_{Supertyphoon Champi · 2026-10-21 · MODIS/Aqua True Color · NASA GIBS}

关于 About · 研究与经历 Research · 科研工程 Stack · 工程 Engineering · 组织 Organizations · 工具箱 Toolbox

研究主线为遥感反演方法：从有限的观测信号中，稳健地还原物理量。

海面粗糙度与波谱、平流层的云与气溶胶、行星大气的散射与辐射收支——问题不同，方法同源。海洋与大气的星载观测是这条主线的验证场，行星探测则是它在尺度上的延伸。

做科学问题的同时，也把「观测 → 反演 → 验证 → 产品」这条路当成一项工程来做：数据要能重现，方法要能被别人跑起来，图要能直接进稿。

My research centers on remote-sensing retrieval: inferring physical variables, robustly, from limited observations.

Sea-surface roughness and wave spectra, stratospheric cloud and aerosol fields, planetary atmospheric scattering and radiation budgets: distinct problems, a shared methodology. Satellite observations of the ocean and atmosphere serve as its proving ground; planetary exploration carries it across scales.

The science is half the job; the other half is engineering the chain observation → retrieval → validation → product so that results are reproducible, runnable by others, and figure-ready.

• 机场终端区雷达数据处理（在研）—— 面向民航气象保障的雷达基数据工程：扫描式相干多普勒测风激光雷达、风廓线雷达等体制的数据质量控制、径向速度与低空风场反演、回波结构的批量产品化，目标是把原始观测变成可直接进入临近预警流程的格点场。
  _{Airport terminal-area radar data engineering (in progress) — quality control, radial-velocity and low-level wind retrieval, and batch product generation across scanning coherent Doppler wind lidars, wind profilers and related systems, turning raw observations into nowcasting-ready gridded fields.}
• 星载单光子激光雷达海面风速反演（在研）—— 基于 ICESat-2 / ATL03 光子点云：去噪与海面信号提取、波形参数反演，联合海面均方斜率与波谱特征建立风速模型，并与 GNSS-R（CYGNSS）产品交叉验证。
  _{Spaceborne single-photon lidar retrieval of sea-surface wind speed (in progress) — built on ICESat-2 / ATL03 photon point clouds: denoising and sea-surface signal extraction, waveform-parameter inversion, and a wind-speed model driven by mean square slope and wave-spectrum features, cross-validated against GNSS-R (CYGNSS) products.}
• 北京大学地球科学夏令营（2024）· 地球系统科学方向
  _{Peking University Earth Science Summer Camp (2024) — Earth System Science}
• CESO 2024 · 金奖 · 全国第 41
  _{CESO 2024 (Chinese Earth Science Olympiad) — Gold Medal, 41st nationally}

从「拿到一份数据」到「交出一张能发表的图」，中间那几十步手动操作，都应该被自动化。

数据源矩阵 / Data feeds — ERA5 · ECMWF IFS / AIFS · ECMWF WAM · Sentinel-2 (STAC) · Argo · GEBCO · HYCOM · GHCND

• DeepSeek Harness 插件开发 —— Host / Client 双侧插件，TypeScript 与 Cordis 依赖注入，自定义 UI Slot 与 patcher 层加载。
  _{DeepSeek Harness plugin development — host- and client-side plugins in TypeScript, with Cordis dependency injection, custom UI slots and patcher-layer loading.}
• 遥感数据处理流水线 —— 光子级点云处理、沿轨剖面重构与波数谱分析，以全流程可复现、结果可验证为工程准则。
  _{Remote-sensing data pipeline — photon-level point-cloud processing, along-track profile reconstruction and wavenumber spectral analysis, engineered for reproducibility and verifiability.}
• 追风工具（自用） —— FastAPI 后端 + Leaflet 前端，全国雷达拼图的透明叠加与风暴单体标注，用来看华南的强对流。
  _{Storm-chasing toolkit (personal use) — FastAPI backend with a Leaflet front end for transparent radar-mosaic overlays and cell annotation.}

Supercell Hunter 粤气爱追风小队 成立于 2025 年，追逐强对流、台风与锋面， 记录并传播华南的激烈天气。 Founded in 2025 · chasing severe convection, typhoons and fronts across Guangdong.

_{© 2026 碳基12 · Banner: NASA GIBS / MODIS-Aqua · Made with care}