Hi all,
I'm Arianna and I'm a beginner with Roms. I'm trying to run the Weddell application with ROMS 2.1 version. I'm working with that version, because I would like to run again the Weddell application coupled with a ice dynamic and thermodynamic module, that works with ROMS 2.1 at the moment.
In order to simplify the application, I defined the climatology of tracers and momentum through analytical expressions instead of a climatology file. Unfortunately the run blows up at the first time step, and after many attempts to vary the initial conditions I got the same result:
blows up....exit flag:1.
The specifications of the application are:
DT == 20.0d0
NDTFAST == 4
GAMMA2=-1
Xsize=4000.0*REAL(Lm, r8 ) m
Esize=4000.0*REAL(Mm, r8 ) m
depth=4500.0 m
Lm = 300
Mm = 5
N = 30
and the cppdfs.h is:
# elif defined WEDDELL
/*
** Options Idealized Weddell Sea Application: Tides and Ice Shelf Test.
*/
#define UV_ADV
#define DJ_GRADPS
#define UV_COR
#define TS_A4HADVECTION
#define TS_A4VADVECTION
#define SOLVE3D
#define SALINITY
#define NONLIN_EOS
#define CURVGRID
#define SPLINES
#define ICESHELF
#define AVERAGES
#define NS_PERIODIC
#define EAST_VOLCONS
#define WEST_VOLCONS
#define RADIATION_2D
#define EAST_FSCHAPMAN
#define EAST_M2FLATHER
#define EAST_M3RADIATION
#define EAST_TRADIATION
#define WEST_FSCHAPMAN
#define WEST_M2FLATHER
#define WEST_M3RADIATION
#define WEST_TRADIATION
#define ANA_M2CLIMA
#define ANA_TCLIMA
#define ANA_GRID
#define ANA_INITIAL
#define ANA_FSOBC
#define ANA_M2OBC
#define ANA_SMFLUX
#define ANA_STFLUX
#define ANA_SSFLUX
#define ANA_SRFLUX
#define ANA_BSFLUX
#define ANA_BTFLUX
All the surface fluxes are zero. The free surface and 2D momentum open boundary conditions and the zeta initial conditions are set up from M2 Tide as in the distributed WEDDELL application.
Initial and climatological data for tracers are rispectively -0.28 degrees for temperature and 34.65 PSU for salinity.
I tried both cases:
1) zeta(i,j,2)=zeta(i,j,1)
and
2) zeta(i,j,2) differs from zeta(i,j,1) because of the time delay in the phase.
Anybody can help me to find the error?
Maybe I'm short-sighted , because I don't see it, but it must be trivial otherwise the model doesn't blow up at the first time step!
Thank you very much in advance
Arianna
Modified Weddell application
Modified Weddell application
Last edited by Arianna on Tue May 20, 2008 9:56 am, edited 2 times in total.
I see you have NS_PERIODIC. Is this on an f-plane with beta=0? Otherwise you might get trouble at the NS ends.
You don't have sea ice turned on at the moment, just the ICESHELF, which is a static shelf sitting over the water, depressing the sea surface by the amount zice. If you want the dynamic/thermodynamic sea ice, one option is to get Dave Robertson to give you access to my svn branch.
Other things to try: turning off the VOLCONS switches and going to a lower order tracer advection scheme.
You don't have sea ice turned on at the moment, just the ICESHELF, which is a static shelf sitting over the water, depressing the sea surface by the amount zice. If you want the dynamic/thermodynamic sea ice, one option is to get Dave Robertson to give you access to my svn branch.
Other things to try: turning off the VOLCONS switches and going to a lower order tracer advection scheme.
Dear all,
I found the reason because the application blows up at the first time step. There is an error in the analitical definition of the distributed Weddell test bathymetry. The later isn't defined for i=IstrR and i=IendR, so h(i,j) drops to zero at the eastern and western edges of the grid (geographical south and north boundaries rispectively for x-axis 90 rotated). So I corrected the Bathymetry paragraph of the analytical.F file adding:
DO j=JstrR,JendR
h(IstrR,j)=h(IstrR+1,j)
h(IendR,j)=h(IendR-1,j)
END DO
Now the application runs until the end .
But your osservations, Kate, are right. In fact watching into the results I think there are some questions, such as the non zero ubar and vbar at the eastern edge, related to the definition of the boundary conditions,which I must study deep. I will try both your suggested changes to the application in order to make more realistic the case. In fact, the Coriolis parameter vaies in the application with i and j and is defined as
f(i,j)=2.0_r8*7.2E-05_r8*SIN((-79.0_r8+REAL(i-1,r8 )*val1)*deg2rad)
where val1=Asin(4000.0_r8*REAL(Lm(ng),r8 )/Eradius)/REAL(Lm(ng),r8 )
and 4000 m is the x-spatial resolution.
Thank you Kate for the reply:
I will be very happy to write to dr. Robertson and to get access to your svn branch, because I'm wondering yet which ice dynamic/thermodynamic model is better or my aims.
Cheers
Arianna
I found the reason because the application blows up at the first time step. There is an error in the analitical definition of the distributed Weddell test bathymetry. The later isn't defined for i=IstrR and i=IendR, so h(i,j) drops to zero at the eastern and western edges of the grid (geographical south and north boundaries rispectively for x-axis 90 rotated). So I corrected the Bathymetry paragraph of the analytical.F file adding:
DO j=JstrR,JendR
h(IstrR,j)=h(IstrR+1,j)
h(IendR,j)=h(IendR-1,j)
END DO
Now the application runs until the end .
But your osservations, Kate, are right. In fact watching into the results I think there are some questions, such as the non zero ubar and vbar at the eastern edge, related to the definition of the boundary conditions,which I must study deep. I will try both your suggested changes to the application in order to make more realistic the case. In fact, the Coriolis parameter vaies in the application with i and j and is defined as
f(i,j)=2.0_r8*7.2E-05_r8*SIN((-79.0_r8+REAL(i-1,r8 )*val1)*deg2rad)
where val1=Asin(4000.0_r8*REAL(Lm(ng),r8 )/Eradius)/REAL(Lm(ng),r8 )
and 4000 m is the x-spatial resolution.
Thank you Kate for the reply:
I will be very happy to write to dr. Robertson and to get access to your svn branch, because I'm wondering yet which ice dynamic/thermodynamic model is better or my aims.
Cheers
Arianna