DIAG speed trouble 20.125553145124616. Blowing-up: Saving latest model state into RESTART file

[Sankar]

Hello everyone. Recently I have encountered with the model blow up issue. I am new to modelling and want to couple wrf and roms using coawst (MCT). For that i want to run roms model individually, but after 6400 steps, the model blows up at a certain point near the coast. I masked that point and then the model runs for 13510 steps and then it blows up again at a different point which is also near the coastline. Although the model blowing up issue is sorted after masking that point, I fear that the model will be blowing up too many times if my no of time steps is more. So in that case, I guess that masking or unmasking the point is not efficient. So can you please have a look at the log file and let me know if this is the only way or there is an another way out.
I have created my grid file using croco tools and initial and boundary file using model2roms tool. I am using the etopo2 bathymetry data.

Thanking you for your attention to my post.

[jcwarner]

this configuration is for ROMS by itself.
the model runs for a while, as you say, and then

13510 2010-01-04 03:03:20.00 5.896200E-02 1.757339E+04 1.757345E+04 1.160390E+16
(127,016,32) 1.571469E-02 1.362909E-02 7.575998E+00 2.012555E+01
DIAG speed trouble 20.125553145124616
...
Blowing-up: Saving latest model state into RESTART file
WRT_RST - wrote re-start fields (Index=1,1) in record = 0000003

this is saying near i=127 j=16 and z level 32 there is a large speed (velocity).
I suggest you look at the rst file, time index 3 and see what the fields look like. Where is 127, 16? at the coast? near a sharp bathy gradient? did something creep in from the boundary?
you need to become a slight detective to look at the output and time evolution in the his files. you have 226 his time steps so look at how the flow was developing.

-j

[Sankar]

Thank you Dr. Warner for your quick response. As you suggested, I have checked the point (127,16) and it lies near the coast more specifically the Somalia coast (as my domain is Arabian Sea). The u momentum and v momentum component has very large deviation there, i.e., -21m/s and 15 m/s respectively which is very unusual. At all the depth levels this value is quite high in the restart file. At that point the bathymetry data does not have any sharp gradient and the depth is around 100m.
In the history file, the u component has velocity more than 15 m/s at 329,12,12 to 329,12,32 and 130,22,12 to 130,12,32, this increase starts after 202 time steps. Both of these points are located near the coast, one is near the Somalia coast and another at the south west coast of India. u component has velocity less than -15m/s around 128,18 which also includes 127,16 and almost at all depth levels and this decrease starts from 172 time steps.
The v momentum component also possess similar kind of unusual values but these are mostly restricted to the Somalia coast and in the Gulf of Aden.
From checking the output history files, it is very evident that some variables have unrealistic values at some points which lie near the coastline and most probably that's the reason behind the blowing up of the model. Can you please help me how to get rid of this issue if any, other than masking those points manually.

Thank you for your attention.

[susonic]

Dear Sankar,

I just saw that your E-P file shows large value which is possibly filled with blank or NaN value.
Please check out fresh water flux file.

-Joonho

[Sankar]

Thank you Dr. Joonho for your quick reply. I have checked the freshwater input flux and the values are similar to other forcing files i.e., shortwave radiation and taux. I created an another forcing file using era5 total precipitation and evaporation but the model is still showing same large minimum and maximum value and it blows up at the same point. I guess I need to dig in a little bit more as Dr. Warner suggested. In the meantime if you think that there may be some other reasons for the model to blow up then please let me know.

Thanking you for your valuable attention and suggestion to my post.

[susonic]

Well if that is the case,

First of all, I would plot every snapshot of temperature, salinity, current and sea surface height variable.

And also I would run the model without surface forcing and lateral boundary condition(closed).

If the model is stable without forcing(surface/lateral), then I would add forcing one after another.

-JH

[Sankar]

Thank you Dr. Joonho and Dr. Warner for your attention to my post. It took some days for me to resolve this blow up issue. I have changed the boundary schemes and now the model is running fine.