Archive planning files for issue #58

Author: NewGraphEnvironment

planning/archive/2026-05-issue-58-temperature-lit-review/.gitkeep

@@ -0,0 +1,105 @@
+# Archive: Temperature-departure methodology lit review (#58)
+
+## Outcome
+
+Established the citation backbone for the **temperature-departure**
+sections of the regional vignettes — Issue 1 of the climate-departure
+3-split lit reviews (precip+drying = #61, interpretation framing
+forthcoming). 10 peer-reviewed papers added to the
+`NewGraphEnvironment/climate` Zotero collection (key `8MH9LCC9`),
+local ragnar DuckDB at `data/rag/temp_methodology.duckdb` (gitignored,
+677 chunks, 10 sources), and a synthesis-plus-citation-map
+`findings.md` ready for the downstream vignette branch to consume
+selectively per the plain-language philosophy
+(`feedback_vignette_citations_sparse.md`). Released as **v0.2.2**
+(patch).
+
+## Headline findings
+
+1. **`cd_trend()`'s raw Mann-Kendall + Theil-Sen aligns with Vincent
+   et al. 2018's Canadian temperature trend methodology** (Sen slope
+   + Kendall's τ). Vincent's iterative lag-1 AC handling is a
+   refinement we don't apply, but raw MK is the correct call for our
+   76-year strong-trend series per Yue & Wang 2002 (cross-rag from
+   snow methodology store). **Our methodology is consistent with the
+   literature consensus on the trend-test question for both snow and
+   temperature.**
+2. **Karl 1993 is the canonical DTR-asymmetry quote**: minimum
+   temperature rose at 3× the rate of maximum (0.84 vs 0.28 °C
+   1951–90 NH); Vose 2005 updates to 0.141 / 0.204 / -0.066 °C/dec
+   for max / min / DTR over 1950-2004 (71% global land coverage).
+3. **EDW (elevation-dependent warming) is real but heterogeneous**
+   per Pepin 2015 + Rangwala & Miller 2012 — important caveat for any
+   "BC mountain AOI warmed faster" claim in the vignette. Mechanism
+   review covers snow-albedo, water vapour, latent heat, aerosols.
+4. **Salmonid thermal envelope citations are now bedded down**:
+   Eaton & Scheller 1996 (foundational US 57-species analysis),
+   Mantua 2010 (PNW salmon climate-stress framing), and Richter &
+   Kolmes 2005 (PNW maximum-temperature criteria for chinook, coho,
+   chum, steelhead). Direct backing for the v0.1.1 "summer daytime
+   maximum is the salmonid thermal envelope" claim.
+5. **No direct ERA5-Land 2m-temperature validation paper for BC** in
+   our corpus — documented as a known caveat. Cross-references via
+   Vincent 2018 / Karl 93 / Vose 05 trend benchmarks instead.
+
+## Citation map
+
+The "cite this for that" table at the end of `findings.md` is an
+**18-row menu**, not an order. Per the plain-language vignette
+philosophy (memory `feedback_vignette_citations_sparse.md`), the
+downstream vignette branch picks sparingly — citations only for
+authorities on findings actually visible in the AOI's graphs/tables.
+BBT-auto-derived keys are baked into the table, ready for `[@key]`
+markers downstream.
+
+## Reproducing the rag store
+
+`data/rag/temp_methodology.duckdb` is gitignored. To rebuild:
+
+1. Download the 10 PDFs from Zotero Web API to
+   `data/rag/temp_methodology_pdfs/` (one file per local label,
+   filename `<label>.pdf`). The `attachKey` for each paper is
+   hardcoded in `scripts/rag_temp_methodology_build.R`.
+2. `ollama serve` + `ollama pull nomic-embed-text` (one-time).
+3. `Rscript scripts/rag_temp_methodology_build.R` — ~28 s, produces
+   677 chunks across 10 sources.
+
+To re-mine: `Rscript scripts/rag_temp_methodology_query.R` →
+regenerates `temp_methodology_quotes.md` (raw retrieval, archived
+here at the time of merge).
+
+## Side-channel issues filed during this work
+
+- **`soul#43`** — convention update for `/lit-search` + `/zotero-api`
+  skills: citation keys must be BBT-auto-derived, never manually set
+  in the `extra` field. Caught when the user reviewed my initial
+  Phase 2 adds; PATCH'd all 10 to clear the override. Future runs
+  will avoid this by following the updated skill guidance.
+- **`#59`** — BEC zone-shift lit-review tracker, parked until after
+  the 3-split lands.
+
+## Notable execution wrinkles
+
+- **BBT plugin compat split** caught during key-capture: BBT 8.x is
+  for Zotero 7, BBT 9.x is for Zotero 8/9. User had BBT 8.0.25 on
+  Zotero 9, which auto-disabled the plugin. Updating to BBT 9.x
+  restored key generation.
+- **Corporate-only authorship** caught for Pepin 2015 (Mountain
+  Research Initiative EDW Working Group). CrossRef returned no
+  individual authors, so BBT initially fell back to a title-based
+  key. PATCH'd in the 21 working-group members from the paper roster
+  (N. Pepin first), BBT regenerated to
+  `pepin_etal2015Elevationdependentwarming` matching convention.
+- **Phase 0 prep** renamed the existing `rag_build_*.R` /
+  `rag_query_*.R` scripts to `rag_*_build.R` / `rag_*_query.R`
+  (`noun_verb` per cd convention), establishing the pattern before
+  adding the 2 new temperature scripts. Touches CLAUDE.md and the
+  snow-archive README's "how to reproduce" section.
+
+## Closing ref
+
+- Issue: NewGraphEnvironment/cd#58
+- PR: NewGraphEnvironment/cd#60 (squash `fa8d74a`, merged 2026-05-05)
+- Release: v0.2.2 (tag pushed)
+- Downstream consumer: future vignette wire-up branch (out of scope
+  for this issue)