Closed agrouaze closed 2 years ago
So we have
s1meta.pixel_xtrack_m
that is self.image['ground_pixel_spacing'][0]
at sensor level (full resolution).
I agree that we should use this value at dataset level. Something like:
s1meta.pixel_xtrack_m * xtrack_step
where xtrack_step
is the resolution, in pixels. (i.e something like np.unique(np.round(np.diff(self._dataset['xtrack'].values), 1))[0]
)
This issue is to spot the fact that 1) currently there is 2 way to fetch the pixel spacing values of the products, and they do not gives the same values: On one side we have the
Sentinel1Dataset
attributepixel_atrack_m
andpixel_xtrack_m
computed like this:And on the other hand we have the information given by
Sentinel1Meta.image
:slant_pixel_spacing
andground_pixel_spacing
. These last variables are coming from the annotations and given at mid sub-swath. For user wanting to have very precise range pixel spacing, there is a variable:range_ground_spacing
. 2) Only thepixel_atrack_m
andpixel_xtrack_m
re available in the attribute of thexsar
datasets, which could be a problem for external libraries; see xsarsea discussion