gridpak: BIG in rect must be at least?

[Lee2015]

Hi,

When I run the Carribean example using gridpak, I got the following information. How to fix it?

~/roms/SCS$ ./grid < Caribbean/grid.in
BIG in rect must be at least
ierr = 8192

Many thanks!

[kate]

In rect.F, there is:

Code: Select all

      integer    BIG
      parameter (BIG=8000)

The check on it is giving you the message from:

Code: Select all

!  Check on work array size
      if (N .gt. BIG) then
         call crash('BIG in rect must be at least ', N)
      endif

The error is telling you to change BIG here to make it larger (8192 or more).

[Lee2015]

kate,

Thanks for your reply!

There is no rect.F file in gridpak folder. In Drivers folder, there is a grid.F file, in which there is

integer BIG
parameter (BIG=3000)

I made the BIG larger than 8192, but the error still exist.

[kate]

Did you try "make clean" and "make" before trying again? There's some problem with the dependencies so you need a "make clean" after changes.

[Lee2015]

Thanks! It works! But when I try to run USwest and Caribbean examples, I got different errors.

~/roms/SCS$ ./grid < USwest/grid.in
rectangularity error in mapped contour at iteration 1 is 5.4265E-02
rectangularity error in mapped contour at iteration 2 is 4.1462E-03
rectangularity error in mapped contour at iteration 3 is 1.9877E-03
rectangularity error in mapped contour at iteration 4 is 9.0973E-04
rectangularity error in mapped contour at iteration 5 is 3.8280E-04
rectangularity error in mapped contour at iteration 6 is 1.5849E-04
rectangularity error in mapped contour at iteration 7 is 6.5328E-05
rectangularity error in mapped contour at iteration 8 is 2.6908E-05
sepeli failed while computing x solution
ierr = 11
~/roms/SCS$ ./grid < Caribbean/grid.in
rect: tracking error

[kate]

If you look at Caribbean/grid.h, it has:

Code: Select all

      parameter (     ITMAX=8, IBIG=400    )

You need to compile against that file.

The ITMAX story is that the core routine of "rect" needs to iterate until convergence, in this case 8 times. If you call it too few times, the solution is not as orthogonal as it could be. The last time I looked at Seagrid, it called rect exactly once.

[Lee2015]

Thanks kate!

It is the grid.h file in Include folder, not grid.h file in Caribbean folder. But I got the same error to both ITMAX > 8 and < 8.

:~/roms/SCS$ ./grid < USwest/grid.in
rectangularity error in mapped contour at iteration 1 is 5.4265E-02
rectangularity error in mapped contour at iteration 2 is 4.1462E-03
rectangularity error in mapped contour at iteration 3 is 1.9877E-03
rectangularity error in mapped contour at iteration 4 is 9.0973E-04
rectangularity error in mapped contour at iteration 5 is 3.8280E-04
rectangularity error in mapped contour at iteration 6 is 1.5849E-04
rectangularity error in mapped contour at iteration 7 is 6.5328E-05
rectangularity error in mapped contour at iteration 8 is 2.6908E-05
rectangularity error in mapped contour at iteration 9 is 1.1082E-05
rectangularity error in mapped contour at iteration 10 is 4.5637E-06
rectangularity error in mapped contour at iteration 11 is 1.8795E-06
sepeli failed while computing x solution
ierr = 11

[kate]

But you show an ITMAX of 11, not 8. Oops, I see the failure is in sepeli, not rect. I don't remember what to do about that one.

[Lee2015]

Whether ITMAX > 8, = 8 or < 8, I got errors.

[kate]

The Caribbean case is rectangular and therefore grid won't be happy. Instead, use sqgrid:

Code: Select all

sqgrid < Caribbean/grid.in

Oh, I see you are trying USWEST now.

[Lee2015]

kate,

Thanks for your help!!! Both Caribbean and USwest work! The root cause of the problem is that the files in Caribbean or USwest must be copied into the Include folder, not the BIG or ITMAX.

~/roms/SCS$ ./grid < USwest/grid.in
rectangularity error in mapped contour at iteration 1 is 7.7235E-02
rectangularity error in mapped contour at iteration 2 is 5.6629E-03
rectangularity error in mapped contour at iteration 3 is 2.5965E-03
rectangularity error in mapped contour at iteration 4 is 1.1331E-03
rectangularity error in mapped contour at iteration 5 is 4.6255E-04
rectangularity error in mapped contour at iteration 6 is 1.8686E-04
rectangularity error in mapped contour at iteration 7 is 7.5318E-05
rectangularity error in mapped contour at iteration 8 is 3.0351E-05
area 0.1162E+13

Activated C-preprocessing Options:

DCOMPLEX Double complex
DBLEPREC Double precision
DRAW_COASTS Draw the coastlines on some plots
KEEP_SHALLOW Keep shallow areas shallow
PLOTS Write out NCAR graphics plots
minimum dx = 5257.24316
minimum dy = 4230.10059


~/roms/SCS$ ./sqgrid < Caribbean/grid.in
area 0.1054E+14

Activated C-preprocessing Options:

DCOMPLEX Double complex
DBLEPREC Double precision
ELLIPSOID Distances computed with ellipsoidal terms
ETOPO2 ETOPO2 bathymetry
DRAW_COASTS Draw the coastlines on some plots
PLOTS Write out NCAR graphics plots
minimum dx = 4401.2580729161855
minimum dy = 2292.8465624996461

Lee

[kate]

Whew! It's hard for me to fix these things while I'm traveling. There's not been much call for me to clean that stuff up - maybe you should look into that GridBuilder thing instead (not that I can run windows stuff).