eoio.processors.interpolate.processor module

eoio.processors.interpolate.processor

Top-level interpolation processor.

This processor provides a single, stable user-facing interface (interpolate).

User config example

processors = {

“interpolate”: {

“coords”: [“x_5000m”, “y_5000m”], “target_grid”: [“x_60m”, “y_60m”], “method”: “linear”, },

}

class eoio.processors.interpolate.processor.Interpolate(params: Dict[str, Any] | None = None, context: Dict[str, Any] | None = None)

Bases: BaseProcessor

Interpolate measurement variables in a dataset to a new coordinate grid.

Processor parameters

The following parameters can be provided in the params dict:

param coords:

List of coordinate names to interpolate along (e.g. ["x", "y"]).

param target_grid:

List of target coordinate values to interpolate to. If str, must be the name of an existing coordinate in the dataset (e.g. "x_60m"). If xr.DataArray, new coord name will be set to name of DataArray. If other (np.Array, list), coord name will be unchanged (e.g. "x_5000m" or "x_5000m_interp" depending on value of ‘inplace’).

param data_vars:

List of variable names to interpolate. If omitted, all variables with any of the interpolation coordinates as a dimension will be interpolated.

param method:

Interpolation method (e.g. "linear", "nearest", etc.). Must be supported by the underlying interpolation implementation (e.g. xarray.interp()).

param inplace:

Bool, if False, new interpolated variables are added to the dataset with the suffix ‘_interp’ (e.g. ‘B02_interp’). If True, original variables are replaced by interpolated ones. Default is False. Either way the source-grid coordinates (those named in coords) are left in place, so the same grid can feed several interpolate steps in one pipeline.

param on_missing:

Behaviour if required metadata for conversion is missing. Supported values are "error" (default, if omitted) or "skip".

param drop_original:

Bool, if True, drop the source variable(s) (the ones named in data_vars, or every variable that had the interpolation coords as a dimension if data_vars was omitted) after producing the interpolated copy. Only meaningful when inplace is False – when inplace is True there’s no separate original left to drop, so this is a no-op in that case. Default is False, matching StackCubes’s drop_originals default.

Be careful when the same source variable is interpolated to multiple target grids across several interpolate steps in one pipeline (e.g. a shared solar angle interpolated onto three different band resolutions): only set drop_original=True on the last such step, since dropping the source after the first would leave nothing for the remaining steps to interpolate from.

Notes

• This processor is intended to run after reading.

name = 'interpolate'

run(ds: Dataset) → Dataset

Run interpolation on the dataset.

Parameters:

ds – Input dataset.

Returns:

Output dataset with interpolated variables.

class eoio.processors.interpolate.processor.InterpolateConfig(coords: Sequence[str], target_grid: Sequence, data_vars: Sequence[str] | None = None, method: str = 'linear', inplace: bool | None = False, on_missing: str = 'error', drop_original: bool = False)

Bases: object

Parameters for the interpolate processor.

coords: Sequence[str]

data_vars: Sequence[str] | None = None

drop_original: bool = False

inplace: bool | None = False

method: str = 'linear'

on_missing: str = 'error'

target_grid: Sequence