Expected speed for SAR calibration
SimJeg opened this issue · comments
Hello !
First thank you for developing and maintaining this snapista package. To install the package I had to first run conda install -c terradue -c conda-forge snap=8.0.0
, it may be useful to precise it in the README.
I am applying the operations listed below on Sentinel-1 data (~1.5GB) and it took ~10mins on a machine with 8 CPUs. Is it what I sould expect ?
- Manifest name: S1A_IW_GRDH_1SDV_20220113T015151_20220113T015216_041436_04ED43_BAD7.SAFE
- Operations
- Read
- Apply-Orbit-File"
- Remove-GRD-Border-Noise
- Calibration
- Write
I am currently adding the Terrain-Flattening and Terrain-Correction to see how long it takes
Thanks
There may be a problem on my side as running the following code takes 6 minutes :
import os
from snapista import Graph, Operator
path_to_manifest = '/data/sentinel1/S1A_IW_GRDH_1SDV_20220113T015151_20220113T015216_041436_04ED43_BAD7.SAFE'
g = Graph()
g.add_node(
operator=Operator(
"Read",
formatName="SENTINEL-1",
file=path_to_manifest,
),
node_id="read",
)
g.add_node(
operator=Operator("Write", file='output.tif',
formatName="GeoTIFF-BigTIFF"),
node_id="write",
source="read",
)
g.run()
Hello @SimJeg,
You can customize the performance of gpt
, see https://github.com/snap-contrib/snapista/blob/master/src/snapista/graph.py#L378