gee-growth-diary

# GEE Growth Diary Use this skill as a source-aware companion for the GEEer成长日记 WeChat album. It is not the canonical EasyGEE method router. ## Trust Boundary 1. Official Earth Engine docs and catalog pages win. 2. The `easygee` skill and its references win for current workflow shape, dataset QA, exports, quotas, authentication, and Map Console behavior. 3. This skill is useful for finding article-inspired task ideas, Chinese phrasing, historical examples, and source attribution. 4. Article code and dataset ids must be re-verified before reuse. Many GEE tutorials age quickly because catalog versions, QA bands, and best practices change. 5. Do not reproduce source article text. The skill contains distilled guidance and a source index only. ## When To Use - The user asks about GEEer成长日记, this WeChat album, or "那 153 篇文章". - The user wants a Chinese GEE cookbook-style brainstorm, and EasyGEE's canonical references do not already answer the method choice. - You need article titles, links, or rough topic tags for attribution. - You want to compare an article-inspired method against current official docs. For routine GEE work such as NDVI, water, classification, time series, export, Map Console layers, auth, quota, or dataset search, start with `easygee`. ## Resource Map - Read `references/source-index.md` for article titles, links, dates, and heuristic topic tags. Tags are retrieval aids, not reviewed classifications. - Read `references/workflow-patterns.md` for article-observed workflow motifs and known age-related risks. - Read `references/dataset-method-router.md` only as secondary candidate notes; verify candidates through EasyGEE dataset search and official catalog docs. ## Review Checklist Before turning any article-derived idea into EasyGEE code or documentation: 1. Verify the dataset id in the official Earth Engine catalog or current community catalog docs. 2. Replace outdated collections with current versions when appropriate, for example Landsat Collection 2 and MODIS/061. 3. Re-check scale factors, units, QA bands, masks, and date coverage. 4. Decide whether the method is product-backed, derived remote sensing, supervised/unsupervised classification, thresholding, or current-image recognition. 5. Add a visual probe and a small numeric probe before export. 6. Label any unverified article-derived step as provisional.