Necessary grid components.

[hetland]

I am working more on using Sakov's gridgen utility for ROMS grids. The output of this utility only returns 'wet' cells, that is, there is no information about the grid at all in some locations. Generally, this is not a problem -- these points would be masked anyways. However, it leads me to the question of:

Are there any masked grid properties that are used? I know that masked values of h are used to get the depth of the cells at the coast. More difficult to deal with is pm and pn. Are masked values of pm and pn sometimes used? What is the appropriate value to set for cells that are masked, and not defined by the grid?

[kate]

I don't think it matters what they are as long as they are non-zero. I would guess that the big three that shouldn't be zero are h, pm, pn. From there the thing to do is try it and tell us

[hetland]

Well, I already know that it ..works.. if you fill in pm and pn with non-zeros. I don't know if the solution is sensitive to the fill value. I suspect that it can really only come in through the viscous/diffusive terms, if at all.

Rather than try the empirical approach, and see what happens with different fill values, I might actually try to >gasp< read the code...

-r