Kootenay vignette: five sentences inherited from Peace contradict the Kootenay data #146

If we do it: every sentence in the Kootenay Lake article agrees with the numbers it renders. If we never do: the article states five things its own tables contradict, and the fish passage report appendices ported from it inherit them.

Problem

The Kootenay Lake vignette was templated from the Peace one, and several sentences that are true for Peace are false for Kootenay. Each was measured against the committed inst/vignette-data/kootenay_lake.rds (v0.6.10); line numbers are as of the #102 branch's merge, with the sentence quoted so a shifted line can still be found.

1. Soil moisture is not flat. vignettes/kootenay-lake.Rmd:273 ("Soil moisture is roughly flat") and :992 ("Soil moisture is essentially flat despite the precipitation decline"). The regional annual trend is negative and significant: 1951 slope −0.029 %/yr, Mann-Kendall p = 0.040, and p = 0.004 from 1981. By ecoregion, NCM (p 0.041) and PTR (0.016) are significant from 1951, and all four are significant from 1981 (p 0.002–0.02).
2. PTR is outside the stated warming range. :964 says "All four ecoregions warmed between +1.6 °C and +1.9 °C". The annual tmean window differences (2015–2025 minus 1951–1980) are NCM 1.60, SBF 1.88 and PTR 1.49; TOP (1.71) is not shown in the article (item 5).
3. "Three findings" introduces four. This is at :962, and in Peace at peace-fwcp.Rmd:854: four bold findings follow in each.
4. The WSG drying is consistent in direction, not in strength. :977-979 says all four watershed groups sit "in the same drying signal". All four annual prcp slopes from 1951 are negative, but only KOTL is significant (p 0.015); DUNC is 0.094, LARL 0.054 and SLOC 0.051.
5. The prose counts four ecoregions; the article shows three. kootenay-lake.Rmd:98 keeps ecoregions over 50 km², so Thompson-Okanagan Plateau (TOP, 26 km², about one ERA5-Land cell) is left out of every per-ecoregion trend figure and the roll-up table. (The WSG × ecoregion overlap table does list it, as TOP %, since its data comes from the unfiltered set.) Peace uses 100 km² (peace-fwcp.Rmd:92; since Vignettes: per-ecoregion prcp trend is % of normal per year but labelled mm/yr #102 the Peace data script reads that threshold from the vignette rather than restating it). The Kootenay data script has no filter, so TOP is in the rds but never shown. "Four ecoregions" survives where only three are shown: :614 ("All four ecoregions warmed at roughly the same rate"), :964 ("All four ecoregions warmed between") and :976 ("consistent across all four ecoregions of the AOI"). (:886, "They span the four ecoregions", introduces the overlap table, which does show four.) The Vignettes: per-ecoregion prcp trend is % of normal per year but labelled mm/yr #102 branch already fixed the prcp captions and the intro (:603-611), which now says TOP is left out.

(An earlier version of this item said the Kootenay article had no area filter. That was wrong: the filter is in the vignette, not the data script.)

Proposed Solution

• Rewrite sentences 1–4 from the data. Prefer inline numbers or a data-raw assertion, as northern-bc.Rmd does, over static text.
• Make the count in the prose follow the filter (three ecoregions shown), and decide whether the two articles share one area threshold.

Decided in PR #149: both articles use 100 km². For Kootenay this draws the same three ecoregions as the old 50 km² filter (TOP 26 km² is out either way, PTR 495 km² is in), so the two articles, and the appendices ported from them, follow one rule.

Found by the plan review for #102.