DIAG speed trouble 20.6038127549509_Blowing-up

[kunal.c]

Dear All,

I am writing in connection to a problem which we are facing during ROMS-CoSiNE coupled run.

It is a Indian Ocean basin wide quarter degree model with 40 vertical layers. We have derived initial and boundary condition for the physical variables (temp,salt,u,v and zeta) from ECCO. The biological variables are initialized with the help of available WOA climatological data.

In our set-up, North and West boundaries are closed and South and East boundaries are open. Currently, closed boundary condition is used for biological variables.

Further, we have set following boundary conditions:

LBC(isFsur) == Clo Cha Cha Clo ! free-surface
LBC(isUbar) == Clo Fla Fla Clo ! 2D U-momentum
LBC(isVbar) == Clo Fla Fla Clo ! 2D V-momentum
LBC(isUvel) == Clo RadNud RadNud Clo ! 3D U-momentum
LBC(isVvel) == Clo RadNud RadNud Clo ! 3D V-momentum
LBC(isMtke) == Clo Red Red Clo ! mixing TKE
LBC(isTvar) == Clo RadNud RadNud Clo \ ! temperature
Clo RadNud RadNud Clo ! salinity

However, with 300 seconds time step, the model is blowing up after 17:50 hours. As suggested in ROMS wiki, I was trying a shorter time step (30 seconds) but still it is blowing up after 22:18 hours.

For your information, we used following parameters during grid preparation

Vtransform == 2 ! transformation equation
Vstretching == 4 ! stretching function

THETA_S == 07.0d0 ! surface stretching parameter
THETA_B == 0.1d0 ! bottom stretching parameter
TCLINE == 150.0d0 ! critical depth (m)

We observe that zeta increases, and reaches unrealistic values (reaches >9m!). Subsequently, zonal component of current is also reaching to a unrealistic value.

Please suggest me about the possible reasons of blowing up. Your suggestions and comments will be greatly appreciated.

Thank you very much in advance for your time and help.

Best,

Kunal

[kate]

Pictures - we want pictures. It is not a time-step issue, but something else.

[kunal.c]

Hi Kate,

Please find figures (prepared from ocean_rst.nc file) and logfile from following location:

URL:--ftp://ftpser.incois.gov.in

User Name:--temp

Password:--incoistemp

PATH: /home1/temp/kunal/*.gif

Best,

Kunal

[kate]

I would turn off the COSINE model until you get stable physics. I would change DSTART to match your initial time so that you don't overflow the timestep counter.

What are the units on your rain field? Mine is:

float rain(rain_time, lat, lon) ;
rain:long_name = "Total surface precipitation flux" ;
rain:units = "kg/m2/s" ;
rain:coordinates = "lon lat" ;
rain:time = "rain_time" ;
rain:_FillValue = 1.e+15f ;
rain:missing_value = 1.e+15f ;
rain:valid_range = -0.00017217f, 0.00189387f ;

[kunal.c]

Hi Kate,

I wiil let you know after performing following experiments:

The model run without CoSiNE.
After stable physics coupled (with CoSiNE) run.

Please find the following details:

float PRATE(ocean_time, lat, lon) ;
PRATE:missing_value = -1.e+34f ;
PRATE:_FillValue = -1.e+34f ;
PRATE:long_name = "PRECIP[GT=TAXIS]" ;
PRATE:history = "From precip.mon.mean" ;
PRATE:coordinates = "lon lat" ;



'PRATE' ! Input
'rain fall rate'
'kilogram meter-2 second-1' ! [kg m2/s]
'rain, scalar, series'
'ocean_time'
'idrain'
'r2dvar'
1.0d0 ! 1/rhow (0.001 m3/kg) if m/s

Best,

Kunal

[kate]

I assume you showed us surface T,u,v? What is the salinity doing? You didn't answer about the units on rain - your values are larger than mine. Perhaps you should change its scale factor?

[kunal.c]

Yes, surface plots.
Sea surface salinity is relaxed to the monthly climatological values derived from WOA 2009.
units on rain - kilogram meter-2 second-1.

[kate]

Are you listening? My global rain file says: rain:valid_range = -0.00017217f, 0.00189387f ;
Your rain was much, much larger than that. If your surface salinity goes nuts from rain, you need to be looking at it, not counting on a surface nudging to fix it.

[kunal.c]

I was running the model while changing the scaling factor of rain and waiting to see whether it is getting stable physics or blowing up.
Meanwhile, I answered your other questions.
Anyway, almost two month run is completed and it is working fine.
Thanks for your attention.