load_data: add mintpy.multilook.method for mean/median
#797
Conversation
|
Looks like there are lots of conflicts with #796. I will try to fix them. |
+ mintpy.multilook. suggested from insarlab#797 (comment) + bugfix of size in creating ionosphere dataset in stackDict.py
|
Hi @yunjunz, this PR is updated regarding the template names and the dataset size bug we discussed offline. Take a look when you have some time. Thanks! The Ci/Circleci test failed just because |
Comparing the two circle CI runs (from the latest main branch and this PR) shows that the loaded ifgramStack.h5 is incorrectly resized from 390 x 390 to 780 x 781, thus, causing the pre-staged ERA5.h5 not usable. I will take a deeper look later today. |
yunjunz
changed the title
mintpy.multilook.method for mean/median
+ mintpy.multilook. suggested from insarlab#797 (comment) + bugfix of size in creating ionosphere dataset in stackDict.py
Allow different options of multilook interpolation while loading data by parsing a method variable + 'nearest' method (default) + 'mean' method + 'median' method Propagate the method variable from the template file named as mintpy.load.stepMethod Merge and resolve conflicts with insarlab#796
+ mintpy.multilook. suggested from insarlab#797 (comment) + bugfix of size in creating ionosphere dataset in stackDict.py
to be more consistent with the order of execution in the code
+ utils.attribute.py: add update_attribute4resize() + prep_isce.prepare_stack(): call update_attribute4resize() to simplify the code + objects.stackDict.ifgramStackDict.write2hdf5(): update metadata if resizing is applied (to ionStack.h5).
There was a problem hiding this comment.
This PR looks all good to me now. @yuankailiu please let me know if your geocoding issue from our offline discussion is resolved or not with the latest commits. Then we could go ahead and merge this PR.
Great, thank you! Now the metadata is correct and the geocoding of the |
Description of proposed changes
Allow different options of multilook interpolation while loading data by parsing a method variable
This applies to multilook the
unwrapPhase,coherence,wrapPhase. But not theconnectedComponent.Propagate the method variable from the template file named as
mintpy.load.stepMethodBy setting the multilook while loading data to mean or median, the noise in the loaded datasets is reduced. So this can be used for a smoother visualization of interferograms and the resulting time series, and the velocity.
Use with care. Because using mean or median multilook on the unwrapped phase can smoothen the integer of$2\pi$ phase jumps (due to unwrapping errors). This may result in issues in unwrapping errors identification and correction. Average multilooking will also introduce a lack of phase closure over an interferometric circuit. So we better still use 'nearest' as the default setting.
Reminders