AGENT SKILL

light-curve-preprocessing

Clean and detrend a TESS-like light curve (quality filtering, outlier removal, flattening) while preserving transit signals.

License 许可
Apache-2.0
最近更新
2026-11-14
Stars
★ 0
Forks
1

01 前言 Front Matter

name
light-curve-preprocessing
description
Clean and detrend a TESS-like light curve (quality filtering, outlier removal, flattening) while preserving transit signals.

03 SKILL.md 正文

Steps

  1. Load the columns: time (MJD), normalized flux, quality flag, flux uncertainty.
  2. Apply quality filtering first (commonly quality == 0 for good data); drop NaNs/infs.
  3. Remove gross outliers (sigma clipping); use a conservative threshold if transits are shallow (e.g., sigma≈5–10).
  4. Detrend stellar/instrumental variability with lightkurve.LightCurve.flatten(window_length=...):
    • Choose window_length long compared to transit duration but short enough to remove rotation/spot modulation.
  5. Optionally run a second outlier removal pass on the detrended light curve (often slightly tighter than the first).
  6. Inspect before/after plots to ensure transits are not over-smoothed or removed.

Expected Result

A cleaned, detrended light curve where stellar activity is largely removed and transit-like dips are preserved for TLS/BLS searches.

04 同仓库更多 Skills