What this gen_be error is due to?

[metroid2012]

Hi,
i want to produce a be.dat file from a month simulation period. the use of cv option 5 gave the fllowing error in the gen_be_diags.
At line 2551 of file da_gen_be.f (unit = 23, file = 'gen_be_stage3.chi_u.dat')
Fortran runtime error: End of file

Error termination. Backtrace:
#0 0x7f087509d2da in ???
#1 0x7f087509dec5 in ???
#2 0x7f087509e68d in ???
#3 0x7f0875214a33 in ???
#4 0x7f087521536f in ???
#5 0x7f0875215435 in ???
#6 0x7f0875217cec in ???
#7 0x56102b35f531 in ???
#8 0x56102b381665 in ???
#9 0x56102b31e3f2 in ???
#10 0x7f08746fcb96 in ???
#11 0x56102b31e429 in ???
#12 0xffffffffffffffff in ???

[1] Gather regression coefficients.

[2] Gather vertical error eigenvectors, eigenvalues.

waht is this due to? here is also the gen_be_wrapper.ksh script i used.

 

[kwthomas]

The error is "end of file". That means file "gen_be_stage3.chi_u.dat" is empty or doesn't have enough
data for the program.

 

[metroid2012]

The error is "end of file". That means file "gen_be_stage3.chi_u.dat" is empty or doesn't have enough
data for the program.

Does this mean my wrfout files are wrong? or where this error came from?

 

[kwthomas]

It's file "gen_be_stage3.chi_u.dat" that has the problem.

Since I've not used "gen_be", I can't give you more.

 

[metroid2012]

Thank you for the reply. I'll recheck everything and track the error.

 

[jorge_p]

Hi, did you finally track the error?

 

[metroid2012]

Yes, i managed to solve the problem. In fact i produced a single one long forecast and copy the wrfout files to the correct directories. But i noticed that i had to produce separate 24 hours forecasts. One needs also to try some ncl plots to check the results ( the gen_be_plot_wrapper.ksh would be a good starting point). An explanation on how to generate this be.dat is found here:


http://www2.mmm.ucar.edu/wrf/users/wrfda/OnlineTutorial/Help/gen_be_forecasts.html

 

[jorge_p]

Yes, i managed to solve the problem. In fact i produced a single one long forecast and copy the wrfout files to the correct directories. But i noticed that i had to produce separate 24 hours forecasts. One needs also to try some ncl plots to check the results ( the gen_be_plot_wrapper.ksh would be a good starting point). An explanation on how to generate this be.dat is found here:


http://www2.mmm.ucar.edu/wrf/users/wrfda/OnlineTutorial/Help/gen_be_forecasts.html

Hi, metroid thanks for updating your progress.

So If we need to generate a be.dat file from a month long series of 24hr forecasts, do you have a practical and fast way of running all these single 24hr runs in a more "automated" ,"simultaneous" way, if you are running wrf on a single multicore machine?

Thanks

 

[metroid2012]

hi,
I've got access to a HPC, so i run the mpi jobs on between 1 and 6 nodes (each node is a xeon 12 cores/24threads cpu) using batch scripts. you can reduce the amount of time from 1 month to at least 15 days (this this the minimum for scientific research purposes). Another good option is to use cv_options 7 or 6.
Good luck!