205 bottom zeros #210
205 bottom zeros #210
Conversation
…looding. Some of this has been moved from interp_vertical()
…be the same as the old function.
…vertical because this has now been taken out of the interp_vertical function
… bottom layer and gdept being smaller. Max_gdept has decreased because it is not marked one cell shallower, this has increased Sum_mask and decreased Sum_unmask. Mean_temp, Mean_sal, Mean_u and Mean_v have increased because they don't entrain zeros near the bottom.
jdha
left a comment
jdha
left a comment
There was a problem hiding this comment.
I think I get what's happening - just a little concerned that if there's no data from the source grid (which is using 9 grid points around the destination bdy point) that things might get messy - I think the same applies when flood filling down as you could be putting surface waters to depth. I generally horizontally fill before vertically to avoid this last instance as much as possible. But I think you're vertically flood filling before horizontally doing so... this needs more thought.
|
The code is doing the horizontal flood filling on a given chunks so won't be across any land barriers. Do you mean if there is a chunk without source data? I understand what you mean with doing it horizontally first instead of vertically first. I'll swap it. |
…curs in the centre point of 9 first, then in surrounding points. valid_index() has been moved inside flood_fill().
…no the horizontal flood fill happens first.
…me of the 9 surrounding points have data but not the centre so only flood filling the centre can change the weighted averaging.
|
I've updated the branch. |
There was a problem with zeros or FillValues in the bottom layer and cells near land because the flood fill wasn't happening correctly. Shown in issue #205
This has been fixed by making sure land in source data is masked before flood fill downwards. Horizontal flood filling now also occurs.