[Wrf-users] UPP V2.2 - Unipost Error

[Dominikus Heinzeller]

Hi Warren,

I had this problem for different reasons (with V2.1, though, but from the release notes I take that this is still valid).

(1) In one particular case I could track it back to the source file w3fi72.f. Basically, the solution was to pass modify the allocation of the the arrays KBUF and BMS around lines 190-200:

+!      CHARACTER * 1   KBUF(*)
+      CHARACTER * 1   KBUF(ITOT)
+      INTEGER, INTENT(INOUT) :: ITOT
       CHARACTER * 1   PDS(*)
       CHARACTER * 1   GDS(200)
-      CHARACTER(1),ALLOCATABLE:: BMS(:)
+      CHARACTER(1), DIMENSION(1:6+(IBLEN+16)/8) :: BMS

This however required a few modifications to the code since the interface to this particular routine is changed (ITOT required).

(2) Are you using the USGS LU database? I found that in lots of places the USGS 24-class data is hardcoded (in V2.1) and the code ends up treating a water field as non-water field and vice-versa. This can lead to nasty problems, including the double-free or corruption ones you have.

(3) A modified wrf_cntrl.parm can cause UPP to assume that some data, required for other derived variables, has been calculated beforehand but in fact it hasn’t. This needs quite a few patches to various MDL2xxx.f files.

Please contact me directly if you want further details or if you want me to test your data with my modified version of UPP.

Cjeers

Dom

On 24/04/2014, at 9:12 pm, Pettee, Warren <wpettee at uncc.edu> wrote:

> Hi again,
> 
> Does anyone know what would cause this error to be returned from unipost:
> 
>  *** glibc detected *** /home/wpettee/Models/WRF_NMM/UPPV2.2/bin/unipost.exe: double free or corruption (!prev): 0x0000000053140070 ***
> 
> When I look at one of the unipost_*.out files, nothing serious seems to stick out before the
> message:
> 
> BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
> EXIT CODE: 134
> 
> I can send my compile/configure logs if needed, but everything seemed to compile nicely. I've been trying multiple strategies to diagnose the
> problem, but I must be overlooking something. System is x64 Ubuntu 12.04, Xeon 2 GHz. UPP is configured for serial
> gcc/gfortran.
> 
> 
> Thanks,
> 
> 
> Warren Pettee
> wpettee at uncc.edu
> 
> _______________________________________________
> Wrf-users mailing list
> Wrf-users at ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/wrf-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/wrf-users/attachments/20140424/4fa6fcba/attachment-0001.html