Changing values in iris cube based on coordinates instead of index

[chuaxr]

I would like to be able to change values in iris based on the coordinate, instead of the index.

For example, consider the following cube and say that I wish to set values from -45N to 45N and 160E to 240E to 1:

import iris
import numpy as np
from iris.coords import DimCoord
from iris.cube import Cube

latitude_vals = np.linspace(-90, 90, 4)
longitude_vals = np.linspace(45, 360, 8)
latitude = DimCoord(latitude_vals, standard_name="latitude", units="degrees")
longitude = DimCoord(longitude_vals, standard_name="longitude", units="degrees")
cube = Cube(
    np.zeros((4, 8), np.float32), dim_coords_and_dims=[(latitude, 0), (longitude, 1)]
)

In this example, what I want can be done by invoking xarray:

import xarray as xr
da = xr.DataArray.from_iris(cube)
da.loc[dict(latitude=slice(-45, 45), longitude=slice(160, 240))] = 1

But can this be done entirely within iris, without having to resort to specifying the indices manually?

Example of specifying the indices manually:

cube.data[1:3, 3:5] = cube.data[1:3, 3:5] + 1

Cross-posted on Stack Overflow.

[trexfeathers]

Hi @chuaxr , we definitely agree with you! There has been much discussion on this topic - see IEP 1 (and the discussion: #1988). There are plans in the near future to focus more attention on IEP's, so hopefully we will see some movement soon.

[rcomer]

I think #3172 is also relevant here: however we currently slice our cube, we get a copy. So assigning values to the slice’s data will not affect the original cube. We have to slice the numpy array (i.e. cube.data) instead, to get get a view.