cf.Field.regrids

Field.regrids(dst, method=None, src_cyclic=None, dst_cyclic=None, use_src_mask=True, use_dst_mask=False, fracfield=False, src_axes=None, dst_axes=None, axis_order=None, ignore_degenerate=True, return_operator=False, check_coordinates=False, min_weight=None, weights_file=None, inplace=False, i=False, _compute_field_mass=None)[source]

Regrid the field to a new latitude and longitude grid.

Regridding is the process of interpolating the field data values while preserving the qualities of the original data, and the metadata of the unaffected axes. The metadata for the regridded axes are taken from the dst parameter.

The 2-d regridding takes place on a sphere, with the grid being defined by latitude and longitude spherical polar coordinates.

Latitude and longitude coordinates

The source and destination grids of the regridding must both be defined by latitude and longitude coordinates, which may be 1-d dimension coordinates or 2-d auxiliary coordinates. These are automatically detected from the field being regridded and the specification of the destination grid given by the dst parameter.

When a grid is defined by 2-d latitude and longitude coordinates, it is necessary for their X and Y dimensions to be defined. This is either automatically inferred from the existence of 1-d dimension coordinates, or else must be specified with src_axes or dst_axes parameters.

Curvilinear Grids

Grids in projection coordinate systems can be regridded as long as two dimensional latitude and longitude coordinates are present.

Tripolar Grids

Connections across the bipole fold are not currently supported, but are not necessary in some cases, for example if the points on either side are together without a gap (as is the case for NEMO model outputs).

Cyclicity of the X axis

The cyclicity of the X (longitude) axes of the source and destination grids (i.e. whether or not the first and last cells of the axis are adjacent) are taken into account. By default, the cyclicity is inferred from the grids’ defining coordinates, but may be also be provided with the src_cyclic and dst_cyclic parameters.

Masked cells

By default, the data mask of the source data is taken into account during the regridding process, but the destination grid mask is not. This behaviour may be changed with the use_src_mask and use_dst_mask parameters.

In general the source data may be arbitrarily masked, meaning that the mask for the regridding axes may vary along the non-regridding axes. The exceptions to this are for second-order conservative, patch recovery regridding, and nearest source to destination methods, for which the mask of the regridding axes must be the same across all non-regridding axes. In these special cases an exception will be raised if the source data mask does not meet this requirement.

Implementation

The interpolation is carried out using regridding weights calculated by the esmpy package, a Python interface to the Earth System Modeling Framework regridding utility (ESMF, https://earthsystemmodeling.org/regrid). Outside of esmpy, these weights are then modified for masked cells (if required) and the regridded data are created as the dot product of the weights with the source data. (Note that whilst the esmpy package is also able to create the regridded data from its weights, this feature can’t be integrated with the dask framework that underpins the field’s data.)

Logging

Whether esmpy logging is enabled or not is determined by cf.regrid_logging. If it is enabled then logging takes place after every call. By default logging is disabled.

New in version 1.0.4.

See also

regridc

Parameters
dst: Field, Domain, RegridOperator or sequence of Coordinate

The definition of the destination grid on which to regrid the field’s data. One of:

  • Field: The grid is defined by the latitude and longitude coordinates of the field construct’s domain.

  • Domain: The grid is defined by the latitude and longitude coordinates of the domain construct.

  • Sequence of Coordinate: The grid is defined by two 1-d dimension coordinate constructs, or two 2-d auxiliary coordinate constructs, that define the spherical latitude and longitude coordinates (in any order) of the destination grid.

    In the 2-d case, both coordinate constructs must have their axes in the same order, which must be specified by the dst_axes parameter.

  • RegridOperator: The grid is defined by a regrid operator that has been returned by a previous call with the return_operator parameter set to True.

    Unlike the other options, for which the regrid weights need to be calculated, the regrid operator already contains the weights. Therefore, for cases where multiple fields with the same source grids need to be regridded to the same destination grid, using a regrid operator can give performance improvements by avoiding having to calculate the weights for each source field. Note that for the other types of dst parameter, the calculation of the regrid weights is not a lazy operation.

    Note

    The source grid of the regrid operator is immediately checked for compatibility with the grid of the source field. By default only the computationally cheap tests are performed (checking that the coordinate system, cyclicity and grid shape are the same), with the grid coordinates not being checked. The coordinates check will be carried out, however, if the check_coordinates parameter is True.

method: str or None, optional

Specify the regridding interpolation method. This parameter must be set unless dst is a RegridOperator, when the method is ignored.

The method parameter may be one of the following:

  • 'linear': Bilinear interpolation.

  • 'bilinear': Deprecated alias for 'linear'.

  • 'conservative_1st': First order conservative interpolation. Preserves the area integral of the data across the interpolation from source to destination. It uses the proportion of the area of the overlapping source and destination cells to determine appropriate weights.

  • 'conservative': Alias for 'conservative_1st'

  • 'conservative_2nd': Second-order conservative interpolation. As with first order conservative interpolation, preserves the area integral of the field between source and destination using a weighted sum, with weights based on the proportionate area of intersection. In addition the second-order conservative method takes the source gradient into account, so it yields a smoother destination field that typically better matches the source data.

  • 'patch' Patch recovery interpolation. A second degree 2-d polynomial regridding method, which uses a least squares algorithm to calculate the polynomials. This method typically results in better approximations to values and derivatives when compared to bilinear interpolation.

  • 'nearest_stod': Nearest neighbour interpolation for which each destination point is mapped to the closest source point. Useful for extrapolation of categorical data. Some destination cells may be unmapped.

  • 'nearest_dtos': Nearest neighbour interpolation for which each source point is mapped to the destination point. Useful for extrapolation of categorical data. All destination cells will be mapped.

  • None: This is the default and can only be used when dst is a RegridOperator.

src_cyclic: None or bool, optional

Specifies whether or not the source grid longitude axis is cyclic (i.e. the first and last cells of the axis are adjacent). If None (the default) then the cyclicity will be inferred from the source grid coordinates, defaulting to False if it can not be determined.

dst_cyclic: None or bool, optional

Specifies whether or not the destination grid longitude axis is cyclic (i.e. the first and last cells of the axis are adjacent). If None (the default) then the cyclicity will be inferred from the destination grid coordinates, defaulting to False if it can not be determined.

Ignored if dst is a RegridOperator.

use_src_mask: bool, optional

By default the mask of the source field is taken into account during the regridding process. The only possible exception to this is when the nearest source to destination regridding method ('nearest_stod') is being used. In this case, if use_src_mask is False then each destination point is mapped to the closest source point, whether or not it is masked (see the method parameter for details).

Ignored if dst is a RegridOperator.

use_dst_mask: bool, optional

If dst is a Field and use_dst_mask is False (the default) then the mask of data on the destination grid is not taken into account when performing regridding. If use_dst_mask is True then any masked cells in the dst field construct are transferred to the result. If dst has more dimensions than are being regridded, then the mask of the destination grid is taken as the subspace defined by index 0 of all the non-regridding dimensions.

Ignored if dst is not a Field.

src_axes: dict, optional

When the source grid’s X and Y dimensions can not be inferred from the existence of 1-d dimension coordinates, then they must be identified with the src_axes dictionary, with keys 'X' and 'Y'.

The dictionary values identify a unique domain axis by passing the given axis description to a call of the field construct’s domain_axis method. For example, for a value of 'ncdim%x', the domain axis construct returned by f.domain_axis('ncdim%x') is selected.

Parameter example:

{'X': 'ncdim%x', 'Y': 'ncdim%y'}

Parameter example:

{'X': 1, 'Y': 0}

dst_axes: dict, optional

When the destination grid’s X and Y dimensions can not be inferred from the existence of 1-d dimension coordinates, then they must be identified with the dst_axes dictionary, with keys 'X' and 'Y'.

If dst is a Field or Domain, then the dictionary values identify a unique domain axis by passing the given axis description to a call of the destination field or domain construct’s domain_axis method. For example, for a value of 'ncdim%x', the domain axis construct returned by f.domain_axis('ncdim%x') is selected.

If dst is a sequence of Coordinate, then the dictionary values identify a unique domain axis by its position in the 2-d coordinates’ data arrays, i.e. the dictionary values must be 0 and 1:

Ignored if dst is a RegridOperator.

Parameter example:

{'X': 'ncdim%x', 'Y': 'ncdim%y'}

Parameter example:

{'X': 1, 'Y': 0}

ignore_degenerate: bool, optional

For conservative regridding methods, if True (the default) then degenerate cells (those for which enough vertices collapse to leave a cell as either a line or a point) are skipped, not producing a result. Otherwise an error will be produced if degenerate cells are found, that will be present in the esmpy log files.

For all other regridding methods, degenerate cells are always skipped, regardless of the value of ignore_degenerate.

Ignored if dst is a RegridOperator.

return_operator: bool, optional

If True then do not perform the regridding, rather return the RegridOperator instance that defines the regridding operation, and which can be used in subsequent calls. See the dst parameter for details.

New in version 3.10.0.

check_coordinates: bool, optional

If True and dst is a `RegridOperator`then the source grid coordinates defined by the operator are checked for compatibility against those of the source field. By default this check is not carried out. See the dst parameter for details.

Ignored unless dst is a RegridOperator.

New in version 3.14.0.

min_weight: float, optional

A very small non-negative number. By default min_weight is 2.5 * np.finfo("float64").eps, i.e. 5.551115123125783e-16. It is used during linear and first-order conservative regridding when adjusting the weights matrix to account for the data mask. It is ignored for all other regrid methods, or if data being regridded has no missing values.

In some cases (described below) for which weights might only be non-zero as a result of rounding errors, the min_weight parameter controls whether or a not cell in the regridded field is masked.

The default value has been chosen empirically as the smallest value that produces the same masks as esmpy for the use cases defined in the cf test suite.

Define w_ji as the multiplicative weight that defines how much of Vs_i (the value in source grid cell i) contributes to Vd_j (the value in destination grid cell j).

Linear regridding

Destination grid cell j will only be masked if a) it is masked in the destination grid definition; or b) w_ji >= min_weight for those masked source grid cells i for which w_ji > 0.

Conservative first-order regridding

Destination grid cell j will only be masked if a) it is masked in the destination grid definition; or b) the sum of w_ji for all non-masked source grid cells i is strictly less than min_weight.

New in version 3.14.0.

weights_file: str or None, optional

Provide a netCDF file that contains, or will contain, the regridding weights. If None (the default) then the weights are computed in memory for regridding between the source and destination grids, and no file is created.

If set to a file path that does not exist then the weights will be computed and also written to that file.

If set to a file path that already exists then the weights will be read from this file, instead of being computed.

Note

No checks are performed on an existing file to ensure that the weights are appropriate for the source field and the values of the keyword parameters. Inappropriate weights will produce incorrect results.

However, when regridding using weights from a file, ensuring that the source field has the same shape over the regridding axes, and the parameter settings are the same as those used when the weights file was created, will ensure correct results.

A netCDF regridding weights file created directly by ESMF has the same structure and variable names (S, row, and col for the weights, destination/row indices, and source/col indices respectively), so may be provided as a weights_file, noting that no checks will be applied to it.

Performance

The computation of the weights can be much more costly than the regridding itself, in which case reading pre-calculated weights can improve performance.

New in version 3.15.2.

inplace: bool, optional

If True then do the operation in-place and return None.

axis_order: sequence, optional

Deprecated at version 3.14.0.

fracfield: bool, optional

Deprecated at version 3.14.0.

_compute_field_mass: dict, optional

Deprecated at version 3.14.0.

i: deprecated at version 3.0.0

Use the inplace parameter instead.

Returns
Field or None or RegridOperator

The regridded field construct; or None if the operation was in-place; or the regridding operator if return_operator is True.

Examples

>>> src, dst = cf.example_fields(1, 0)
>>> print(src)
Field: air_temperature (ncvar%ta)
---------------------------------
Data            : air_temperature(atmosphere_hybrid_height_coordinate(1), grid_latitude(10), grid_longitude(9)) K
Cell methods    : grid_latitude(10): grid_longitude(9): mean where land (interval: 0.1 degrees) time(1): maximum
Field ancils    : air_temperature standard_error(grid_latitude(10), grid_longitude(9)) = [[0.76, ..., 0.32]] K
Dimension coords: atmosphere_hybrid_height_coordinate(1) = [1.5]
                : grid_latitude(10) = [2.2, ..., -1.76] degrees
                : grid_longitude(9) = [-4.7, ..., -1.18] degrees
                : time(1) = [2019-01-01 00:00:00]
Auxiliary coords: latitude(grid_latitude(10), grid_longitude(9)) = [[53.941, ..., 50.225]] degrees_N
                : longitude(grid_longitude(9), grid_latitude(10)) = [[2.004, ..., 8.156]] degrees_E
                : long_name=Grid latitude name(grid_latitude(10)) = [--, ..., kappa]
Cell measures   : measure:area(grid_longitude(9), grid_latitude(10)) = [[2391.9657, ..., 2392.6009]] km2
Coord references: grid_mapping_name:rotated_latitude_longitude
                : standard_name:atmosphere_hybrid_height_coordinate
Domain ancils   : ncvar%a(atmosphere_hybrid_height_coordinate(1)) = [10.0] m
                : ncvar%b(atmosphere_hybrid_height_coordinate(1)) = [20.0]
                : surface_altitude(grid_latitude(10), grid_longitude(9)) = [[0.0, ..., 270.0]] m
>>> print(dst)
Field: specific_humidity (ncvar%q)
----------------------------------
Data            : specific_humidity(latitude(5), longitude(8)) 1
Cell methods    : area: mean
Dimension coords: latitude(5) = [-75.0, ..., 75.0] degrees_north
                : longitude(8) = [22.5, ..., 337.5] degrees_east
                : time(1) = [2019-01-01 00:00:00]
>>> x = src.regrids(dst, method='linear')
>>> print(x)
Field: air_temperature (ncvar%ta)
---------------------------------
Data            : air_temperature(atmosphere_hybrid_height_coordinate(1), latitude(5), longitude(8)) K
Cell methods    : latitude(5): longitude(8): mean where land (interval: 0.1 degrees) time(1): maximum
Dimension coords: atmosphere_hybrid_height_coordinate(1) = [1.5]
                : latitude(5) = [-75.0, ..., 75.0] degrees_north
                : longitude(8) = [22.5, ..., 337.5] degrees_east
                : time(1) = [2019-01-01 00:00:00]
Coord references: standard_name:atmosphere_hybrid_height_coordinate
Domain ancils   : ncvar%a(atmosphere_hybrid_height_coordinate(1)) = [10.0] m
                : ncvar%b(atmosphere_hybrid_height_coordinate(1)) = [20.0]
                : surface_altitude(latitude(5), longitude(8)) = [[--, ..., --]] m

>>> r = src.regrids(dst, method='linear', return_operator=True)
>>> y = src.regrids(r)
>>> y.equals(x)
True