[RegCNET] Errors in "./icbc regcm.in" process

[Huanghe, Gu]

Dear Graziano and others,

I'm glad to see your comment on this question. I have a similar problem when reading the NNRP2 data when execute the command "./icbc regcm.in". 

I modify the "mod_ncep.f90" according to illustrations in "ncep_diff". I delete the lines with '-' and add the lines with '+' in place. At the same time, I change the array 'ivar(7)' to 'ivar7(7)' in "istatus = nf90_get_var(inet,ivar(7),work,istart,icount) " . Then I recompiled the regcm4 with new "mod_ncep.f90" file. But the same error still exist.

This same NNRP2 data could be read by regcm3. Did I make any mistake when correct the mod_ncep.f90 or any incorrect setting in regcm.in?

Please give me some suggestions. 
Thank you very much!

Huanghe
________________________________________
From: regcnet-bounces at lists.ictp.it [regcnet-bounces at lists.ictp.it] on behalf of Graziano Giuliani [graziano.giuliani at aquila.infn.it]
Sent: Tuesday, November 09, 2010 10:33 AM
To: regcnet at lists.ictp.it
Subject: Re: [RegCNET] Errors in "./icbc regcm.in" process

On 11/09/2010 03:26 PM, 炳亮庄 wrote:
> Dear all
>   I want to run regcm4 over East Asia.  However, I was hold up in process of
> generating ICBC* file. The error might be led to the Terrain Elevation over
> West China is list as followed:
>   ../scratch/RCM4simulation/DATA/NNRP1/2010/uwnd.2010.nc
>
>      9.9999997764825821E-003    202.6600036621094
>             10
>   ../scratch/RCM4simulation/DATA/NNRP1/2010/vwnd.2010.nc
>
>      9.9999997764825821E-003    202.6600036621094
>   READ IN fields at DATE:   2010031500
>   Error: the RCM surface is above the GCM
>   model top at i,j=            1           42
>   This might reasonably happen if you have a
>   very tall mountain in your domain. Setting
>   p top to a smaller value could help.
>   Otherwise, this indicates a bug somewhere...
>     3.9377335E-02  =>      2.937734     ,     1.000000
> FORTRAN STOP
>
> According to the statements above, I adjusted the top pressure to a very
> small value, it still has no work.
[...]

 From previous reports, it seems that this error is triggered from a faulty read
from the input NetCDF file: we have hardwired in the code that variable of
interest in files is the fifth one (look at nf90_get_var calls in mod_ncep),
Instead, NCEP dataset have been changed, and, as we did discard error checking,
processing proceeds to interpolation on bogus data (and correctly finds vertical
interpolation errors...).

As this errors have been corrected in the upcoming release, my suggestion can be:

1) Wait for the 4.1 release
2) Correct the mod_ncep.f90 using the following patch.

Ciao!

Graziano.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: mod_ncep.f90
URL: <https://lists.ictp.it/pipermail/regcnet/attachments/20101110/1a9305a6/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: regcm.in
URL: <https://lists.ictp.it/pipermail/regcnet/attachments/20101110/1a9305a6/attachment-0002.in>