[ncl-talk] Converting hybrid sigma-pressure coordinate to pressure level

[Gustavo Correa]

Hi Xi

I am speaking about GFDL AM3/CM3, so if you're using AM2.1/CM2.1 
or older, you need to check if this is correct for those older versions.

I just went through the same problem a couple of weeks ago.

In the history/postprocessed files, 
GFDL renames as  "pk" what is in fact "ak".

Note however, that contrary to NCAR models,
which uses this conversion formula:

pressure(i,j,k,n)=ak(k)*p0 + bk(k)*ps(i,j,n)

AM3 (maybe older versions as well) use

pressure(i,j,k,n)=ak(k) + bk(k)*ps(i,j,n)  


where ak is just renamed
to pk in the netCDF output files.

In the formulas above, i,j,k,n are respectively the indices for longitude,
latitude, level, and time, and p0 is the reference pressure (at the surface).
Although the mean sea level pressure p0 is 1013hPa, models
normally use p0=1000hPa.

Note that in the NCAR scheme both ak and bk are 
non-dimensional.
However, in the GFDL AM3 scheme bk is non-dimensional,
but ak (renamed pk) has dimensions of pressure.
Actually ak (or pk) is in Pascal (NOT mb or hPa).

This is probably handled directly by vinth2p, or at most you need
to rescale ak=pk/p0 before you use vinth2p.
Please, check the vinth2p documentation before you use it.

A good reference is the Donner et. al. 2013 Journal of Climate 
article about AM3.
It has a table clarifying the meaning of ak, bk, etc, although
it doesn't mention that in the output files ak is renamed pk.

I hope this helps,
Gus Correa

On Apr 28, 2017, at 12:07 AM, Xi Chang wrote:

> Hi NCL team,
> 
> I am currently running some simulations with GFDL model + topography.
> So, I aware that pressure level_interpolation is needed when topography is non-zero due to terrain-following sigma coordinates. I know that I can use this function from NCL "vinth2p" but I only have "bk", "pk", and "ps". from the output file, but not the "ak". Can you suggest me how to solve this issue?
> 
> Thanks
> Dr. Chang
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk