blows up at 1st time step even after decreasing time step

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Message
Author
komet7
Posts: 11
Joined: Sun Aug 22, 2010 5:49 am
Location: university of florida

blows up at 1st time step even after decreasing time step

#1 Post by komet7 » Mon Feb 13, 2012 7:22 pm

Hi Rommers,

I am desperately seeking for your help. My model blows up at the first time step, even though I tried smaller time steps. My domain is 5km to 20km. Here is my configuration:

Tides: I analytically specified surface elevation (ana_fsobc), with 3 constituents.

Wind: I specified analytical constant wind sustr=.01 N/m2, svstr=.002 N/m2

River: discharge of 20m3/s. For depth is 1 m all at the carved area and until 10 m offshore.

Initial condition: S0=32; T0=10

Do you think it is because of the advective scheme due to grid size and wind correlation? or do you think I should carve more cells across river to let advective processes to evolve? Or should I play with the bathymetry so that just after the mouth it drops from 1 meter to 2 meters for example (instead of having 1 m depth untill 10 meters offsore)?

I attached my log file and below is my cppdefs. I appreciate your help.

#define UV_ADV
#define UV_COR
#define UV_QDRAG
#define UV_PSOURCE
#define DJ_GRADPS
#define TS_A4HADVECTION
#define TS_A4VADVECTION
#define TS_DIF2
#define MIX_GEO_TS
#define TS_PSOURCE
#define NONLIN_EOS
#define CURVGRID
#define SALINITY
#define MASKING
#define SOLVE3D
#define SPLINES
#define AVERAGES

#define LMD_MIXING
#ifdef LMD_MIXING
# define LMD_RIMIX
# define LMD_CONVEC
# define LMD_SKPP
# define LMD_BKPP
# define LMD_NONLOCAL
#endif

#define ANA_INITIAL
#define ANA_FSOBC
#define ANA_PSOURCE
#define ANA_SMFLUX
#define ANA_SRFLUX
#define ANA_SSFLUX
#define ANA_STFLUX
#define ANA_BSFLUX
#define ANA_BTFLUX
Attachments
my5.txt
(21.41 KiB) Downloaded 105 times
Last edited by komet7 on Wed Jun 06, 2012 6:53 pm, edited 1 time in total.

User avatar
kate
Posts: 3780
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: blows up at 1st time step even after decreasing time ste

#2 Post by kate » Mon Feb 13, 2012 8:24 pm

You don't want a change in bathymetry from 1 to 10 meters without a transition zone. A jump will show up as a "bad" value for rx0 (search the output for Beckmann). For a setup like this, I would start very simple and add one thing at a time. If it runs with winds, then add tides. If that runs, add the river. Build up a toolkit of how to look at the output, perhaps starting with ncview. When ROMS blows up, you usually get a restart record of the "bad" state which you can investigate. Is the trouble in the barotropic mode or baroclinic mode? At the bottom, surface, or the middle of the water column?

lanerolle
Posts: 157
Joined: Mon Apr 28, 2003 5:12 pm
Location: NOAA

Re: blows up at 1st time step even after decreasing time ste

#3 Post by lanerolle » Tue Feb 14, 2012 12:54 am

Why not run with a flat bathymetry of say 10m everywhere? If this works then there is an issue with your bathymetry but if it also blows-up then it is either your initialization fields or forcing fields. Also, in case you are not currently doing so, its best to ramp-up the winds (in the surface forcing), velocities in the open boundary forcing and the river discharge in the river/point source forcing. You can ramp-up linearly in time say over a period of 1-5 days and this may also help to stabilize your run.

nganju
Posts: 82
Joined: Mon Aug 16, 2004 8:47 pm
Location: U.S. Geological Survey, Woods Hole
Contact:

Re: blows up at 1st time step even after decreasing time ste

#4 Post by nganju » Wed Feb 15, 2012 1:43 pm

you have a dx between 13 and 58 m, and a dy between 13 and 84 m, a timestep of 100 s, and a Courant number of 4.
I would try reducing the timestep. Also make sure you have no depth values less than zero, even if they are masked.

komet7
Posts: 11
Joined: Sun Aug 22, 2010 5:49 am
Location: university of florida

Re: blows up at 1st time step even after decreasing time ste

#5 Post by komet7 » Mon Mar 12, 2012 6:10 pm

Thank you so much for all the replies. I have followed each suggestion. I removed all the other forcing, just kept the tides. I introduce the tides not abruptly but by using a function. I reduced the time step to 1 second. I now have sloping idealized shelf with straight coastline. However the model still blows up, not at the first time step tho. I checked the velocities until it blows up and they increase near the shore where my smaller sized grids are.

If you could give me any advice that would be great. I appreciate your time.

User avatar
kate
Posts: 3780
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: blows up at 1st time step even after decreasing time ste

#6 Post by kate » Mon Mar 12, 2012 6:18 pm

Perhaps you still need a smaller timestep. What is your Courant number now? Your spatial scales are much smaller than anything I have ever done - you should expect to need tiny timesteps as well.

nganju
Posts: 82
Joined: Mon Aug 16, 2004 8:47 pm
Location: U.S. Geological Survey, Woods Hole
Contact:

Re: blows up at 1st time step even after decreasing time ste

#7 Post by nganju » Mon Mar 12, 2012 6:22 pm

could you post an image of your grid, and perhaps a zoom-in of where the blow-up is happening?

komet7
Posts: 11
Joined: Sun Aug 22, 2010 5:49 am
Location: university of florida

Re: blows up at 1st time step even after decreasing time ste

#8 Post by komet7 » Tue Mar 20, 2012 7:46 pm

We realized that there is a bug in inp-par.F. After line 1370 [S(ibry,ifield,igrid)%Chapman = .TRUE.] the code missed the line [S(ibry,ifield,igrid)%acquire = .TRUE.]. However after fixing the code, my model still blows up after some time. I checked the mask and grid many times and I am even using homogeneous bathymetry with a straight coastline. I also realized that at the intersections of the open boundaries ubar increases before it blow up.

The way I defined the analytical boundary conditions is constant phases at north-south boundary and a varying one at the east boundary. I introduced the surface change slowly by using a function. Would you recommend me to use netcdf files (derived from larger scale tidal info) instead of using analytical functions.

Thank you so much for all your help and sincere responses.

Sincerely,

nganju
Posts: 82
Joined: Mon Aug 16, 2004 8:47 pm
Location: U.S. Geological Survey, Woods Hole
Contact:

Re: blows up at 1st time step even after decreasing time ste

#9 Post by nganju » Tue Mar 20, 2012 7:56 pm

are you missing a boundary cell in the analytical forcing? plot the output for ubar/vbar/zeta along the boundary and see if one cell is being held at zero instead of varying with your function.

Post Reply