Fix erroneous masks for Coreg classes #197
Conversation
- warp_dem() now has mask dilation by default. - get_array_and_mask() takes and properly respects Rasters now (it was used with rasters before, but the mask was not respected). - -9999 values in warp_dem were replaced by np.nan - Added a warning catcher in apply_matrix() (due to skimage + numpy) - Improved test to validate that the above changes fixes the problem.
tests/test_coreg.py
Outdated
|
|
||
| # Validate that the zshift is not something crazy high and that no negative values exist in the data. | ||
| assert np.nanmax(np.abs(zshift)) < 50 | ||
| assert np.count_nonzero(aligned.data.filled(0) < -50) == 0 |
There was a problem hiding this comment.
I've seen you use 'filled' quite often. I believe your intention is to exclude nodata from the statistics. If so, I recently found out you can use data.compressed(), which returns only the valid values as a 1D array (I know, the name is not very intuitive...).
There was a problem hiding this comment.
A better test would be to check that the elevation range before/after coregistration is more or less identical.
There was a problem hiding this comment.
Changed to use compressed() in 2d13111!
I also added tests to check that coregistration improved. I see now that I misread your suggestion haha, but this should be more than apt, no?
There was a problem hiding this comment.
Well, the overall statistics might not be too sensitive to just a few outliers. But the range will be, so I would still recommend checking the range.
But it's a good idea to check that the dispersion is reduced too!
|
Thanks for fixing the bugs! |
Fixes #195 .
Changes:
warp_dem()now has mask dilation by default.get_array_and_mask()takes and properly respects Rasters now (it was used with rasters before, but the mask was not respected).warp_dem()were replaced by np.nanapply_matrix()(due to skimage + numpy deprecation warning)Something that is not tested automatically but was tested manually was the example by @adehecq in #195 in
examples/plot_blockwise_coreg.py. As seen in the screenshot in #195, these fixes solve that problem.