1

Is there a straightforward way to extract a region from an Iris cube which is described by 2D latitude and longitude variables, for example using NEMO ocean model data?

I found this workaround but was wondering if there was a way to do this in 'pure' Iris, without having to resort to defining a new function?

For example, if I have this cube...

In [30]: print(cube)
mole_concentration_of_dimethyl_sulfide_in_sea_water / (mol m-3) (time: 780; cell index along second dimension: 330; cell index along first dimension: 360)
     Dimension coordinates:
          time                                                       x                                       -                                      -
          cell index along second dimension                          -                                       x                                      -
          cell index along first dimension                           -                                       -                                      x
     Auxiliary coordinates:
          latitude                                                   -                                       x                                      x
          longitude                                                  -                                       x                                      x

... and then try to extract a region using intersection, I get this...

>>> subset = cube.intersection(longitude=(-10, 10))

CoordinateMultiDimError: Multi-dimensional coordinate not supported: 'longitude'

Thanks!

jonnyhtw
  • 61
  • 8

1 Answers1

2

As you can see from the error messsage, iris does not currently support subsetting by multi-dimensional coordinates, so you have to write a function similar to bbox_extract_2Dcoords() in that blog post. All it does is creates a boolean mask with values set to True within your region of interest and False outside. Then the boundaries of this region are used as indices to subset the cube.

An alternative would be to regrid the data to a regular grid defined by 1D longitude and latitude and then subset the data using the standard Constraint() method.

Denis Sergeev
  • 1,052
  • 1
  • 8
  • 19
  • Thanks a lot for that! Could this regridding to 1D longitude and latitude within Iris? – jonnyhtw Feb 19 '21 at 01:20
  • @jonnyhtw yes, it should be possible with using https://scitools-docs.github.io/iris/master/generated/api/iris/experimental/regrid.html#iris.experimental.regrid.regrid_weighted_curvilinear_to_rectilinear – Denis Sergeev Feb 20 '21 at 18:38