Published June 5, 2026
| Version v0.6.0
Software
Open
sdo-clv-pipeline
Authors/Creators
- 1. Flatiron Insitute
- 2. Center for Astrophysics | Harvard & Smithsonian
- 3. Astrophysics Group, University of Exeter
Description
v0.6.0
Performance, correctness, and robustness release. Per-epoch processing is roughly 2× faster, two unit bugs in the velocity fit are fixed, and batch runs are more robust and resumable.
Highlights
- ~2× faster per epoch — the slow per-pixel sunpy/astropy coordinate transforms are replaced with an equivalent analytic, numba-accelerated path, validated against the sunpy implementation.
- Fixed two unit (degree/radian) bugs in the Doppler bulk-velocity fit
Performance
- Analytic helioprojective → heliographic/heliocentric geometry (replaces the
per-pixel
SkyCoordframe transforms). - Coarse-grid AIA → HMI reprojection.
- numba-accelerated geometry, Legendre velocity-fit design matrix, and bilinear resampling.
- Vectorized region-statistics aggregation.
- Opt-in within-epoch multithreading via the
SDO_THREADSenvironment variable (default 1, which is the safe setting under the across-epoch process pool).
Robustness & tooling
- Clearer end-of-run summaries and explicit per-epoch skip reasons; QUALITY-flag decoding with tolerated "soft" flags recorded in the output.
- New
scripts/run_one.py(idempotent, resumable single-epoch processing), andscripts/stitch_tmp.pyfor robust, resumable, failure-isolated batch runs.
Plotting
- Image and mask plots now stamp the observation timestamp (formatted from
DATE-OBS) in the upper-left corner, clear of the solar disk.
Full Changelog: https://github.com/palumbom/sdo-clv-pipeline/compare/v0.5.3...v0.6.0
Notes
Files
palumbom/sdo-clv-pipeline-v0.6.0.zip
Files
(3.3 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:2d0292d09bde8146d208639ba6aa8c02
|
3.3 MB | Preview Download |
Additional details
Related works
- Is supplement to
- Software: https://github.com/palumbom/sdo-clv-pipeline/tree/v0.6.0 (URL)
Software
- Repository URL
- https://github.com/palumbom/sdo-clv-pipeline