Releases: umr-lops/xsar
v0.9.8
v0.9.7
v0.9.6
v0.9.5
v0.9.4
v0.9.3
v0.9.2
- add capability to get geolocation fields at specific image coordinates without loading full HR variable
- add
lazy_loading
option to turn "on" of "off" themap_block_coords()
mechanism that currently suffer memory leak - add
load_luts
option to turn "on" of "off" the decoding and rasterization of LUTs. - prevent the usage of resolution when reading a IW or EW SLC product (this feature is out of scope for
xsar
)
v0.9.1
v0.9 xsar with datatree
What's Changed
This new 2022 release will provide the users a nice overview of the mapped SAR Sentinel-1 product into xarray-datatree
to support groups/nodes of variables.
The original xarray.Dataset
containing the high resolution variables is still available for users but extra xarray.Dataset
from .xml annotations are now provided to the users. These annotations are :
- Doppler estimates,
- bursts information,
- image information,
- low resolution geo-location,
- FM rate information,
- ...etc...
Rationales
The main motivations for this release are:
- making
xsar
outputs more self discoverable - starting to separate as much as possible the decoding of the SAR product and the "interpretation layer" (i.e. interpolation, calibration denoising). In particular this release will enable expert SAR users to perform their own operations on the data using directly the annotations.
- keeping the traceability of each variable in
xsar
using:- the attributes
origin
(xpath), - the .xsd files to have official ESA definition of variables
- keeping the exact same names for the variables
- the attributes
Disclaimer
For now the datatree v0.0.9 is not supporting the double mutability, it means that only the changes in the datatree
will affect the dataset
s but opposite way is not yet supported (see: xarray-contrib/datatree#80 ).
It means that modifications done on thedataset
s should be assign
explicitely with such syntaxe:
datatree['measurement'] = datatree['measurement'].assign(dataset)
Contributors
@agrouaze , thanks to @keewis for his hints and to @lanougue for his feedbacks