MIR: regridding numpy arrays

This example shows how to interpolate numpy arrays using the mir backend.

from earthkit.regrid.array import regrid
import numpy as np

First, we generate random data matching the size of an O32 octahedral reduced Gaussian grid.

values = np.random.random(5248)

Next, interpolate the array onto a 1x1 degree global regular latitude-longitude grid using regrid(). The input and output grids are defined by a gridspec. Both the input and the output values are numpy arrays.

res, res_grid = regrid(values, in_grid={"grid": "O32"}, out_grid={"grid": [1,1]})
res.shape, res_grid

((181, 360),
 {'east': 359, 'grid': [1, 1], 'north': 90, 'south': -90, 'west': 0})

regrid() returns a tuple with the regridded array and the resulting gridspec. This latter might be different than the one specified in out_grid because MIR can perform normalisation and other adjustments on it.