Surface data skipped in wrfvar_output

Post Reply
Kpeter
Posts: 1
Joined: Mon May 20, 2019 12:34 pm

Surface data skipped in wrfvar_output

Post by Kpeter » Mon May 20, 2019 1:17 pm

I would like to ask your assistance for my 3D-VAR data assimilation experiment. All elements of my experiment seem ok, the log files tell appropriate work, and wrfvar_output generated correctly.
I have checked the increments (difference between first guess and 3D-VAR output file) and they also seem correct on higher level. BUT on surface level something may be wrong. :cry:

Considering T2m there is slight and singular increment field, moreover U,V fields are identical in both files (no increments), however Jo calculation for synop_u and synop_v appears in log files:

jo_synop 0.211335854820E+03
jo%synop_u 0.107599699190E+03
jo%synop_v 0.186768876058E+01
jo%synop_t 0.101610705821E+03
jo%synop_p 0.257418691903E+00
jo%synop_q 0.342357448447E-03
jo_airep 0.362222449565E+02
jo%airep_u 0.113257960778E+02
jo%airep_v 0.000000000000E+00
jo%airep_t 0.248964488787E+02
jo%airep_q 0.000000000000E+00
jo_profiler 0.346260589824E+02
jo%profiler_u 0.261924596285E+02
jo%profiler_v 0.843359935392E+01

My first idea was that surface observation assimilation results are not written out, but I did not find this option in namelists. What could be wrong?
I have put FG, namelist, and output files together:

https://drive.google.com/file/d/1Sh70hr ... =drive_web


Thank you for your help in advance,
Péter

RCarpenter
Posts: 48
Joined: Tue Jun 12, 2018 11:18 pm

Re: Surface data skipped in wrfvar_output

Post by RCarpenter » Fri Jun 14, 2019 8:15 pm

I am having the same issue (perhaps). Fields like T2 and U10 are not being updated. But the model levels seem to be reasonably updated. How can I get the derived near-surface fields to update?

hclin2018
Posts: 31
Joined: Wed Jun 13, 2018 5:38 pm

Re: Surface data skipped in wrfvar_output

Post by hclin2018 » Wed Jul 24, 2019 8:31 pm

Please see the description about update_sfcdiags in var/README.namelist.

It is better to let the model diagnose and output those surface variables from the first couple time steps.
For example, in WRF namelist.input,
&time_control
auxhist6_outname = "./surface_d<domain>_<date>"
iofields_filename = "./surface_out.txt"
io_form_auxhist6 = 2
auxhist6_begin_m = 3 ! set a proper value according to your time step
auxhist6_end_m = 6 ! set a proper value according to your time step

Edit surface_out.txt file to include the following one line.

+:h:6:t2,q2,u10,v10

Post Reply

Return to “Operational Implementation”