Input error when including boundary conditions in 4DVar

Discussion about tangent linear and adjoint models, variational data assimilation, and other related issues.

Moderators: arango, robertson

Post Reply
Message
Author
Arvin_Yin
Posts: 15
Joined: Wed Apr 23, 2014 12:05 pm
Location: Tsinghua University

Input error when including boundary conditions in 4DVar

#1 Post by Arvin_Yin » Tue Jul 21, 2015 9:14 am

Hi, everyone!

I have successuflly run the I4D-Var model without "ADJUST_BOUNDARY", , then I defined "ADJUST_BOUNDARY", but I got this error:

<<<< 4D Variational Data Assimilation, Outer = 001, Inner = 000 >>>>

INITIAL: Configuring and initializing forward nonlinear model ...
*******

NETCDF_CHECK_DIM - inconsistent size of dimension: obc_adjust 10 41
in file: ./in/ini_hycom_daily_130101.nc

The file is my initial condition file.

I have checked mode_netcdf.F, and I found:

#ifdef ADJUST_BOUNDARY
CASE ('obc_adjust')
IF (dim_size(i).ne.Nbrec(ng)) THEN
IF (Master) WRITE (stdout,10) TRIM(dim_name(i)), &
& dim_size(i), Nbrec(ng), &
& TRIM(ncname)
exit_flag=2
EXIT
END IF

But I still not clearly understand why the error happen.

Anyone help?

Arvin_Yin
Posts: 15
Joined: Wed Apr 23, 2014 12:05 pm
Location: Tsinghua University

Re: Input error when including boundary conditions in 4DVar

#2 Post by Arvin_Yin » Fri Jul 31, 2015 3:11 am

Hi, all.

Now I know that this error occur because I have wrong 'NOBC' number in ocean.in file.

Nbrec=1+NTIMES/NOBC should be 10 as the error say.

I don't know why the size of 'obc_adjust' have to be 10. So, it's a new question.

Thanks for your help.

Gabriel_Carvalho
Posts: 9
Joined: Tue Jul 05, 2011 2:47 pm
Location: Prooceano

Re: Input error when including boundary conditions in 4DVar

#3 Post by Gabriel_Carvalho » Wed Dec 14, 2016 8:20 pm

Hi,
I recently had the same problem.

I got this error changing the assimilation window lenght.

My ini file comes from my last assimilation cycle (with different NTIMES) and has the vars: zeta_obc, salt_obc, temp_obc, u_obc, v_obc, ubar_obc and vbar_obc, all of them with obc_adjust dimension.

The obc_adjust dimension of this initial file is the number of boundaries adjustments made in the last DA cycle (Nbrec=1+NTIMES/NOBC).

I didn´t want to change NOBC, but as I changed NTIMES, check_dims complains.

My solution was to erase all the variables that has the *_obc from my initial file (I think that they were there unnecessarily anyway).

regards,

Post Reply