Stream: python-questions

Topic: Interpolating onto regular geopotential grid

view this post on Zulip Daniel Marsh (Dec 22 2021 at 22:43):

Hello all,

I seem to be having a hard time figuring out how to put a vertical timeseries onto a regular geopotential/geometric grid for comparison to ground-based observations.
If I have a model output on pressure levels, e.g., T(lev,time) and geopotential height Z3(lev,time) how best to create an new variable Tz interploted onto a regular Z3 grid (e.g. every 1e3m)? I tried something like:

from scipy import interpolate

zkm = np.arange(1.,50.,2.)
f = interpolate.interp2d(Z3*1e-3, X, zkm)

where Z3 and X are single location timeseries from a CESM run, and so just level and time. In this example, I am trying to put it on a 2 km grid.
But it expects zkm to be the same size as Z3,X. I really just need to do 1-D interpolation in the vertical, so interp2d seems overkill. Apart from looping over time and doing 1d interpolation is there an elegant way to do this?

thanks,

Dan

view this post on Zulip Daniel Marsh (Dec 23 2021 at 00:13):

Well, looping over time seems to be pretty quick, so I have gone with that. Just had to remember that np.interp expects x to be increasing, so had to np.flip the vertical arrays.

Last updated: Jan 30 2022 at 12:01 UTC