Analytical passive tracers on the center of a model domain

[d.kobashi]

Dear all,

I am trying to run a simple analytical model (a rectangular-shape domain with uniform depth). I put M2 tidal elevation and current along the eastern boundary and radiation OBC along western boundary so that tides are radiated out. The northern and southern boundaries are closed. The grid size is 100 x 50. Temperature and salinity are constant throughout the domain.

I want to to put passive tracers (dyes) for a square-area on the center of the domain (11x11 grids) and the following are what I did.

1. I defined T_PASSIVE and ANA_PASSIVE in cppdefs file

Code: Select all

#defined T_PASSIVE
#defined ANA_PASSIVE

2. In ocean.in, I changed NPT=0 to NPT=1 (one dye area) and Hout(inert) == T

Code: Select all

NPT=1
Hout(inert) == T

3. I modified ana_passive.h as follow.

Code: Select all

      DO ip=1,NPT
        itrc=inert(ip)
        DO k=1,N(ng)
          DO j=20,30
            DO i=40,50
              t(i,j,k,1,itrc)=20.0_r8
!              t(i,j,k,1,1)=20.0_r8
!              t(i,j,k,2,itrc)=t(i,j,k,1,itrc)
            END DO
          END DO
        END DO
      END DO

Then when I checked the model output, I saw 6 tracer areas (and 3 more near the southern boundary) not 1 (see attached figure). I wonder what I did wrong.

My cppdef file is also attached. I use COAWST, but the cppdef is pretty much the same as that for a recent Rutgers ROMS.
Any comments would be appreciated.

Thanks in advance.

DJ Kobashi
TAMU

[jcwarner]

did you tile it 3x2?
try tile it 1x1 and use 1 processor. the way you specified the point sources are not correct for a tiled application.

[kate]

Indeed, you need something more like this:

Code: Select all

          DO j=MAX(JstrT,20),MIN(JendT,30)
            DO i=MAX(IstrT,40),MIN(IendT,50)

[d.kobashi]

Thanks John and Kate for input. Always helpful.

tile!? my bad. Why didn't I think of that...

Yes. the single core run worked and so did the modification that Kate mentioned for multiple core run. Thanks a lot.

When I re-ran my analytical model, I saw abnormally high concentration at the bottom right corner of the tracer area as shown in the attached figure (for both horizontal snapshot and cross-section for xi direction). The initial dye concentration is 20 kg/m3 and the concentration of the area is around 60 kg/m3, which does not make sense to me. I used TS_MPDATA for this run, but the 3rd order upstream scheme (TS_U3HADVECTION) gave a similar result. I am puzzled by this. I wonder why.

Any comments would be appreciated.

Thanks in advance.

DJ

[wilkin]

It may be significant that you commented out the initialization of the second time level.

Code: Select all

!              t(i,j,k,2,itrc)=t(i,j,k,1,itrc)

Try restoring that step.