研究笔记 · Research Note
Walker Circulation Dynamics @ PKU
Walker Circulation Dynamics @ PKU
基本信息
- 时间: 2020.11 — 2021.10 (~12 months active research; paper draft v6 finalized 2021-10-22)
- 角色: 本科独立研究 (First author)
- 组织: 北京大学物理学院大气与海洋科学系
- 指导教师: Ji-Nie-聂绩 (primary PI) + (co-advisor, 胡组讨论平台)
- 地点: 北京 / Tianhe-II 国家超算广州中心 (NSCC-GZ)
- 资助: PKU President Fellowship (undergraduate research)
核心问题
Walker Circulation 对 oceanic forcing 宽度 的 systematic 响应: 在 CAM3 aqua-planet 里把海洋热源 (Warm Pool) 与冷源 (Cold Pool) 的距离 D 从 60° 变化到 180° (step 20°), 加上 Warm-only / Cold-only / Aqua baseline, 共 10 cases × 30 年 runs, 探索 Walker 强度 + 大尺度 SST/precip/ω/SW 响应的规律。
🌟 Zhenyu 核心原创贡献 (2021-07-13 breakthrough)
Linear Superposition framework — 用 forcing-response operator 的 2nd-order Taylor expansion 在 forcing 距离较远时 drop interaction term, 得到:
4-field quantitative validation (D80-D180 bulk):
- SST corr 0.9063
- Precip corr 0.9215
- ω@500mb corr 0.8282
- SW corr 0.9324 (all > 0.82)
2021-07-13 breakthrough 现场 verbatim (见 slide 24):
After advisor discussion, the linear superposition framework proposed by Zhenyu provided a satisfactory explanation, and was endorsed.
Emergent prediction: Walker strength 非单调 peaks at D80-100° — 这个 peak 恰好对应 Wonly 的 natural descending area (200-220°E) 与 cold pool 几何重合的位置。不需要 nonlinear mechanism 解释。
Linear superposition 失效的边界 (D60-D80): low-cloud positive feedback loop —— CP descending↑ → suppress convection → trap humidity → low cloud↑ → albedo↑ → SW absorption↓ → SST↓ → ΔSST↑ → loop; 4-term energy budget diagnosed (SW+LH dominate, ΔSW D60: -66.5141 W/m² → D180: -110.6316 W/m²)。
7-Month Intuition Accumulation Trajectory (从 framing 到 breakthrough)
| 时间 | 事件 | 累积状态 |
|---|---|---|
| 2020-12-29 slide 5 | Walker 左右非对称 framing (7-width sweep) | 早期 framework seed |
| 2021-04-26 master PPT 295 slides | 拼贴萌芽 slides 32-65 (3 个月 BEFORE breakthrough) | first systematic superposition attempt |
| 2021-05-16 胡组 15 slides | D180 self-consistent linear superposition explicit | 3 个月前 explicit |
| 2021-05-26 / 5-28 / 6-2 进展报告 | 84 figure groups systematic (5-28 拼贴系统化到 6 fields) | 2 个月前接近成形 |
| 2021-07-13 进展报告 29 slides | ⭐ breakthrough — Zhenyu 当场提出 linear superposition + Ji Nie endorsement | ”之后就可以开始写论文” |
| 2021-07-16 胡组 28 slides | Post-breakthrough 给 Hu group 的 echo (slide 20 “线性叠加可以解释 Q1”) | multi-advisor platform delivery |
关键成就
- Linear Superposition framework 提出 + 4-field 0.83-0.93 correlation 验证 (framework-level 贡献, 非 incremental), 详见 Linear-Superposition-Framework
- Emergent non-monotonic Walker strength 解释 (D80-100° peak = natural descending 与 cold pool 重合)
- Low-cloud feedback failure-boundary 诊断 (4-term energy budget SW+LH dominate)
- ω-based vs u-based methodology bug catch (4-26 slide 26 发现 omega 不满足 “无散”; 两 NCL module 并列 preserved:
msf_top2bottom_u2_Walker.ncl61 lines +msf_top2bottom_omega2_WalkerStream.ncl63 lines) — engineering rigor signature, 归于 Multi-Method-Comparison-Spine - Paper draft v6 (3663 words, 11 refs, 2021-10-22): Zhenyu He¹, Ji Nie¹*, supported by the PKU President Fellowship — 未投稿:
Project not submitted by mutual decision.
非 research failure, 是合作者的 publishing decision。paper 草稿存档见 。 6. 8 周 / 9-version paper auto-diff: v1 → v2_2 → v2_2-nj (Ji Nie 第 1 轮 +13 中文 inline comments) → v3 (Zhenyu 实施 +3 equation tables) → v4 (+465 words / +8 refs in 2 days, 大扩充) → v5 → v5-nj (Ji Nie 第 2 轮) → v6 (+3 refs, 最终) 7. Ji Nie methodology arc arc signature: Project paper draft retained internally; advisor and student arrived at a mutual decision not to submit.
技术深度 (Resume-grade specifics)
- 5-stage HPC pipeline architecture: (1) CAM3 aqua-planet with prescribed Q-flux → (2) 30-year spin-up on Tianhe-II (undergraduate Tianhe-II account) account (
yhrun -p work -n 4848-PE SLURM) → (3) 25cal_X/MATLAB diagnostic folders → (4) NCL stream function + ω analysis → (5) paper figure refinement MATLAB + Python - 27,882 MATLAB+NCL lines / 5 languages: MATLAB (main diagnostic) + NCL (stream function) + Fortran (namelist + CAM3 source mods) + Bash (SLURM runscripts) + Python (paper figure polish)
- 14 Q-flux variants preserved (D60 / D80 / D100 / D120 / D140 / D160 / D180 + Wonly + Conly + Aqua + variations)
- 20 NCAR CAM3 .F90 source mods (
SourceMods.zip143 KB in Walkercam3_task/), aggregate: cldinti / cldsav / cloud_fraction / comctl / ghg_defaults / hb_diff / history / hycoef / ice_constants / inidat / initext / ocn_srf / physpkg / radiation / radlw / runtime_opts / somoce (MLDANN 50→30m slab ocean, key dep) / stratiform / tphysbc — 本科生 source-code 级别的 GCM 修改
学术 / 职业 意义
- 与 PKU-Aerosol-Senior-Thesis + Caltech-DSCOVR-Research 共同构成 3-track parallel Sep-Dec 2021 本科大四上 (Walker paper v3→v6 + DSCOVR HBP + nwp_hw 4 个作业 ≈ 3-4 deliverable/week)
- 是 PKU-Walker-Hadley-Response-Research (2019.10 — 2020.05, Ji Nie 组 first project) 的 successor + 是 PhD 阶段 3D LES + 1-D DSE+SWTG quasi-2D 等 multi-model 工作的 competence foundation
- Walker paper v6 的 research-grade specificity 被 Zhenyu 后续用于 UC Berkeley PhD 申请 + Caltech/UCR Yuk Yung/King-Fai Li 合作 + Princeton SGS 邀请的底色
- Moment 2 of Independent Judgment narrative (held until 2026-06-10) spine (2021-07-13 在导师 MSE budget 解释失败后, 自己提出 linear superposition 并 defend 到 Ji Nie endorsement)
使用的技能
- 气候物理与大气科学 (Walker/Hadley + 4-term energy budget + low-cloud feedback)
- 高性能计算-HPC (Tianhe-II (undergraduate Tianhe-II account) 48-PE + CAM3 aqua-planet 30-yr)
- Python-科学计算 (paper figure + 部分 diagnostic)
- 数值模式源代码修改 (20 NCAR CAM3 .F90 mods)
相关页面
- PKU-Undergraduate-Research — 本科研究 umbrella page
- PKU-Walker-Hadley-Response-Research — 前序 Ji Nie 组 project (2019.10 — 2020.05)
- Ji-Nie-聂绩 — Walker 项目 primary PI
- — Walker co-advisor, 胡组讨论平台
- Linear-Superposition-Framework — Zhenyu 原创 framework concept 页
- — breakthrough artifact source page
- — paper draft source page
- PKU advisor methodology companion (private) — acknowledgment 改写 arc, 与 Jing-Li-李婧 共同构成
- Multi-Method-Comparison-Spine — ω-vs-u methodology rigor 归入此 spine
- Independent Judgment narrative (held until 2026-06-10) — Moment 2 (meta-level)
- Academic-Honors-PKU — supported by the PKU President Fellowship
来源
research_wiki/projects/walker_hadley/overview.md(~55k words / 1303 行)research_wiki/resume_angles/problem_solving.md§Project 5research_wiki/resume_angles/technical_depth.md§Section 9