How to calculate the horizontal gradient in longitude/latitude coordinates?

[NMC-DAVE]

📰 Custom Issue

Hi,
I want to calculate the horizontal gradient in longitude/latitude coordinates. So can you show me
an example to do this? Thanks very much!
Dave

[trexfeathers]

Hi @NMC-DAVE, thanks for getting in touch.

Iris has no functionality for horizontal temperature gradients, but it should be achievable:

• I expect using numpy.gradient() on the Cube.data array will give you the raw numbers you need.
• You'll need to derive the coordinates (since each point will be half-way between the original points?).
  • You could use numpy.interp() on the DimCoord.points arrays for this.
  • Create new DimCoords using the resulting arrays.
• Create a new Cube using the array from the first step, and the new DimCoords from the second step.

[pp-mo]

@trexfeathers Iris has no functionality for horizontal temperature gradients

I'm not totally sure what you mean, though.
Would iris.analysis.calculus.differentiate be of any use in your case ?

If so, I think it does do some of the work to make new 'derived coordinates', as just mentioned.

[NMC-DAVE]

Very thanks @pp-mo ,
iris.analysis.calculus.differentiate may do the job if the cube coordinates are an equal distance like 5km delta. But for longitude/latitude coordinates, I have to calculate the temperature gradient on spheric coordinates with an unequal delta. The Metpy has metpy.calc.lat_lon_grid_deltas function to compute. So iris have the same function. Thanks again.
Dave