Repository navigation
Conversation
| raise ValueError("The dimensions of the arrays are not consistent") | ||
|
|
||
| # Interpolate using the xarray interp function twice: first across, then along the scan | ||
| # Interpolate using the xarray interp function twice: first along, then across the scan |
There was a problem hiding this comment.
@sjoro @pepephillips @ameraner I want to confirm this change with you. The first interp call below uses the "dim_alt" dimension and is along track, the second call uses "dim_act" so that is across track. Right?
There was a problem hiding this comment.
correct yes, thanks for spotting
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #134 +/- ##
==========================================
+ Coverage 89.74% 90.15% +0.41%
==========================================
Files 20 20
Lines 1541 1565 +24
==========================================
+ Hits 1383 1411 +28
+ Misses 158 154 -4
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
|
can you explain where the problem originates from? is the data we interpolate in F-ordering to start from? |
|
For the record, the mentioned fixes on pyresample side are pytroll/pyresample#723 fixed by pytroll/pyresample#725 |
@mraspaud It comes from the scipy interp and the order that it is done in. At first in one of my earlier commits Claude had swapped it as it would fix the memory order but it makes the chunking be column based instead of row based and messes up the chunking relative to the non-interpolated datasets in the Satpy reader. See #135 for the huge rewrite version of this PR which fixes all of that at the cost of complexity. |
3 participants
I had a case where a user of my code was using an old version of pyresample and didn't have the C-contiguous fixes that live in pyresample now (maybe just the
mainbranch). This PR adds a C-contiguous reorder to the VII interpolator so that it doesn't have to happen else where in Satpy processing. This has the downside of data being copied an extra time when it would have been copied during a later operation anyway. It has the upside of not needing a copy later when it is needed.Even if we decide (@ameraner, @sjoro, @pepephillips) to not merge the C-contiguous part of this I would like to commit the test changes because they are pretty helpful.
git diff origin/main **/*py | flake8 --diff