Fix depth slicing with non-ocean cells #948
Conversation
|
@maltrud, this is the fix for the issue you uncovered. After I have merged this, I will share my config file and some instructions on how to build your own conda environment for using the |
Previously, when there were cells with `maxLevelCell = 0`, this caused a failure because there was no valid depth coordinate for that cell and therefore no closest depth index could be found. This fix simply skips cells that are not in the ocean and masks them out.
xylar
force-pushed
the
fix_depth_slicing_with_nonocean_cells
branch
from
149f8ea
to
3116204
Compare
We only want to drop times were all values are NaN, not when *any* value is NaN (e.g. below the bathymetry in a vertical average).
TestingI reran the analysis of a run on the I also ran the test suite. All analysis completed successfully and can be seen here: |
Previously, when there were cells with
maxLevelCell = 0, this caused a failure because there was no valid depth coordinate for that cell and therefore no closest depth index could be found.This fix simply skips cells that are not in the ocean and masks them out.
This merge also fixes an issue with rolling averages that was discovered in the same run. Rolling averages should drop
NaNvalues only if all values at a given time point areNaN, not if any are (the default behavior). This was presumably not noticed previously because we performed rolling averages on fields without anyNaNvalues.