Blowing-up when using COAWST to couple ROMS and SWAN

[Rocky_Wang]

Hello everyone,
I used to run the ROMS model with SWAN output files(wave-orbital velocity ub,wave period T, and wave-propagation direction,) to investigate sediment transported processes dominated by currents and waves. now I want to use COAWST model to couple them together. The C-preprocessing Options are according to inlet_test.h. In my ROMS model,the boundary forcing files data are from　HYCOM(the open boundarys are south and east), the atmosphere forcing data are from ECMWF. My SWAN model is forced by CCMP wind. The GLS turbulence closure use K-omega values.
the coupled model run successfully, but it Blowing-up after two month calculation.And I can't find the reason. The attachment are my .h file and log file. Can you give me some clues. Thank you very much!

[jcwarner]

Your stdout says " 17974 62 09:50:00 9.874450E-02 2.791750E+02 2.792738E+02 2.775544E+13
(095,245,20) 0.000000E+00 1.108514E-02 8.481048E+00 6.956400E+01
DIAG speed trouble 69.5640000538570
Blowing-up: Saving latest model state into RESTART file
DIAG speed ijk 98.3770376595520 95 246 1 1"

So you need to look at the history file and the RST file and see where the ocean model is getting large velocities. maybe near point 95 245. Is it along the open boundary or at some wet/dry point... is the water level really low some place and causing large velocities, etc.
-j

[Rocky_Wang]

Thanks John. Sorry for the late reply, I attended a field survey in recent days.
My model is calculated from 2007 to 2016, It always Blowing-up in 2007/3/04. So I have made some test to find the reason. For example, use the 2016 wind file instead of the 2007 wind file for swan Model. I find only in this way, the model can pass the day 2007/3/04. I think wind is main reason to result in the Blowing-up. I checked the wind data of that day, the max wind speed can reach 25 m/s, and the significant wave height can reach 4m. they are vary large numbers to my model domain.
According to your guidance, I find the blowing-up point near the land, the bath is 5m. Maybe the large wind speed caused low water level of that point and then the model figure out a large speed. How to deal with this situation? I didn't turn on WET_DRY cpp in my .h file, If it works when turn on it?

[Rocky_Wang]

Your stdout says " 17974 62 09:50:00 9.874450E-02 2.791750E+02 2.792738E+02 2.775544E+13
(095,245,20) 0.000000E+00 1.108514E-02 8.481048E+00 6.956400E+01
DIAG speed trouble 69.5640000538570
Blowing-up: Saving latest model state into RESTART file
DIAG speed ijk 98.3770376595520 95 246 1 1"

So you need to look at the history file and the RST file and see where the ocean model is getting large velocities. maybe near point 95 245. Is it along the open boundary or at some wet/dry point... is the water level really low some place and causing large velocities, etc.
-j

Thanks for your guidance!

I turn on #define WET_DRY in my .h file, and set DCRIT == 1.0d0 in my .in file. and the model passed the day 2007/3/4 successfully.

I read your paper A wetting and drying scheme for ROMS. But there are still something puzzled me.

1) When turn on WET_DRY, the computation speed was slowing down obviously.

2) I have coupled ROMS to SWAN model, they exchange water level data every 3 hours, so if the WET_DRY scheme influence the SWAN model?

3) I change depmin=0.05 to depmin=1 (same as DCRIT )in my swan.in, to exclude dry points during wave computations. why some parameters of SWAN turn into zero during the data exchange.

[jcwarner]

the water levels go from roms to swan, so wet dry will affect swan.

when you increase the min depth for swan to 1 m, then there will be locations that will turn dry and this will change the wave dynamics in the very shallow parts. suggest you do 2 runs with the different depths, and then pcolor some of the fields to see where the differences are. this might make it m ore obvious.
-j