Problem with ROMS-SWAN Coupling

[nirnimesh]

Hello Everybody

I am trying to couple ROMS AND SWAN for a nearshore application like shoreface example distributed with the ROMS code. The program compiles without any error, but when running the executable I get the following error first time:

STEP Day HH:MM:SS KINETIC_ENRG POTEN_ENRG TOTAL_ENRG NET_VOLUME

0 0 00:00:00 0.000000E+00 2.438617E+01 2.438617E+01 4.466439E+05
p1_29972: p4_error: net_recv read: probable EOF on socket: 1
p1_29972: p4_error: net_recv read: probable EOF on socket: 1
-bash-3.2$ p3_29984: (0.503906) net_recv failed for fd = 6
p3_29984: p4_error: net_recv read, errno = : 104
rm_l_3_29985: (0.503906) net_send: could not write to fd=5, errno = 32
rm_l_2_29979: (0.601562) net_send: could not write to fd=6, errno = 9
rm_l_2_29979: p4_error: net_send write: -1
PGFIO-F-231/formatted read/unit=50/error on data conversion.
File name = force.mat-001 formatted, sequential access record = 17
In source file swanparll.f90, at line number 3791
p4_error: latest msg from perror: Bad file descriptor
p3_29984: (2.507812) net_send: could not write to fd=5, errno = 32


****************************************************************************************************
After this, I remove all *.mat, swaninit etc generated using SWAN. On recompiling and running the program next time I get the following error:


STEP Day HH:MM:SS KINETIC_ENRG POTEN_ENRG TOTAL_ENRG NET_VOLUME

0 0 00:00:00 0.000000E+00 2.438617E+01 2.438617E+01 4.466439E+05
p0_6211: p4_error: net_recv read: probable EOF on socket: 1
Killed by signal 2.
Killed by signal 2.
Killed by signal 2.
*******************************************************************************************************

I looked through previous posts showing similar errors, but was not able to find a solution.

I am attaching the ROMS output on the screen(results.txt) with this post. It would be nice, if someone can take a look at it and guide me with this problem.

Thanking You
Nirnimesh

[jcwarner]

i searched the roms pages for
"p4_error: net_recv read"
and came up with 2 related issues
1) was about a fix for iopt(ilen) which has been released in the latest version.
2) the second was an issue with intel ifort v 10.

Are you using the latest version of the code, and what flavor of ifort are you using?

[nirnimesh]

Hi John

Thanks for the reply

1) I checked the line in Swanparll.F file in Waves/SWAN/Src/ folder. The iopt(ilen) correction has already been done there.
2) I am not using IFORT but PGI compiler.
I am using ROMS v 3.0 , SVN revision 240:
When I run the previous examples discussed as a part of ROMS Workshop in Savannah(May 2008), they work perfectly fine with ROMS and SWAN coupling

Can you suggest other reasons which cause similar error?

Thanking You
Nirnimesh

[jcwarner]

so let me get this correct.
If you compile, lets say, inlet test that is part of the distributed code, does it work correctly? Even tiled 2 processor for roms and 2 for swan.
But your own application is having troubles?

[nirnimesh]

John

Yes, INLET_TEST, SHOREFACE work fine for distributed processors (2 for ROMS and 2 for SWAN). But my own application doesn't work. I am checking one of the grid input files for SWAN right now. Let me get back to you after working on it for a while.

Thank you for looking into the problem

Nirnimesh

[nirnimesh]

John

I found the error .
There was a mismatch between the grid dimensions for ROMS and SWAN grid ( ). When I change the SWAN grid input file, the model works fine other than blowing up after 10 steps ( ).
I think I will play around a bit with the boundary conditions etc. to solve the blowing up.

Thank you once again for looking into the problem.

Nirnimesh

[dukhovskoy]

Hi,

I compiled and ran the INLET test case, however, simulated wave heights and velocities look suspeciously small. On the wiki page describing this experiment, there are no figures or any output results (quantitative) that could be used for comparison. Can somebdy who has successfuly run the test, post their figures of the output fields (e.g., SSH, Wave height, U/V)? Or some values of these characteristics at several locations that can be used for validation?

Thank you,
Dmitry

[dukhovskoy]

Hi,
I compiled and ran the INLET test case, however, simulated wave heights and velocities look suspeciously small.

In case somebody else is having problems with the INLET TEST case this may help. I've figured out what caused small SSH in the ocean model. In the ROMS version that I downloaded couple weeks ago, the OB conditions used in this INLET test case (for the ocean component ) do not work. The prescribed tide wave at the OB does not propagate into the domain. In the inlet_test.h the OBCs at the Northern boundary are:
#define FSOBC_REDUCED
#define NORTH_FSGRADIENT
#define NORTH_M2REDUCED
#define NORTH_M3GRADIENT

The following OBCs allows tidal wave propagation in the domain:
#define FSOBC_REDUCED
#define NORTH_FSNUDGING
#define NORTH_M2FLATHER
#define NORTH_M3RADIATION
#define WEST_FSCHAPMAN
#define WEST_M2RADIATION
#define WEST_M3RADIATION
#define EAST_FSCHAPMAN
#define EAST_M2RADIATION
#define EAST_M3RADIATION

Also, in the input file (ocean_inlet_test.in) I modified:
OBCFAC == 1.5d0 (it was 0.0d)
and prescribed a strong nudging (a very short, less than a day, relaxation time scale) which was 0 before.

The SWAN component works fine.

Dmitry

[jcwarner]

Dmitry-

thank you for bringing this to our attention. Actaully, about a month ago a change was made to the way certain boundary routines were activated (i just posted src ticket 227 in response to src 214). This is being evaluated now.

When this is corrected, then the tide will be imposed as I had originally designed with the FSOBC and NORTH_M2REDUCED . Your changes are to use a different way to impose a free surface displacement at the northern end. That may work fine, but I do not recommend that users switch to that. I recommend that the trunk be modified back to a way that allows those BC's as before. Please stay tuned.
Also, I will try to get some info on the wiki as to the results for that test case.

[jcwarner]

Dmitry-

The trunk has been updated to correct the issue that you identified. ( a fix in less than 1 hour - great service !! Thank you Hernan !).

I am running the code now. It looks like the trunk is now correct. I will compare the results to my default set of output at the office tomorrow. Please update your code and try the inlet_test as it is originally distributed. I will post results online so that you can compare.
-john

[dukhovskoy]

John, thank you for your help.
Dmitry

[jcwarner]

I uploaded some images of ROMS ssh and SWAN Hwave to
https://www.myroms.org/wiki/index.php/INLET_TEST_CASE
I am NOT the best at creating these wiki page images, but hopefully these images will help.
-j