Question for 'ana_mask.h'

[dhtjtks]

Hi all,

I tried to use 'ana_mask.h' to block some area of my domain, which have 100(i)*10(j) grids in interior rho point. When I looked at the results it didn't work like I wanted. Extra areas were blocked when I used parallel mode. However, if I used a single processor, it worked fine but also gave some strange numbers at the masking area. I remember I didn't get this error before when I used other computing system. I had zero values at the masking area before. Is this the compiler problem? I attach my 'ana_mask.h' and output file. Please help me.

Regards,
Lee

[kate]

Maybe it's about your expectations. In serial mode, the size of the arrays is smaller than in parallel mode. It might even be unsafe to be setting Istr-2 in serial mode. In parallel mode you are guaranteed to have 2 or 3 halo points around all the edges on each tile.

As for the 9.9e36, that's in the NetCDF output, using the _FillValue for masked areas. Are those values from the NetCDF output or from print statements in the code?

[dhtjtks]

Maybe it's about your expectations. In serial mode, the size of the arrays is smaller than in parallel mode. It might even be unsafe to be setting Istr-2 in serial mode. In parallel mode you are guaranteed to have 2 or 3 halo points around all the edges on each tile.

As for the 9.9e36, that's in the NetCDF output, using the _FillValue for masked areas. Are those values from the NetCDF output or from print statements in the code?

Thanks Kate. The values from the 'ocean_his.nc'. But I remember the value was "0" for masked areas when I used ROMS3.0. Wasn't it?

[kate]

I'm sure it was, but things change in the ROMS world. 0.0 is a valid temperature so not a very good _FillValue.

[dhtjtks]

I'm sure it was, but things change in the ROMS world. 0.0 is a valid temperature so not a very good _FillValue.

Then, it's o.k. I have to follow the rule.
Thanks Kate!

Lee