Unipost strange high, mid, low cloud cover output

All other topics about postprocessing model data (GrADS and other software), about other numerical weather prediction software (including WRF-NMM and WRF-ARW discussion unrelated to UEMS/WRF EMS), and general meteorology talk go in this forum.
oib
Posts: 118
Joined: Mon Jan 31, 2011 3:29 pm

Re: Unipost strange high, mid, low cloud cover output

Post by oib » Mon Oct 20, 2014 1:08 pm

surgeon wrote:
oib wrote: However wrf.exe from netcdf-3.6.3 seems a little faster (1%) than wrf.exe from netcdf-4.1.1.
This is interesting. :)
Hi, this is not 100% certain,since I've only done ONE test. :D
The difference may instead be the result of a "random" fluctuation from the average value.

meteoadriatic
Posts: 1584
Joined: Wed Aug 19, 2009 10:05 am

Re: Unipost strange high, mid, low cloud cover output

Post by meteoadriatic » Mon Oct 20, 2014 1:09 pm

surgeon wrote:
oib wrote: However wrf.exe from netcdf-3.6.3 seems a little faster (1%) than wrf.exe from netcdf-4.1.1.
This is interesting. :)
Well, I guess less features = little bit faster code. Usually is that way.

Oib,

here is my unipost v2.2:
http://www.meteoadriatic.net/pub/wrf/oib/unipost.exe

I guess it should work on all Centos6 64-bit platforms, at least if CPU supports SSE4.2. Try it and if you still have problem with this one, then your unipost is just OK and you should search for issue elsewhere in the system.

oib
Posts: 118
Joined: Mon Jan 31, 2011 3:29 pm

Re: Unipost strange high, mid, low cloud cover output

Post by oib » Mon Oct 20, 2014 1:26 pm

meteoadriatic wrote:Oib,

here is my unipost v2.2:
http://www.meteoadriatic.net/pub/wrf/oib/unipost.exe

I guess it should work on all Centos6 64-bit platforms, at least if CPU supports SSE4.2. Try it and if you still have problem with this one, then your unipost is just OK and you should search for issue elsewhere in the system.
Many many thanks!
Will try it very soon and let you know results ;)

oib
Posts: 118
Joined: Mon Jan 31, 2011 3:29 pm

Re: Unipost strange high, mid, low cloud cover output

Post by oib » Tue Oct 21, 2014 7:42 am

I have tried to re-build everything using system libraries from yum and the problem is still here. :(

I have used meteoadriatic unipost.exe and the problem is still here. :?

As meteoadriatic said, now I know my unipost.exe is good and the issue must be elsewhere in my system.

Now what direction I could go ???
At this moment I do not know which direction to take...

I'm thinking of abandoning Unified Post Processor and try to pull out the cloud cover directly from wrf_out files using a tool like NCL, that I have already successfully build on my system.
In particular ARW wrf_out netcdf files already contain by default the field CLDFRA, a 3D field representing the cloud fraction (values between 0 and 1) on the model levels.
Perhaps one can use this to calculate the low, middle and high level cloud fractions and plot these to a map...

Or maybe exists a way to instruct unipost to write a CLDFRA field into the GRIB file onto z-levels ?
if it were available a similar field in the GRIB file, then it should not be difficult write down a short GrADS script to calculate the total cloud cover between two arbitrary pressure levels or better between two arbitrary above sea height levels.

Note: my unipost.exe is 64.4 MB, meteoadriatic unipost.exe is 78.9 MB. Why this difference?

surgeon
Posts: 46
Joined: Wed Dec 08, 2010 1:34 pm
Location: Poland
Contact:

Re: Unipost strange high, mid, low cloud cover output

Post by surgeon » Tue Oct 21, 2014 10:16 am

oib wrote: Note: my unipost.exe is 64.4 MB, meteoadriatic unipost.exe is 78.9 MB. Why this difference?
My: 40.3 MB . GCC rox :lol:

oib
Posts: 118
Joined: Mon Jan 31, 2011 3:29 pm

Re: Unipost strange high, mid, low cloud cover output

Post by oib » Tue Oct 21, 2014 12:16 pm

At this point will not help in anything, in any case, to complete the exchange, here is unipost.exe compiled by me. It's 64 MB,

meteoadriatic
Posts: 1584
Joined: Wed Aug 19, 2009 10:05 am

Re: Unipost strange high, mid, low cloud cover output

Post by meteoadriatic » Tue Oct 21, 2014 12:30 pm

Difference in size... my is the biggest one, that's good :lol:

I have no idea what is going on there!

But, can you upload grib file from that wrfout you did for Adriatic region? I would like to look at grib to see what is different from my grib.

oib
Posts: 118
Joined: Mon Jan 31, 2011 3:29 pm

Re: Unipost strange high, mid, low cloud cover output

Post by oib » Tue Oct 21, 2014 1:04 pm

meteoadriatic wrote:Difference in size... my is the biggest one, that's good :lol:
:lol:
meteoadriatic wrote: I have no idea what is going on there!

But, can you upload grib file from that wrfout you did for Adriatic region? I would like to look at grib to see what is different from my grib.
Woilà!

data
ctl
idx

meteoadriatic
Posts: 1584
Joined: Wed Aug 19, 2009 10:05 am

Re: Unipost strange high, mid, low cloud cover output

Post by meteoadriatic » Tue Oct 21, 2014 5:44 pm

Definitely different. Not sure why.

Try also my http://meteoadriatic.net/pub/wrf/oib/wrf_cntrl.parm it has less fields than your, but just to be sure that your defined fields does not conflict unipost somehow.

Do you use multiple CPU threads for unipost?

oib
Posts: 118
Joined: Mon Jan 31, 2011 3:29 pm

Re: Unipost strange high, mid, low cloud cover output

Post by oib » Wed Oct 22, 2014 7:58 am

meteoadriatic wrote:Try also my http://meteoadriatic.net/pub/wrf/oib/wrf_cntrl.parm it has less fields than your, but just to be sure that your defined fields does not conflict unipost somehow.
Thanks for the file!

I have tried going with your wrf_cntrl.parm, but the problem is still here :(
meteoadriatic wrote:Do you use multiple CPU threads for unipost?
My system has two CPU nodes (Intel Xeon Processor E5-26xx), each has 12 cores.

Hyper-Threading is disabled.

UPP is compiled with "Linux x86_64, Intel compiler (dmpar)" mode.

I execute unipost as the following:

Code: Select all

/path/to/mpich/bin/mpiexec -n 24 /path/to/UPP/bin/unipost.exe >& ouput.log
I have also tried executing only on one core:

Code: Select all

/path/to/mpich/bin/mpiexec -n 1 /path/to/UPP/bin/unipost.exe >& ouput.log

Post Reply