Click here for the API documentation
This project aims to provide a bridge between Iris and ESMF. This takes the form of regridder classes which take Iris cubes as their arguments and use ESMF to perform regridding calculations. These classes are designed to perform well on cubes which have multiple non-horizontal dimensions and lazy (Dask) data. Both rectilinear and curvilinear grids as well as UGRID meshes have been supported.
There are a range of regridder classes (e.g MeshToGridESMFRegridder and
GridToMeshESMFRegridder). For an example of the regridding process, the
MeshToGridESMFRegridder class works as follows:
import iris
from iris.experimental.ugrid import PARSE_UGRID_ON_LOAD
from esmf_regrid.experimental.unstructured_scheme import MeshToGridESMFRegridder
# An example such a file can be found at:
# https://github.com/SciTools/iris-test-data/blob/master/test_data/NetCDF/unstructured_grid/data_C4.nc
with PARSE_UGRID_ON_LOAD.context():
source_mesh_cube = iris.load_cube("mesh_cube.nc")
# An example of such a file can be found at:
# https://github.com/SciTools/iris-test-data/blob/master/test_data/NetCDF/global/xyt/SMALL_hires_wind_u_for_ipcc4.nc
target_grid_cube = iris.load_cube("grid_cube.nc")
# Initialise the regridder with a source mesh and target grid.
regridder = MeshToGridESMFRegridder(source_mesh_cube, target_grid_cube)
# use the initialised regridder to regrid the data from the source cube
# onto a cube with the same grid as `target_grid_cube`.
result = regridder(source_mesh_cube)Note that this pattern allows the reuse of an initialised regridder, saving significant amounts of time when regridding. To make use of this efficiency across sessions, we support the saving of certain regridders. We can do this as follows:
from esmf_regrid.experimental.io import load_regridder, save_regridder
# Save the regridder.
save_regridder(regridder, "saved_regridder.nc")
# Load saved regridder.
loaded_regridder = load_regridder("saved_regridder.nc")
# Use loaded regridder.
result = loaded_regridder(source_mesh_cube)