the [combine_topo] section in the configuration file. Key options include:

• resolution_latlon: Target resolution for lat-lon grids (in degrees).

• resolution_cubedsphere: Target resolution for cubed-sphere grids (e.g., 3000 for NExxx grids).

• latmin and latmax: Latitude range for blending datasets.

• ntasks and min_tasks: Number of MPI tasks for remapping.

• method: Remapping method (e.g., bilinear).

• lat_tiles and lon_tiles: Number of tiles to split the global dataset for parallel remapping.

• renorm_thresh: Threshold for renormalizing Antarctic variables during blending.

For the low-resolution version, additional configuration options are provided in the combine_low_res.cfg file.

Workflow

1. Setup: The step downloads required datasets and sets up input/output files.

2. Modification: Antarctic and global datasets are modified to include necessary variables and attributes.

3. Remapping: Datasets are remapped to the target grid using SCRIP files and weight generation.

4. Blending: The datasets are blended across the specified latitude range.

5. Output: The combined dataset is saved in NetCDF format.

6. Optional Field Plotting: Each field in the dataset is rasterized and saved as an image with a colorbar.

Example Usage

Below is an example of how the CombineStep can be added to a Polaris task:

from polaris.tasks.e3sm.init.topo.combine import CombineStep

component = task.component
subdir = CombineStep.get_subdir(low_res=False)
if subdir in component.steps:
    step = component.steps[subdir]
else:
    step = CombineStep(component=component, low_res=False)
    component.add_step(step)
task.add_step(step)

To create a CombineTask for caching combined datasets:

from polaris.tasks.e3sm.init.topo.combine import CombineTask

combine_task = CombineTask(component=my_component, low_res=False)
my_component.add_task(combine_task)

Below is an example of how the VizCombinedStep can be added to a Polaris task:

from polaris.tasks.e3sm.init.topo.combine import VizCombinedStep

viz_step = VizCombinedStep(component=my_component, combine_step=combine_step)
my_component.add_step(viz_step)

Since there is a single shared step for each pair of Antarctic and global datasets, the step should be added only once to the component and the existing step (identifiable via its subdir) should be used subsequently.

The VizCombinedStep is typically added only when visualization is explicitly required, as it is not part of the default workflow.

For more details, refer to the source code of the polaris.tasks.e3sm.init.topo.combine.CombineStep and polaris.tasks.e3sm.init.topo.combine.CombineTask classes.

Note

Since this step is expensive and time-consuming to run, most tasks will want to use cached outputs from this step rather than running it in full.

Remapping Topography to MPAS Base Meshes

The e3sm/init component includes a workflow for remapping combined topography datasets to MPAS base meshes at a range of resolutions. This process ensures that bathymetry, land-ice thickness, and related fields are accurately mapped from the cubed-sphere grid onto the unstructured MPAS mesh used in E3SM simulations.

Remapping Workflow Overview

The remapping workflow is composed of several modular steps, each responsible for a specific part of the process:

• Masking: The polaris.tasks.e3sm.init.topo.remap.MaskTopoStep applies ocean and land masks to the combined topography dataset on the cubed-sphere grid. This step generates masked versions of all topography fields, as well as fractional land and ocean coverage.

• Remapping: The polaris.tasks.e3sm.init.topo.remap.RemapTopoStep remaps the masked topography fields from the cubed-sphere grid to the MPAS mesh. This step can be run with or without smoothing, depending on configuration.

• Smoothing: Smoothing is optionally applied during remapping. The workflow supports both unsmoothed and smoothed topography, with the smoothed step depending on the unsmoothed results.

• Visualization: The polaris.tasks.e3sm.init.topo.remap.VizRemappedTopoStep can be added to generate plots of the remapped fields for quality control.

The polaris.tasks.e3sm.init.topo.remap.RemapTopoTask orchestrates these steps for each supported MPAS base mesh.

Step Dependencies

The typical dependency chain is:

1. CombineStep (combines global and Antarctic topography)

2. MaskTopoStep (applies land/ocean masks)

3. RemapTopoStep (remaps to MPAS mesh, optionally with smoothing)

4. VizRemappedTopoStep (optional visualization)

Configuration Options

The remapping steps are configured through the [remap_topography] section in the configuration file. Key options include:

• ntasks and min_tasks: Number of MPI tasks for remapping.

• renorm_threshold: Fractional threshold for renormalizing elevation variables.

• expand_distance and expand_factor: Smoothing parameters (set to 0 and 1 for no smoothing).

• Additional options for visualization colormaps and normalization.

For the low-resolution version, additional configuration options are provided in the remap_low_res.cfg file.

Workflow

1. Masking: The MaskTopoStep applies land and ocean masks to the combined topography dataset, producing masked fields and fractional coverage variables.

2. Remapping: The RemapTopoStep remaps the masked topography fields to the MPAS mesh. If smoothing is enabled, both unsmoothed and smoothed steps are created, with the smoothed step depending on the unsmoothed output.

3. Smoothing: Smoothing parameters are controlled via configuration. If no smoothing is requested, the smoothed step simply symlinks the unsmoothed results.

4. Visualization: The VizRemappedTopoStep can be added to plot each remapped field using configuration-driven colormaps and normalization.

5. Output: The final remapped topography is saved as topography_remapped.nc for each mesh and smoothing option.

Example Usage

Below is an example of how the remapping steps can be added to a Polaris task:

from polaris.tasks.e3sm.init.topo.remap import get_default_remap_topo_steps

steps, config = get_default_remap_topo_steps(
    component=component,
    base_mesh_step=base_mesh_step,
    combine_topo_step=combine_topo_step,
    low_res=low_res,
    smoothing=True,
    include_viz=True,
)
for step in steps:
    component.add_step(step)

To add the full remapping workflow as a task for each supported mesh:

from polaris.tasks.e3sm.init.topo.remap import add_remap_topo_tasks

add_remap_topo_tasks(component)

This will add a RemapTopoTask for each supported base mesh, including all necessary steps (masking, remapping, smoothing, and optional visualization).

Example: Ocean Fraction on Icos480km Mesh

Below is an example of the ocean fraction field (ocean_frac) remapped to the Icos480km MPAS mesh:

This field indicates the fraction of each MPAS cell that is covered by ocean after remapping, which is important for subsequent steps such as mesh culling and mask generation.

Customization

Remapping options, including smoothing parameters and the number of MPI tasks, are controlled via the [remap_topography] section in the configuration file. The workflow supports both unsmoothed and smoothed topography, and can generate visualizations for each remapped field.

For more details, see the source code for polaris.tasks.e3sm.init.topo.remap.RemapTopoTask and related steps, as well as the configuration files remap.cfg and remap_low_res.cfg.

Extending Masking and Smoothing Functionality

Developers may wish to customize how land/ocean masks are generated or how smoothing is applied during remapping. This can be accomplished by subclassing the relevant step classes and overriding their methods:

Customizing Mask Generation

To implement a custom approach for generating land and ocean masks, create a subclass of polaris.tasks.e3sm.init.topo.remap.MaskTopoStep and override the define_masks() method. This method receives an xarray.Dataset and should return two xarray.DataArray objects representing the ocean and land masks, respectively.

Example:

from polaris.tasks.e3sm.init.topo.remap import MaskTopoStep

class MyCustomMaskStep(MaskTopoStep):
    def define_masks(self, ds):
        # Custom logic for mask generation
        ocean_mask = ...  # compute ocean mask as DataArray
        land_mask = ...   # compute land mask as DataArray
        return ocean_mask, land_mask

You can then use your custom step in place of the default MaskTopoStep when constructing your workflow.

Customizing Smoothing Behavior

To implement custom smoothing logic, subclass polaris.tasks.e3sm.init.topo.remap.RemapTopoStep and override the define_smoothing() method. This method receives the unsmoothed topography dataset and should return the expand_distance and expand_factor (either as scalars or xarray.DataArray objects) to control the smoothing applied during remapping.

Example:

from polaris.tasks.e3sm.init.topo.remap import RemapTopoStep

class MyCustomRemapStep(RemapTopoStep):
    def define_smoothing(self, ds_unsmoothed):
        # Custom logic for spatially varying smoothing
        expand_distance = ...  # scalar or DataArray
        expand_factor = ...    # scalar or DataArray
        return expand_distance, expand_factor

This allows for advanced smoothing strategies, such as spatially varying parameters based on mesh properties or scientific requirements.

Integration

To use your custom steps, simply instantiate them in your workflow or override the step creation logic in your task or workflow setup.

For more details, refer to the docstrings and source code of polaris.tasks.e3sm.init.topo.remap.MaskTopoStep and polaris.tasks.e3sm.init.topo.remap.RemapTopoStep.