Releases: inrae/SISPPEO
v1.1.2
v1.1.0
SISPPEO 1.1.0 Release Notes
This release includes new features, bug fixes, and a small performance improvement. We recommend that all users upgrade to this version.
A new version of the documentation (hosted using GitHub pages) is also shipped with it.
New features
SISPPEO can now handle multi-variables algorithms
Algorithms that return multiple arrays (/variables) can now be implemented in SISPPEO. For instance, you can now return :
-
the variable of interest + its uncertainty ;
-
a given variable (e.g., the absorption coefficient of CDOM) at several wavelengths.
This feature will (soon) be used to add an algorithm that estimates Kd [transparency] using QAA (Lee, 2002) for L8 and S2. It will be a multi-var algorithm returning Kd at 443, 490, 550, and 660 nm.
Added a new reader for handling C2RCC (C2X, C2X-COMPLEX) products
C2RCC products (generated with Snap) can now be use as input products.
A new (water quality) algorithm is available
NDCI (Normalized Difference Chlorophyll Index) [chl-a]
New CLI commands and alias
-
commands:
-
algorithms
-> returns the list of algorithms that can be used. -
masks
-> returns the list of masks that can be created. -
create-l3
to create [masked] products
-
-
option alias:
-
-i
for--input_product
-
-t
for--product_type
-
-im
for--input_product_mask
-
-tm
for--product_type_mask
-
-a
for--algo
-
-m
for--mask
-
-o
for--out_product
-
New timelapse
method
Added a method to generate timelapses (as gif files) from TimeSeries objects.
Improvements
-
GRSReader can now read L4, L5, L7 and L8 products.
-
extract_points
:-
can now be use with L3MaskProducts ; the method has been moved from L3Product to CoordinatesMixin.
-
returns NaN instead of 0 in corners (if a rounded buffer is used).
-
-
L3Product.plot
: you can choose the variable to be plotted. -
added SWIR bands specs to both L8 and S2 in sat_config.yaml
-
improved the readibility of the code by using nametuples
-
updated error messages of
xy
andindex
methods.
Performance improvements
Improved both the readibility and the memory efficiency of the code with the use of dataclasses and slots.
Changes
-
got rid of the ObjectFactory class and used dictionnaries instead.
-
loaded TimeSeries ObjectFactory cla (from multiple files) now use a
xarray.DataArray
instead of axarray.DaskArray
. -
extract_points
has a new signature.
Bug fixes
-
filenames of output products
-
use of nanmedian in
TimeSeries.compute_stats
-
image encoding (for empty or small images)
-
batch processing
-
Mutex class (CLI)
Dependencies
pip is not used anymore (for installing pvlib, ray and s2cloudless) in the environment.yaml conda file. Instead, pvlib-python, ray-core, and s2cloudless librairies (availables on conda-forge) are used.
SISPPEO v1.0
This is the first official release of SISPPEO.