AGENT SKILL

seismic-picker-selection

Choose an appropriate phase picking strategy (deep learning vs STA/LTA vs template matching) and tune it to balance recall/precision for catalog-quality picks.

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

01 前言 Front Matter

name
seismic-picker-selection
description
Choose an appropriate phase picking strategy (deep learning vs STA/LTA vs template matching) and tune it to balance recall/precision for catalog-quality picks.

03 SKILL.md 正文

Steps

  1. Prefer deep learning pickers (e.g., PhaseNet via SeisBench) when you need high sensitivity/generalizability across many traces without templates.
  2. Use thresholding and peak-distance rules to control false positives; do not assume exactly one event or one P/S pair in a continuous window.
  3. If deep learning outputs are unstable due to data scale, rescale/normalize waveforms before inference (especially when raw amplitudes are extremely small).
  4. Consider simple signal-based checks (e.g., STA/LTA-style gating) only as a supplement for rejecting obvious noise segments, not as the primary picker if high timing precision is required.

Expected Result

A method choice and tuning approach that targets high F1 by improving timing robustness and reducing forced false positives on low-SNR or phase-absent traces.

04 同仓库更多 Skills