regrid

New in version 0.5.0.

regrid(values, in_grid=None, out_grid=None, interpolation='linear', output='values_gridspec', backend='precomputed-local', inventory_path=None)

Regrid the values using precomputed weights stored at a local path.

Parameters:

• values (ndarray, fieldlist, field) –

  the following input data types are supported:

  • an ndarray representing a single field defined on the in_grid. A valid in_grid must be specified.

  • an earthkit-data GRIB fieldlist (requires earthkit-data >= 0.6.0). The input grid is automatically detected from the data (in_grid is ignored). It only works when the out_grid is a regular latitude-longitude grid.

  • an earthkit-data GRIB field (requires earthkit-data >= 0.6.0). The input grid is automatically detected from the data (in_grid is ignored). It only works when the out_grid is a regular latitude-longitude grid.

• in_grid (dict) – the gridspec describing the grid that values are defined on. Ignored when values is not an ndarray.

• out_grid (dict) – the gridspec describing the target grid that values will be interpolated onto

• interpolation (str) – the interpolation method. Possible values are linear and nearest-neighbour. For nearest-neighbour the following aliases are also supported: nn, nearest-neighbor.

• output (str) –

  define what is returned when the input is an array, ignored otherwise. Possible values are as follows:

  • ”values_gridpec”: return a tuple with the interpolated values and the gridspec of the output grid. This is the default option.

  • ”values”: return the interpolated values only

  • ”gridpec”: return the gridspec of the output grid only

Returns:

see the output parameter for details

The interpolation only works when a pre-computed interpolation weights are available for the given in_grid, out_grid and interpolation combination in the local inventory specified by inventory_path.

If the weights are available the interpolation is performed by multiplying the values vector with it (matrix-vector multiplication).