Project

General

Profile

Interpolation between ICON's hybrid sigma height grids

Added by Arthur Grundner about 4 years ago

Hi everyone,

I need to interpolate data between two different ICON hybrid sigma height grids.
Luckily, the remapeta function seems to offer this functionality.
As far as the z-axis description goes I have the vertical level height descriptions for every grid-point (also the orography).
However, the remapeta function expects an ASCII-file with the description of the z-axis.

How can I convert the explicit vertical level description into an ASCII-file so as to be able to use remapeta?

Thank you

Arthur


Replies (10)

RE: Interpolation between ICON's hybrid sigma height grids - Added by Ralf Mueller about 4 years ago

hi Arthur!

I really doubt that remapeta does the right thing with the vertical coordinate of ICON. it certainly does for ECHAM, but ICON has a different vertical coordinate (zg: strandard_name = geometric_height_at_full_level_center I think), which is not based on VCTs anymore. Instead it is a free 3d hight field. Originally I implemented the operator intlevel3d for that purpose, but that's some time ago and new developments have been done.

Depending on what kind of data you have, please contact the corresponding groups to clarify this. Could be Atmosphere department at MPI or someone at DWD.

sorry
ralf

RE: Interpolation between ICON's hybrid sigma height grids - Added by Arthur Grundner about 4 years ago

Thanks for your answer, Ralf. At least it solves some of my confusion.

I guess intlevel3d could be helpful for some variables, but in general I'm afraid I need a non-linear interpolation method.
I was thinking that maybe I could use vertintap/ap2hl and piece together the sigma height grid.

RE: Interpolation between ICON's hybrid sigma height grids - Added by Arthur Grundner about 4 years ago

Arthur Grundner wrote:

Thanks for your answer, Ralf. At least it solves some of my confusion.

I guess intlevel3d could be helpful for some variables, but in general I'm afraid I need a non-linear interpolation method.
I was thinking that maybe I could use vertintap/ap2hl and piece together the sigma height grid.

Although, the ICON-vertical grid is not completely free, is it? Seeing that it follows the SLEVE coordinates.

RE: Interpolation between ICON's hybrid sigma height grids - Added by Ralf Mueller about 4 years ago

Arthur Grundner wrote:

Arthur Grundner wrote:

Thanks for your answer, Ralf. At least it solves some of my confusion.

I guess intlevel3d could be helpful for some variables, but in general I'm afraid I need a non-linear interpolation method.
I was thinking that maybe I could use vertintap/ap2hl and piece together the sigma height grid.

Although, the ICON-vertical grid is not completely free, is it? Seeing that it follows the SLEVE coordinates.

sure - what I mean was, that there is no formula like in hybrid sigma pressure coordinates

Why do you think, you need non-linear interpolation?

RE: Interpolation between ICON's hybrid sigma height grids - Added by Arthur Grundner about 4 years ago

Hi Ralf,

well, for pressure, wind and temperature fields.

I tried using intlevel3d and noticed that I'm running into this error: https://code.mpimet.mpg.de/issues/9468
It seems it takes at least two weeks until intlevel3d works properly on Mistral. :-/

RE: Interpolation between ICON's hybrid sigma height grids - Added by Ralf Mueller about 4 years ago

if the fix is in the development version, I can install it on mistral for u in no time

RE: Interpolation between ICON's hybrid sigma height grids - Added by Arthur Grundner about 4 years ago

Ralf Mueller wrote:

if the fix is in the development version, I can install it on mistral for u in no time

Thanks, Ralf!

RE: Interpolation between ICON's hybrid sigma height grids - Added by Ralf Mueller about 4 years ago

alright, here you go: you can grab the binary from mlogin108 under /tmp/cdo. it should work everywhere on mistral. so just copy it to a place u like and don't forget to set PATH or an alias .... yh, u probably know that ;-)

hth
ralf

    (1-10/10)