From 7646328c4f63aa698f6d3e946cb70e7fe36cf198 Mon Sep 17 00:00:00 2001
From: almac2022 <al@newgraphenvironment.com>
Date: Sat, 9 May 2026 13:50:43 -0700
Subject: [PATCH 1/7] Initialize PWF baseline for #20 + #43

Combined planning baseline for cd_compare defaults (#20) and
window-vs-window p-value (#43). Issues are explicitly coupled in
#20's issue body ('Tightly coupled with #43; both should land
together'), so single branch / single PR.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 planning/active/findings.md  | 115 +++++++++++++++++++++++++++++++++++
 planning/active/progress.md  |  11 ++++
 planning/active/task_plan.md |  69 +++++++++++++++++++++
 3 files changed, 195 insertions(+)
 create mode 100644 planning/active/findings.md
 create mode 100644 planning/active/progress.md
 create mode 100644 planning/active/task_plan.md

diff --git a/planning/active/findings.md b/planning/active/findings.md
new file mode 100644
index 0000000..1242571
--- /dev/null
+++ b/planning/active/findings.md
@@ -0,0 +1,115 @@
+# Findings — cd_compare defaults + window-vs-window p-value (#20 + #43)
+
+## Issue #20 — context
+
+`cd_compare()` requires the user to specify both time windows
+explicitly. There's no sensible default that helps a first-time
+user understand what climate departure looks like for their AOI.
+
+**Status (post-v0.2.1):** vignettes have settled on a de-facto
+framing — recent **2015-2025** vs WMO-style **1951-1980**, paired
+with both 75-yr and 45-yr Theil-Sen trends. Both `peace-fwcp` and
+`kootenay-lake` use this consistently. The methodology lit review
+from #54 (archived at
+`planning/archive/2026-05-issue-53-snow-lit-review/findings.md`)
+backs the choice — Mann-Kendall + Theil-Sen with no prewhitening
+per Yue & Wang 2002 is correct for our 76-yr series with strong
+trends.
+
+**Remaining scope:** wire those windows in as `cd_compare()`
+defaults so a first-time user gets a sensible answer without
+having to supply both windows. Document the rationale (cumulative
+impact framing — "2 °C warmer" punchline rather than per-year
+rate) in roxygen + a brief note in the vignettes.
+
+Tightly coupled with #43.
+
+## Issue #43 — context
+
+`cd_compare()` returns a window-vs-window difference (mean_diff /
+pct_change) but no measure of statistical confidence. Readers of
+the vignette want to know "is this window-shift real or noise?"
+
+The Mann-Kendall p-value from `cd_trend()` is currently used as a
+proxy in the FWCP Peace vignette but answers a different question
+("is there a monotonic trend?") — not "do the two windows differ?"
+Two windows can differ significantly without a monotonic trend
+(step change, U-shape), and a non-significant trend doesn't always
+mean the windows are statistically indistinguishable.
+
+**Proposal:** add `test = c("t", "wilcox", NULL)` to `cd_compare()`.
+Welch t-test (default) or Mann-Whitney U. Output gains `p_value`
+column (NA when `test = NULL`). Small samples (< 8 in either
+window) → return NA with a warning. Document the independence
+assumption.
+
+## Codebase exploration
+
+### `R/cd_compare.R` (current)
+
+```r
+cd_compare <- function(x, window_a, window_b, method = "mean_diff") {
+  method <- match.arg(method, c("mean_diff", "pct_change"))
+  mean_a <- x |>
+    dplyr::filter(.data$year %in% window_a) |>
+    dplyr::summarise(mean_a = mean(.data$value, na.rm = TRUE), .by = c("variable", "period"))
+  mean_b <- x |>
+    dplyr::filter(.data$year %in% window_b) |>
+    dplyr::summarise(mean_b = mean(.data$value, na.rm = TRUE), .by = c("variable", "period"))
+  n_a <- length(unique(x$year[x$year %in% window_a]))
+  n_b <- length(unique(x$year[x$year %in% window_b]))
+  if (n_a < 2) warning("window_a has fewer than 2 years of data", call. = FALSE)
+  if (n_b < 2) warning("window_b has fewer than 2 years of data", call. = FALSE)
+  out <- dplyr::left_join(mean_a, mean_b, by = c("variable", "period"))
+  out$difference <- switch(method,
+    mean_diff = out$mean_a - out$mean_b,
+    pct_change = (out$mean_a - out$mean_b) / abs(out$mean_b) * 100
+  )
+  out$method <- method
+  out
+}
+```
+
+Returns 6 cols: `variable, period, mean_a, mean_b, difference, method`.
+
+### Existing tests (`tests/testthat/test-cd_compare.R`)
+
+4 tests covering mean_diff, pct_change, small-window warning, and
+multi-variable handling. All use `1956:1960 / 1951:1955` explicit
+windows on a 1951:1960 toy series — independent of the new
+defaults.
+
+### Vignette callsites
+
+- `peace-fwcp.Rmd:534` — `cmp <- cd::cd_compare(ts, 2015:2025, 1951:1980)`
+- `peace-fwcp.Rmd:656` — same in per-ecoregion loop
+- `kootenay-lake.Rmd:561` — same
+- `kootenay-lake.Rmd:688` — same in per-ecoregion loop
+- `R/cd_plot_comparison.R:18` — example; same windows
+
+The `cd_plot_comparison()` example windows can stay explicit
+since it shows a synthetic series (`1956:1960` vs `1951:1955`) —
+or can drop to defaults. Decision to make in Phase 1.
+
+### Reused patterns
+
+- `cd_trend()` uses `match.arg` + per-row dispatch and rounds
+  p-values to 4 decimals (`round(mk$sl[1], 4)`) — Phase 2 should
+  mirror.
+- Existing `n_a < 2` + `n_b < 2` warnings already use
+  `call. = FALSE` and a clean message — pattern reused for the
+  `< 8` test guard.
+
+## Reference periods
+
+The 1951–1980 window is the WMO-style reference. ERA5-Land
+catalogues run 1950 → present (~9 km). Recent 2015–2025 is 11
+years. Both far exceed the n=8 guard, so per-ecoregion + per-WSG
+loops in vignettes will all return real p-values, not NA.
+
+## Open question parked
+
+The issue (#43) suggests `test = "t"` as the default for the next
+minor release — explicit in the issue body. Going with that.
+Plan calls a minor bump (v0.2.8 → v0.3.0) since the schema gains
+a column by default.
diff --git a/planning/active/progress.md b/planning/active/progress.md
new file mode 100644
index 0000000..cc1c5d1
--- /dev/null
+++ b/planning/active/progress.md
@@ -0,0 +1,11 @@
+# Progress — cd_compare defaults + window-vs-window p-value (#20 + #43)
+
+## Session 2026-05-09
+
+- Plan-mode exploration — phases approved by user
+- Created branch `20-43-cd-compare-defaults-pvalue` off main
+  (post-v0.2.8 release commit `51aa556`)
+- Confirmed the two issues are tightly coupled (#20 issue body
+  explicitly says "should land together" with #43)
+- Scaffolded PWF baseline with approved 4-phase plan
+- Next: start Phase 1 — edit `R/cd_compare.R` signature + roxygen
diff --git a/planning/active/task_plan.md b/planning/active/task_plan.md
new file mode 100644
index 0000000..e588e87
--- /dev/null
+++ b/planning/active/task_plan.md
@@ -0,0 +1,69 @@
+# Task: cd_compare defaults + window-vs-window p-value (#20 + #43)
+
+`cd_compare()` requires both `window_a` / `window_b` (no defaults)
+and emits no measure of statistical confidence. Issue #20 settles
+the framing — recent **2015–2025** vs WMO-style **1951–1980** —
+based on what both regional vignettes (`peace-fwcp`, `kootenay-lake`)
+already use consistently and the methodology lit review at
+`planning/archive/2026-05-issue-53-snow-lit-review/findings.md`.
+Issue #43 closes the second gap: a real window-vs-window p-value
+column so `cd_compare()` stops borrowing `cd_trend()`'s
+Mann-Kendall p-value as a proxy. The MK test answers a different
+question (monotonic trend vs window-vs-window difference). Issue
+#20 explicitly says "Tightly coupled with #43; both should land
+together" — single branch, single PR.
+
+## Phase 1 — Add defaults (#20)
+
+- [ ] Edit `R/cd_compare.R` signature: `window_a = 2015:2025`,
+      `window_b = 1951:1980` defaults; update roxygen with the
+      framing rationale (cumulative-impact vs rate-of-change)
+- [ ] Update existing tests in `tests/testthat/test-cd_compare.R` —
+      keep explicit-window tests, add a new test exercising the
+      defaults
+- [ ] `devtools::document()` to refresh `man/cd_compare.Rd`
+- [ ] `devtools::test()` — green
+- [ ] Atomic commit: "Add cd_compare() defaults: 2015:2025 vs 1951:1980 (#20)"
+
+## Phase 2 — Window-vs-window p-value (#43)
+
+- [ ] Edit `R/cd_compare.R` body — add `test = "t"` argument,
+      window-extraction + Welch t / Wilcoxon dispatch, < 8-year
+      small-N guard with batched warning, `p_value` column emitted
+      only when `test != NULL` (default `"t"` so column is present
+      by default)
+- [ ] Add tests covering: clean step-change → tiny p; iid noise →
+      large p; `test = "wilcox"`; `test = NULL` → original 6-col
+      schema; small-N guard fires once with warning; multi-variable
+      input gets per-row p-values
+- [ ] `devtools::document()` to refresh `man/cd_compare.Rd`
+- [ ] `devtools::test()` — green
+- [ ] `lintr::lint_package()` — clean
+- [ ] Atomic commit: "Add window-vs-window p-value to cd_compare() (#43)"
+
+## Phase 3 — Vignette updates
+
+- [ ] `vignettes/peace-fwcp.Rmd` — bare default-driven
+      `cd::cd_compare(ts)` recipe call; comparison table swaps
+      synthesized `trend_p` proxy for new `Δ p (windows)` column
+      (keep `Trend p (75-yr)` alongside — different question);
+      add one-paragraph narrative on the two p-values
+- [ ] `vignettes/kootenay-lake.Rmd` — same edits
+- [ ] Local render of both vignettes — clean
+- [ ] Atomic commit: "Use cd_compare() defaults + window-vs-window p-value in regional vignettes"
+
+## Phase 4 — Release
+
+- [ ] Update `NEWS.md` — single section bullet covering both
+- [ ] Bump `DESCRIPTION` Version: `0.2.8` → `0.3.0` (minor — new
+      arg with default value, output schema gains `p_value` column
+      by default)
+- [ ] PR via `/gh-pr-push`; merge via `/gh-pr-merge`
+- [ ] `/planning-archive` after merge
+
+## Validation
+
+- [ ] Tests pass
+- [ ] `/code-check` clean on each commit
+- [ ] PWF checkboxes match landed work
+- [ ] `/planning-archive` on completion

From 48597cbc09e950ff2c0d49e13a58d1837a4ffd5b Mon Sep 17 00:00:00 2001
From: almac2022 <al@newgraphenvironment.com>
Date: Sat, 9 May 2026 13:51:59 -0700
Subject: [PATCH 2/7] Refresh stale roxygen docs (cd_anomaly, cd_variables)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Stale Rd regenerations from earlier .R-file edits. No source change —
just devtools::document() catching up so subsequent runs don't surface
noise.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 man/cd_anomaly.Rd   | 11 ++++++++---
 man/cd_variables.Rd | 19 +++++++++++++------
 2 files changed, 21 insertions(+), 9 deletions(-)

diff --git a/man/cd_anomaly.Rd b/man/cd_anomaly.Rd
index 1cf059e..209cd88 100644
--- a/man/cd_anomaly.Rd
+++ b/man/cd_anomaly.Rd
@@ -14,7 +14,9 @@ cd_anomaly(x, baseline, cap_pct = 200)
 \code{variable}, \code{period}, \code{baseline_mean}.}
 
 \item{cap_pct}{Numeric. Cap for percent-of-normal anomalies.
-Values beyond +/- \code{cap_pct} are clamped. Default \code{200}.}
+Values beyond +/- \code{cap_pct} are clamped. Default \code{200}. Only
+applies to \code{pct_normal} variables; \code{absolute} and
+\code{pct_point_diff} anomalies are not capped.}
 }
 \value{
 A tibble with columns \code{variable}, \code{period}, \code{year},
@@ -23,8 +25,11 @@ A tibble with columns \code{variable}, \code{period}, \code{year},
 \description{
 Calculates departure from a baseline for each year. Uses
 \code{\link[=cd_variables]{cd_variables()}} to determine the anomaly type: absolute deviation
-for temperature, VPD, and RH; percent of normal for precipitation
-and soil moisture.
+for temperature, VPD, RH, and the annual snow scalars; percent of
+normal for precipitation, soil moisture, and the monthly snow vars
+(\code{swe}, \code{snowfall}, \code{snowmelt}); percentage-point difference for
+variables that are already fractions/percentages (\code{snow_cover},
+\code{snowfall_fraction}).
 }
 \examples{
 catalog <- cd_catalog(
diff --git a/man/cd_variables.Rd b/man/cd_variables.Rd
index 44ecdeb..ff66b68 100644
--- a/man/cd_variables.Rd
+++ b/man/cd_variables.Rd
@@ -11,17 +11,24 @@ A tibble with columns:
 \describe{
 \item{variable}{Short name used throughout the package.}
 \item{long_name}{Human-readable label for plots and tables.}
-\item{unit}{Measurement unit (degree C, percent, Pa).}
-\item{anomaly_type}{"absolute" for direct departures, "pct_normal"
-for percent-of-normal anomalies.}
+\item{unit}{Measurement unit (degree C, percent, Pa, mm, day, mm/wk).}
+\item{anomaly_type}{"absolute" for direct departures (value - baseline,
+reported in the variable's native unit), "pct_normal" for
+percent-of-normal anomalies (100 * value / baseline - 100, capped),
+"pct_point_diff" for departures in percentage points (used for
+variables that are already fractions/percentages, e.g. snow cover,
+snowfall fraction, where pct-of-normal is meaningless and the
+natural delta is value - baseline interpreted as percentage points).}
 \item{era5_name}{ERA5-Land variable name for CDS API requests,
 or NA for derived variables.}
 }
 }
 \description{
-Returns a tibble of metadata for the seven climate variables supported
-by the cd package. Used internally for unit labels, anomaly type routing,
-and ERA5-Land API variable names.
+Returns a tibble of metadata for the climate variables supported by the
+cd package. Used internally for unit labels, anomaly type routing, and
+ERA5-Land API variable names. Covers the seven core climate variables
+plus eight snow-related variables (four monthly natives, four annual
+derived) added in v0.2.0.
 }
 \examples{
 cd_variables()

From 9f56b0494055f66e394da9496eca1ce8f5b48a0b Mon Sep 17 00:00:00 2001
From: almac2022 <al@newgraphenvironment.com>
Date: Sat, 9 May 2026 13:52:20 -0700
Subject: [PATCH 3/7] Add cd_compare() defaults: 2015:2025 vs 1951:1980 (#20)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Wire the de-facto framing both regional vignettes have settled on
into the function signature so first-time users get a sensible
answer without supplying both windows. Roxygen documents the why —
cumulative-impact comparison ('how much warmer is the recent
decade') vs rate-of-change (cd_trend's job), with 1951–1980 as
the WMO-style standard normal reference.

Toy example_catalog only spans 1951:1960 so the in-Rd examples
keep explicit windows; a comment in the example block points users
at the live-catalog defaults.

Fixes #20

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 R/cd_compare.R                   | 22 ++++++++++++++++++----
 man/cd_compare.Rd                | 20 ++++++++++++++++----
 planning/active/progress.md      |  5 ++++-
 planning/active/task_plan.md     | 10 +++++-----
 tests/testthat/test-cd_compare.R | 17 +++++++++++++++++
 5 files changed, 60 insertions(+), 14 deletions(-)

diff --git a/R/cd_compare.R b/R/cd_compare.R
index 673fb20..e69a960 100644
--- a/R/cd_compare.R
+++ b/R/cd_compare.R
@@ -4,10 +4,19 @@
 #' the difference between them. Enables custom comparisons beyond
 #' standard baseline anomalies.
 #'
+#' Defaults compare a recent decade (`2015:2025`) to the WMO-style
+#' standard normal reference period (`1951:1980`) — the framing
+#' used in the package vignettes. This is a *cumulative-impact*
+#' comparison ("how much warmer is the recent decade than the
+#' pre-warming reference?") rather than a rate-of-change
+#' comparison; for the latter use [cd_trend()].
+#'
 #' @param x A tibble from [cd_extract()] with columns `variable`,
 #'   `period`, `year`, `value`.
-#' @param window_a Integer vector of years for the first window.
-#' @param window_b Integer vector of years for the second window.
+#' @param window_a Integer vector of years for the recent / impact
+#'   window. Default `2015:2025`.
+#' @param window_b Integer vector of years for the reference
+#'   window. Default `1951:1980` (WMO-style standard normal).
 #' @param method Character. Comparison method: `"mean_diff"` for
 #'   `mean_a - mean_b`, or `"pct_change"` for percentage change
 #'   relative to window_b.
@@ -25,14 +34,19 @@
 #' )
 #' ts <- cd_extract(catalog, aoi)
 #'
-#' # How has the recent period shifted from the early period?
+#' # The toy example catalog only spans 1951:1960, so the windows
+#' # below override the defaults (2015:2025 vs 1951:1980) — those
+#' # are the right choice on the live STAC catalog.
 #' cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955)
 #'
 #' # Same comparison as percentage change
 #' cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955, method = "pct_change")
 #'
 #' @export
-cd_compare <- function(x, window_a, window_b, method = "mean_diff") {
+cd_compare <- function(x,
+                       window_a = 2015:2025,
+                       window_b = 1951:1980,
+                       method = "mean_diff") {
   method <- match.arg(method, c("mean_diff", "pct_change"))
 
   mean_a <- x |>
diff --git a/man/cd_compare.Rd b/man/cd_compare.Rd
index 2de38cc..d533880 100644
--- a/man/cd_compare.Rd
+++ b/man/cd_compare.Rd
@@ -4,15 +4,17 @@
 \alias{cd_compare}
 \title{Compare arbitrary time windows}
 \usage{
-cd_compare(x, window_a, window_b, method = "mean_diff")
+cd_compare(x, window_a = 2015:2025, window_b = 1951:1980, method = "mean_diff")
 }
 \arguments{
 \item{x}{A tibble from \code{\link[=cd_extract]{cd_extract()}} with columns \code{variable},
 \code{period}, \code{year}, \code{value}.}
 
-\item{window_a}{Integer vector of years for the first window.}
+\item{window_a}{Integer vector of years for the recent / impact
+window. Default \code{2015:2025}.}
 
-\item{window_b}{Integer vector of years for the second window.}
+\item{window_b}{Integer vector of years for the reference
+window. Default \code{1951:1980} (WMO-style standard normal).}
 
 \item{method}{Character. Comparison method: \code{"mean_diff"} for
 \code{mean_a - mean_b}, or \code{"pct_change"} for percentage change
@@ -27,6 +29,14 @@ Computes the mean value for two user-defined time windows and
 the difference between them. Enables custom comparisons beyond
 standard baseline anomalies.
 }
+\details{
+Defaults compare a recent decade (\code{2015:2025}) to the WMO-style
+standard normal reference period (\code{1951:1980}) — the framing
+used in the package vignettes. This is a \emph{cumulative-impact}
+comparison ("how much warmer is the recent decade than the
+pre-warming reference?") rather than a rate-of-change
+comparison; for the latter use \code{\link[=cd_trend]{cd_trend()}}.
+}
 \examples{
 catalog <- cd_catalog(
   system.file("extdata", "example_catalog.json", package = "cd")
@@ -37,7 +47,9 @@ aoi <- sf::st_read(
 )
 ts <- cd_extract(catalog, aoi)
 
-# How has the recent period shifted from the early period?
+# The toy example catalog only spans 1951:1960, so the windows
+# below override the defaults (2015:2025 vs 1951:1980) — those
+# are the right choice on the live STAC catalog.
 cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955)
 
 # Same comparison as percentage change
diff --git a/planning/active/progress.md b/planning/active/progress.md
index cc1c5d1..19c593c 100644
--- a/planning/active/progress.md
+++ b/planning/active/progress.md
@@ -8,4 +8,7 @@
 - Confirmed the two issues are tightly coupled (#20 issue body
   explicitly says "should land together" with #43)
 - Scaffolded PWF baseline with approved 4-phase plan
-- Next: start Phase 1 — edit `R/cd_compare.R` signature + roxygen
+- Phase 1 done — defaults `window_a = 2015:2025`, `window_b = 1951:1980`
+  on `cd_compare()`; roxygen documents cumulative-impact framing;
+  defaults test passes; 16 PASS / 0 FAIL on the cd_compare suite
+- Next: Phase 2 — `test = "t"` argument + p_value column
diff --git a/planning/active/task_plan.md b/planning/active/task_plan.md
index e588e87..5ece5b5 100644
--- a/planning/active/task_plan.md
+++ b/planning/active/task_plan.md
@@ -15,15 +15,15 @@ together" — single branch, single PR.
 
 ## Phase 1 — Add defaults (#20)
 
-- [ ] Edit `R/cd_compare.R` signature: `window_a = 2015:2025`,
+- [x] Edit `R/cd_compare.R` signature: `window_a = 2015:2025`,
       `window_b = 1951:1980` defaults; update roxygen with the
       framing rationale (cumulative-impact vs rate-of-change)
-- [ ] Update existing tests in `tests/testthat/test-cd_compare.R` —
+- [x] Update existing tests in `tests/testthat/test-cd_compare.R` —
       keep explicit-window tests, add a new test exercising the
       defaults
-- [ ] `devtools::document()` to refresh `man/cd_compare.Rd`
-- [ ] `devtools::test()` — green
-- [ ] Atomic commit: "Add cd_compare() defaults: 2015:2025 vs 1951:1980 (#20)"
+- [x] `devtools::document()` to refresh `man/cd_compare.Rd`
+- [x] `devtools::test()` — green
+- [x] Atomic commit: "Add cd_compare() defaults: 2015:2025 vs 1951:1980 (#20)"
 
 ## Phase 2 — Window-vs-window p-value (#43)
 
diff --git a/tests/testthat/test-cd_compare.R b/tests/testthat/test-cd_compare.R
index cbc90a7..315e866 100644
--- a/tests/testthat/test-cd_compare.R
+++ b/tests/testthat/test-cd_compare.R
@@ -41,6 +41,23 @@ test_that("cd_compare warns on small windows", {
   )
 })
 
+test_that("cd_compare uses 2015:2025 vs 1951:1980 defaults", {
+  # Toy series with values that bake in the default-window framing —
+  # 1951:1980 reference at value 0, 2015:2025 recent at value 2.
+  ts <- tibble::tibble(
+    variable = "tmean",
+    period = "annual",
+    year = c(1951:1980, 2015:2025),
+    value = c(rep(0, 30), rep(2, 11))
+  )
+  cmp <- cd_compare(ts)
+
+  expect_named(cmp, c("variable", "period", "mean_a", "mean_b", "difference", "method"))
+  expect_equal(cmp$mean_a, 2)
+  expect_equal(cmp$mean_b, 0)
+  expect_equal(cmp$difference, 2)
+})
+
 test_that("cd_compare handles multiple variables", {
   ts <- tibble::tibble(
     variable = rep(c("tmean", "prcp"), each = 6),

From 4c2d6734dd17d82b19decc7983fda62b8ffe01fb Mon Sep 17 00:00:00 2001
From: almac2022 <al@newgraphenvironment.com>
Date: Sat, 9 May 2026 13:54:37 -0700
Subject: [PATCH 4/7] Add window-vs-window p-value to cd_compare() (#43)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

New `test` argument (default "t") emits a p_value column on the
output. `test = "t"` runs Welch's two-sample t-test on the
annual values in each window; `test = "wilcox"` runs
Mann-Whitney U; `test = NULL` skips the test entirely and drops
the column.

Rows where either window has < 8 non-NA years get
`p_value = NA` and a single batched warning naming the affected
variable/period rows (no per-row warning spam).

The window-vs-window p-value answers a different question than
the cd_trend Mann-Kendall test: "do the two windows differ" vs
"is there a monotonic trend across the full series." Step
changes and U-shapes can produce significant window differences
without monotonic trends, and vice versa. Vignettes will report
both p-values side-by-side in Phase 3.

Roxygen documents the independence assumption; for series with
strong autocorrelation the t-test is mildly anti-conservative —
the Wilcoxon alternative is more robust to non-Gaussian tails
but shares the independence assumption.

Fixes #43

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 R/cd_compare.R                   |  70 ++++++++++++++++++---
 man/cd_compare.Rd                |  45 +++++++++++---
 planning/active/progress.md      |   9 ++-
 planning/active/task_plan.md     |  14 +++--
 tests/testthat/test-cd_compare.R | 103 +++++++++++++++++++++++++++++--
 5 files changed, 213 insertions(+), 28 deletions(-)

diff --git a/R/cd_compare.R b/R/cd_compare.R
index e69a960..e9bd96a 100644
--- a/R/cd_compare.R
+++ b/R/cd_compare.R
@@ -1,8 +1,8 @@
 #' Compare arbitrary time windows
 #'
-#' Computes the mean value for two user-defined time windows and
-#' the difference between them. Enables custom comparisons beyond
-#' standard baseline anomalies.
+#' Computes the mean value for two user-defined time windows, the
+#' difference between them, and (by default) a p-value testing
+#' whether the two windows differ.
 #'
 #' Defaults compare a recent decade (`2015:2025`) to the WMO-style
 #' standard normal reference period (`1951:1980`) — the framing
@@ -11,6 +11,19 @@
 #' pre-warming reference?") rather than a rate-of-change
 #' comparison; for the latter use [cd_trend()].
 #'
+#' The window-vs-window p-value (`test = "t"` Welch t-test or
+#' `test = "wilcox"` Mann-Whitney U) tests whether the means of
+#' the two windows differ — a different question than
+#' [cd_trend()]'s Mann-Kendall test, which checks for a monotonic
+#' trend across the full series. Two windows can differ
+#' significantly without a monotonic trend (step changes, U-shapes)
+#' and a non-significant trend doesn't always mean the windows are
+#' indistinguishable. The test treats annual values as independent;
+#' for series with strong autocorrelation the t-test p is mildly
+#' anti-conservative — the Wilcoxon alternative is more robust to
+#' non-Gaussian tails / outliers but shares the independence
+#' assumption.
+#'
 #' @param x A tibble from [cd_extract()] with columns `variable`,
 #'   `period`, `year`, `value`.
 #' @param window_a Integer vector of years for the recent / impact
@@ -20,9 +33,16 @@
 #' @param method Character. Comparison method: `"mean_diff"` for
 #'   `mean_a - mean_b`, or `"pct_change"` for percentage change
 #'   relative to window_b.
+#' @param test Character or NULL. Window-vs-window significance
+#'   test: `"t"` for Welch's two-sample t-test (default) or
+#'   `"wilcox"` for Mann-Whitney U. Set to `NULL` to skip — output
+#'   then drops the `p_value` column. Rows where either window has
+#'   fewer than 8 non-NA values get `p_value = NA` and a single
+#'   batched warning.
 #'
 #' @return A tibble with columns `variable`, `period`, `mean_a`,
-#'   `mean_b`, `difference`, `method`.
+#'   `mean_b`, `difference`, `method`, and (when `test` is
+#'   non-NULL) `p_value`.
 #'
 #' @examples
 #' catalog <- cd_catalog(
@@ -36,18 +56,22 @@
 #'
 #' # The toy example catalog only spans 1951:1960, so the windows
 #' # below override the defaults (2015:2025 vs 1951:1980) — those
-#' # are the right choice on the live STAC catalog.
-#' cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955)
+#' # are the right choice on the live STAC catalog. test = NULL
+#' # because the toy windows have < 8 years.
+#' cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955, test = NULL)
 #'
 #' # Same comparison as percentage change
-#' cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955, method = "pct_change")
+#' cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955,
+#'            method = "pct_change", test = NULL)
 #'
 #' @export
 cd_compare <- function(x,
                        window_a = 2015:2025,
                        window_b = 1951:1980,
-                       method = "mean_diff") {
+                       method = "mean_diff",
+                       test = "t") {
   method <- match.arg(method, c("mean_diff", "pct_change"))
+  if (!is.null(test)) test <- match.arg(test, c("t", "wilcox"))
 
   mean_a <- x |>
     dplyr::filter(.data$year %in% window_a) |>
@@ -70,5 +94,35 @@ cd_compare <- function(x,
   )
   out$method <- method
 
+  if (!is.null(test)) {
+    p <- vapply(seq_len(nrow(out)), function(i) {
+      v <- out$variable[i]
+      p_ <- out$period[i]
+      a <- x$value[x$variable == v & x$period == p_ & x$year %in% window_a]
+      b <- x$value[x$variable == v & x$period == p_ & x$year %in% window_b]
+      a <- a[!is.na(a)]
+      b <- b[!is.na(b)]
+      if (length(a) < 8 || length(b) < 8) return(NA_real_)
+      pv <- tryCatch(
+        switch(test,
+          t      = stats::t.test(a, b, var.equal = FALSE)$p.value,
+          wilcox = stats::wilcox.test(a, b, exact = FALSE)$p.value
+        ),
+        error = function(e) NA_real_
+      )
+      round(pv, 4)
+    }, numeric(1))
+
+    if (any(is.na(p))) {
+      bad <- which(is.na(p))
+      warning(sprintf(
+        "p_value set to NA for %d row(s) with < 8 years in either window: %s",
+        length(bad),
+        paste(sprintf("%s/%s", out$variable[bad], out$period[bad]), collapse = ", ")
+      ), call. = FALSE)
+    }
+    out$p_value <- p
+  }
+
   out
 }
diff --git a/man/cd_compare.Rd b/man/cd_compare.Rd
index d533880..6ede438 100644
--- a/man/cd_compare.Rd
+++ b/man/cd_compare.Rd
@@ -4,7 +4,13 @@
 \alias{cd_compare}
 \title{Compare arbitrary time windows}
 \usage{
-cd_compare(x, window_a = 2015:2025, window_b = 1951:1980, method = "mean_diff")
+cd_compare(
+  x,
+  window_a = 2015:2025,
+  window_b = 1951:1980,
+  method = "mean_diff",
+  test = "t"
+)
 }
 \arguments{
 \item{x}{A tibble from \code{\link[=cd_extract]{cd_extract()}} with columns \code{variable},
@@ -19,15 +25,23 @@ window. Default \code{1951:1980} (WMO-style standard normal).}
 \item{method}{Character. Comparison method: \code{"mean_diff"} for
 \code{mean_a - mean_b}, or \code{"pct_change"} for percentage change
 relative to window_b.}
+
+\item{test}{Character or NULL. Window-vs-window significance
+test: \code{"t"} for Welch's two-sample t-test (default) or
+\code{"wilcox"} for Mann-Whitney U. Set to \code{NULL} to skip — output
+then drops the \code{p_value} column. Rows where either window has
+fewer than 8 non-NA values get \code{p_value = NA} and a single
+batched warning.}
 }
 \value{
 A tibble with columns \code{variable}, \code{period}, \code{mean_a},
-\code{mean_b}, \code{difference}, \code{method}.
+\code{mean_b}, \code{difference}, \code{method}, and (when \code{test} is
+non-NULL) \code{p_value}.
 }
 \description{
-Computes the mean value for two user-defined time windows and
-the difference between them. Enables custom comparisons beyond
-standard baseline anomalies.
+Computes the mean value for two user-defined time windows, the
+difference between them, and (by default) a p-value testing
+whether the two windows differ.
 }
 \details{
 Defaults compare a recent decade (\code{2015:2025}) to the WMO-style
@@ -36,6 +50,19 @@ used in the package vignettes. This is a \emph{cumulative-impact}
 comparison ("how much warmer is the recent decade than the
 pre-warming reference?") rather than a rate-of-change
 comparison; for the latter use \code{\link[=cd_trend]{cd_trend()}}.
+
+The window-vs-window p-value (\code{test = "t"} Welch t-test or
+\code{test = "wilcox"} Mann-Whitney U) tests whether the means of
+the two windows differ — a different question than
+\code{\link[=cd_trend]{cd_trend()}}'s Mann-Kendall test, which checks for a monotonic
+trend across the full series. Two windows can differ
+significantly without a monotonic trend (step changes, U-shapes)
+and a non-significant trend doesn't always mean the windows are
+indistinguishable. The test treats annual values as independent;
+for series with strong autocorrelation the t-test p is mildly
+anti-conservative — the Wilcoxon alternative is more robust to
+non-Gaussian tails / outliers but shares the independence
+assumption.
 }
 \examples{
 catalog <- cd_catalog(
@@ -49,10 +76,12 @@ ts <- cd_extract(catalog, aoi)
 
 # The toy example catalog only spans 1951:1960, so the windows
 # below override the defaults (2015:2025 vs 1951:1980) — those
-# are the right choice on the live STAC catalog.
-cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955)
+# are the right choice on the live STAC catalog. test = NULL
+# because the toy windows have < 8 years.
+cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955, test = NULL)
 
 # Same comparison as percentage change
-cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955, method = "pct_change")
+cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955,
+           method = "pct_change", test = NULL)
 
 }
diff --git a/planning/active/progress.md b/planning/active/progress.md
index 19c593c..4569272 100644
--- a/planning/active/progress.md
+++ b/planning/active/progress.md
@@ -11,4 +11,11 @@
 - Phase 1 done — defaults `window_a = 2015:2025`, `window_b = 1951:1980`
   on `cd_compare()`; roxygen documents cumulative-impact framing;
   defaults test passes; 16 PASS / 0 FAIL on the cd_compare suite
-- Next: Phase 2 — `test = "t"` argument + p_value column
+- Phase 2 done — `test = c("t", "wilcox", NULL)` argument added;
+  default `"t"` so output gains `p_value` column by default;
+  small-N guard (< 8 years either window) → NA + single batched
+  warning naming affected variable/period rows; 6 new tests
+  covering step-change, iid noise, wilcox, NULL, small-N guard,
+  multi-variable; full suite 181 PASS / 0 FAIL
+- Next: Phase 3 — vignette wire-up (drop synthesized trend_p
+  proxy, surface real `Δ p (windows)` column)
diff --git a/planning/active/task_plan.md b/planning/active/task_plan.md
index 5ece5b5..dc72339 100644
--- a/planning/active/task_plan.md
+++ b/planning/active/task_plan.md
@@ -27,19 +27,21 @@ together" — single branch, single PR.
 
 ## Phase 2 — Window-vs-window p-value (#43)
 
-- [ ] Edit `R/cd_compare.R` body — add `test = "t"` argument,
+- [x] Edit `R/cd_compare.R` body — add `test = "t"` argument,
       window-extraction + Welch t / Wilcoxon dispatch, < 8-year
       small-N guard with batched warning, `p_value` column emitted
       only when `test != NULL` (default `"t"` so column is present
       by default)
-- [ ] Add tests covering: clean step-change → tiny p; iid noise →
+- [x] Add tests covering: clean step-change → tiny p; iid noise →
       large p; `test = "wilcox"`; `test = NULL` → original 6-col
       schema; small-N guard fires once with warning; multi-variable
       input gets per-row p-values
-- [ ] `devtools::document()` to refresh `man/cd_compare.Rd`
-- [ ] `devtools::test()` — green
-- [ ] `lintr::lint_package()` — clean
-- [ ] Atomic commit: "Add window-vs-window p-value to cd_compare() (#43)"
+- [x] `devtools::document()` to refresh `man/cd_compare.Rd`
+- [x] `devtools::test()` — green
+- [x] `lintr::lint_package()` — clean (no new lints from
+      cd_compare; pre-existing `.data` pronoun + vignette caption
+      length unchanged)
+- [x] Atomic commit: "Add window-vs-window p-value to cd_compare() (#43)"
 
 ## Phase 3 — Vignette updates
 
diff --git a/tests/testthat/test-cd_compare.R b/tests/testthat/test-cd_compare.R
index 315e866..8b381ae 100644
--- a/tests/testthat/test-cd_compare.R
+++ b/tests/testthat/test-cd_compare.R
@@ -5,7 +5,7 @@ test_that("cd_compare computes mean_diff", {
     year = 1951:1960,
     value = c(rep(0, 5), rep(10, 5))
   )
-  cmp <- cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955)
+  cmp <- cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955, test = NULL)
 
   expect_s3_class(cmp, "tbl_df")
   expect_named(cmp, c("variable", "period", "mean_a", "mean_b", "difference", "method"))
@@ -22,7 +22,8 @@ test_that("cd_compare computes pct_change", {
     year = 1951:1960,
     value = c(rep(100, 5), rep(150, 5))
   )
-  cmp <- cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955, method = "pct_change")
+  cmp <- cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955,
+                    method = "pct_change", test = NULL)
 
   expect_equal(cmp$difference, 50)
   expect_equal(cmp$method, "pct_change")
@@ -36,7 +37,7 @@ test_that("cd_compare warns on small windows", {
     value = 1:3
   )
   expect_warning(
-    cd_compare(ts, window_a = 1953, window_b = 1951:1952),
+    cd_compare(ts, window_a = 1953, window_b = 1951:1952, test = NULL),
     "window_a has fewer than 2 years"
   )
 })
@@ -50,7 +51,7 @@ test_that("cd_compare uses 2015:2025 vs 1951:1980 defaults", {
     year = c(1951:1980, 2015:2025),
     value = c(rep(0, 30), rep(2, 11))
   )
-  cmp <- cd_compare(ts)
+  cmp <- cd_compare(ts, test = NULL)
 
   expect_named(cmp, c("variable", "period", "mean_a", "mean_b", "difference", "method"))
   expect_equal(cmp$mean_a, 2)
@@ -65,9 +66,101 @@ test_that("cd_compare handles multiple variables", {
     year = rep(1951:1956, 2),
     value = c(0, 0, 0, 10, 10, 10, 100, 100, 100, 200, 200, 200)
   )
-  cmp <- cd_compare(ts, window_a = 1954:1956, window_b = 1951:1953)
+  cmp <- cd_compare(ts, window_a = 1954:1956, window_b = 1951:1953, test = NULL)
 
   expect_equal(nrow(cmp), 2)
   expect_equal(cmp$difference[cmp$variable == "tmean"], 10)
   expect_equal(cmp$difference[cmp$variable == "prcp"], 100)
 })
+
+# ---------------------------------------------------------------------------
+# Window-vs-window p-value tests (#43)
+# ---------------------------------------------------------------------------
+
+test_that("cd_compare returns p_value column when test = 't' (default)", {
+  set.seed(1)
+  # Clean step change between two 30-year windows — Welch t should
+  # return a vanishingly small p.
+  ts <- tibble::tibble(
+    variable = "tmean",
+    period = "annual",
+    year = c(1951:1980, 1991:2020),
+    value = c(rnorm(30, mean = 0, sd = 1), rnorm(30, mean = 5, sd = 1))
+  )
+  cmp <- cd_compare(ts, window_a = 1991:2020, window_b = 1951:1980)
+
+  expect_true("p_value" %in% names(cmp))
+  expect_equal(nrow(cmp), 1)
+  expect_lt(cmp$p_value, 0.001)
+})
+
+test_that("cd_compare t-test returns large p on iid noise", {
+  set.seed(2)
+  ts <- tibble::tibble(
+    variable = "tmean",
+    period = "annual",
+    year = c(1951:1980, 1991:2020),
+    value = rnorm(60, mean = 0, sd = 1)
+  )
+  cmp <- cd_compare(ts, window_a = 1991:2020, window_b = 1951:1980)
+
+  expect_gt(cmp$p_value, 0.05)
+})
+
+test_that("cd_compare supports test = 'wilcox'", {
+  set.seed(3)
+  ts <- tibble::tibble(
+    variable = "tmean",
+    period = "annual",
+    year = c(1951:1980, 1991:2020),
+    value = c(rnorm(30, mean = 0), rnorm(30, mean = 5))
+  )
+  cmp <- cd_compare(ts, window_a = 1991:2020, window_b = 1951:1980, test = "wilcox")
+
+  expect_true("p_value" %in% names(cmp))
+  expect_lt(cmp$p_value, 0.001)
+})
+
+test_that("cd_compare with test = NULL drops p_value column", {
+  ts <- tibble::tibble(
+    variable = "tmean",
+    period = "annual",
+    year = c(1951:1980, 1991:2020),
+    value = c(rep(0, 30), rep(5, 30))
+  )
+  cmp <- cd_compare(ts, window_a = 1991:2020, window_b = 1951:1980, test = NULL)
+
+  expect_named(cmp, c("variable", "period", "mean_a", "mean_b", "difference", "method"))
+})
+
+test_that("cd_compare small-N guard returns NA p_value with one warning", {
+  ts <- tibble::tibble(
+    variable = "tmean",
+    period = "annual",
+    year = 1951:1960,
+    value = c(rep(0, 5), rep(5, 5))
+  )
+  expect_warning(
+    cmp <- cd_compare(ts, window_a = 1956:1960, window_b = 1951:1955),
+    "p_value set to NA"
+  )
+  expect_true(is.na(cmp$p_value))
+})
+
+test_that("cd_compare per-row p_values on multi-variable input", {
+  set.seed(4)
+  # tmean: real shift (small p); prcp: no shift (large p)
+  ts <- tibble::tibble(
+    variable = rep(c("tmean", "prcp"), each = 60),
+    period = "annual",
+    year = rep(c(1951:1980, 1991:2020), 2),
+    value = c(
+      c(rnorm(30, mean = 0, sd = 1), rnorm(30, mean = 5, sd = 1)),
+      rnorm(60, mean = 100, sd = 5)
+    )
+  )
+  cmp <- cd_compare(ts, window_a = 1991:2020, window_b = 1951:1980)
+
+  expect_lt(cmp$p_value[cmp$variable == "tmean"], 0.001)
+  expect_gt(cmp$p_value[cmp$variable == "prcp"], 0.05)
+})

From 46d6792e9207609ee09cafe2f6918d7e4a4d7aeb Mon Sep 17 00:00:00 2001
From: almac2022 <al@newgraphenvironment.com>
Date: Sat, 9 May 2026 13:58:20 -0700
Subject: [PATCH 5/7] Use cd_compare() defaults + window-vs-window p-value in
 regional vignettes
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Both vignettes:
- Compare-recipe chunk now shows bare `cd::cd_compare(ts)` —
  defaults (2015:2025 vs 1951:1980, Welch t) carry the call.
  Cleaner cd showcase; reader sees the recipe in 1 line.
- Compare-table chunk recomputes `cmp` from cached `ts` so the
  new `p_value` column flows through without re-running the
  heavy precompute scripts. The cached `regional` predates
  #43 so doesn't have the column.
- Table swaps the synthesized Mann-Kendall `trend_p` proxy for
  the real window-vs-window `Δ p (windows)` (Welch). `Trend p
  (75-yr)` stays alongside — different question, both belong.
- One-paragraph narrative reframes the two p-values: "do the
  windows differ?" (window p) vs "is there a monotonic ramp?"
  (trend p). Step changes vs gradual ramps read differently.

`data-raw/<vignette>_vignette_data.R` updated so the next
precompute run uses cd_compare defaults.

Both vignettes render clean locally with both column headers
present in the rendered HTML.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 data-raw/kootenay_lake_vignette_data.R | 10 +++----
 data-raw/peace_fwcp_vignette_data.R    |  8 ++---
 planning/active/progress.md            | 11 +++++--
 planning/active/task_plan.md           | 11 ++++---
 vignettes/kootenay-lake.Rmd            | 41 ++++++++++++++++----------
 vignettes/peace-fwcp.Rmd               | 38 ++++++++++++++----------
 6 files changed, 71 insertions(+), 48 deletions(-)

diff --git a/data-raw/kootenay_lake_vignette_data.R b/data-raw/kootenay_lake_vignette_data.R
index 68d127e..1ef1236 100644
--- a/data-raw/kootenay_lake_vignette_data.R
+++ b/data-raw/kootenay_lake_vignette_data.R
@@ -48,12 +48,10 @@ regional_ts  <- cd_extract(catalog, aoi)
 regional_bl  <- cd_baseline(regional_ts, baseline_years = 1951:1980)
 regional_ano <- cd_anomaly(regional_ts, regional_bl)
 regional_trn <- cd_trend(regional_ano, trend_start = c(1951, 1981))
-regional_cmp <- cd_compare(regional_ts,
-                           window_a = 2015:2025, window_b = 1951:1980,
-                           method = "mean_diff")
+regional_cmp <- cd_compare(regional_ts, method = "mean_diff")  # defaults: 2015:2025 vs 1951:1980, Welch t
 regional_cmp_pct <- cd_compare(
   regional_ts[regional_ts$variable %in% pct_normal_vars, ],
-  window_a = 2015:2025, window_b = 1951:1980, method = "pct_change"
+  method = "pct_change"
 )
 
 # -- Per-ecoregion -----------------------------------------------------------
@@ -65,7 +63,7 @@ ecoregion_results <- lapply(seq_len(nrow(ecoregions)), function(i) {
   trn <- cd_trend(ano, trend_start = c(1951, 1981))
   cmp_pct <- cd_compare(
     ts[ts$variable %in% pct_normal_vars, ],
-    window_a = 2015:2025, window_b = 1951:1980, method = "pct_change"
+    method = "pct_change"
   )
   list(ts = ts, ano = ano, trn = trn, cmp_pct = cmp_pct)
 })
@@ -81,7 +79,7 @@ wsg_results <- lapply(seq_len(nrow(wsgs)), function(i) {
   trn <- cd_trend(ano, trend_start = c(1951, 1981))
   cmp_pct <- cd_compare(
     ts[ts$variable %in% pct_normal_vars, ],
-    window_a = 2015:2025, window_b = 1951:1980, method = "pct_change"
+    method = "pct_change"
   )
   list(ts = ts, ano = ano, trn = trn, cmp_pct = cmp_pct)
 })
diff --git a/data-raw/peace_fwcp_vignette_data.R b/data-raw/peace_fwcp_vignette_data.R
index ad9540e..d3728a2 100644
--- a/data-raw/peace_fwcp_vignette_data.R
+++ b/data-raw/peace_fwcp_vignette_data.R
@@ -42,16 +42,14 @@ regional_ts  <- cd_extract(catalog, aoi)
 regional_bl  <- cd_baseline(regional_ts, baseline_years = 1951:1980)
 regional_ano <- cd_anomaly(regional_ts, regional_bl)
 regional_trn <- cd_trend(regional_ano, trend_start = c(1951, 1981))
-regional_cmp <- cd_compare(regional_ts,
-                           window_a = 2015:2025, window_b = 1951:1980,
-                           method = "mean_diff")
+regional_cmp <- cd_compare(regional_ts, method = "mean_diff")  # defaults: 2015:2025 vs 1951:1980, Welch t
 # pct_change for vars with pct_normal anomaly type (#48 added swe / snowfall /
 # snowmelt to this list; snow_cover and snowfall_fraction are pct_point_diff so
 # mean_diff is the right comparison method for those — already in regional_cmp).
 pct_normal_vars <- c("prcp", "soil_moisture", "swe", "snowfall", "snowmelt")
 regional_cmp_pct <- cd_compare(
   regional_ts[regional_ts$variable %in% pct_normal_vars, ],
-  window_a = 2015:2025, window_b = 1951:1980, method = "pct_change"
+  method = "pct_change"
 )
 
 # -- Per-ecoregion -----------------------------------------------------------
@@ -63,7 +61,7 @@ ecoregion_results <- lapply(seq_len(nrow(ecoregions)), function(i) {
   trn <- cd_trend(ano, trend_start = c(1951, 1981))
   cmp_pct <- cd_compare(
     ts[ts$variable %in% pct_normal_vars, ],
-    window_a = 2015:2025, window_b = 1951:1980, method = "pct_change"
+    method = "pct_change"
   )
   list(ts = ts, ano = ano, trn = trn, cmp_pct = cmp_pct)
 })
diff --git a/planning/active/progress.md b/planning/active/progress.md
index 4569272..e555edc 100644
--- a/planning/active/progress.md
+++ b/planning/active/progress.md
@@ -17,5 +17,12 @@
   warning naming affected variable/period rows; 6 new tests
   covering step-change, iid noise, wilcox, NULL, small-N guard,
   multi-variable; full suite 181 PASS / 0 FAIL
-- Next: Phase 3 — vignette wire-up (drop synthesized trend_p
-  proxy, surface real `Δ p (windows)` column)
+- Phase 3 done — both vignettes show bare `cd::cd_compare(ts)`
+  recipe call (defaults take over); compare-table chunks
+  recompute `cmp` from cached `ts` so `p_value` flows through
+  without re-running the heavy precompute scripts; surfaces both
+  `Δ p (windows)` and `Trend p (75-yr)` columns side-by-side;
+  narrative paragraph explains the two-test framing; precompute
+  scripts updated for next refresh; both vignettes render clean
+  with the new column visible in HTML
+- Next: Phase 4 — NEWS + DESCRIPTION bump → v0.3.0; PR + merge
diff --git a/planning/active/task_plan.md b/planning/active/task_plan.md
index dc72339..3450f57 100644
--- a/planning/active/task_plan.md
+++ b/planning/active/task_plan.md
@@ -45,14 +45,17 @@ together" — single branch, single PR.
 
 ## Phase 3 — Vignette updates
 
-- [ ] `vignettes/peace-fwcp.Rmd` — bare default-driven
+- [x] `vignettes/peace-fwcp.Rmd` — bare default-driven
       `cd::cd_compare(ts)` recipe call; comparison table swaps
       synthesized `trend_p` proxy for new `Δ p (windows)` column
       (keep `Trend p (75-yr)` alongside — different question);
       add one-paragraph narrative on the two p-values
-- [ ] `vignettes/kootenay-lake.Rmd` — same edits
-- [ ] Local render of both vignettes — clean
-- [ ] Atomic commit: "Use cd_compare() defaults + window-vs-window p-value in regional vignettes"
+- [x] `vignettes/kootenay-lake.Rmd` — same edits
+- [x] `data-raw/<vignette>_vignette_data.R` — drop explicit
+      windows so future precompute runs use defaults
+- [x] Local render of both vignettes — clean (HTML contains both
+      `Δ p (windows)` and `Trend p (75-yr)` column headers)
+- [x] Atomic commit: "Use cd_compare() defaults + window-vs-window p-value in regional vignettes"
 
 ## Phase 4 — Release
 
diff --git a/vignettes/kootenay-lake.Rmd b/vignettes/kootenay-lake.Rmd
index e6987e9..6282c00 100644
--- a/vignettes/kootenay-lake.Rmd
+++ b/vignettes/kootenay-lake.Rmd
@@ -543,25 +543,33 @@ The table below compares two windows directly — the recent decade
 absolute is the difference of the two means in the variable's native
 units. Δ % is shown only for variables where it is meaningful
 (precipitation, soil moisture, vapour pressure deficit, relative
-humidity). The trend p column is the Mann-Kendall p-value of the
-75-year (1951–present) trend on the same variable and period; it
-tests for a steady year-on-year ramp, which is a related but distinct
-question from "do these two windows differ".
+humidity). Two p-values are reported, answering different
+questions. **Δ p (windows)** is the Welch t-test on the annual
+values within each window — does the recent decade differ from
+the pre-warming reference? **Trend p (75-yr)** is the Mann-Kendall
+test on the full 1951–present series — is there a steady
+year-on-year ramp? Step changes show up as significant Δ p with
+non-significant trend p; gradual ramps show up as both significant.
 
 The recent decade was 1.6 to 1.8 °C warmer than the pre-warming
 reference for annual mean, daytime maximum, and overnight minimum,
-with Mann-Kendall trend p-values below 0.001. Vapour pressure deficit
-is up significantly. Annual precipitation was about 6 to 7 % *lower*
-in the recent decade — and unlike the FWCP Peace just to the north,
-the long-term Mann-Kendall trend test does confirm a steady
-year-on-year decline (p ≈ 0.02). Soil moisture is roughly flat.
-Relative humidity shows a small significant decline.
+with both window and trend p-values below 0.001. Vapour pressure
+deficit is up significantly on both tests. Annual precipitation
+was about 6 to 7 % *lower* in the recent decade — and unlike the
+FWCP Peace just to the north, the long-term Mann-Kendall trend
+test does confirm a steady year-on-year decline (trend p ≈ 0.02).
+Soil moisture is roughly flat. Relative humidity shows a small
+significant decline.
 
 ```{r compare-recipe, eval = FALSE}
-cmp <- cd::cd_compare(ts, window_a = 2015:2025, window_b = 1951:1980)
+cmp <- cd::cd_compare(ts)   # defaults: 2015–2025 vs 1951–1980, Welch t-test
 ```
 
 ```{r compare-table, echo = FALSE}
+# Recompute cmp from the cached ts so the p_value column reflects the
+# current cd_compare() defaults — the precomputed rds was generated
+# before #43 added the test argument.
+cmp <- cd::cd_compare(ts)
 trn_p <- trn[trn$trend_start == 1951, c("variable", "period", "mk_pvalue")]
 names(trn_p)[3] <- "trend_p"
 trn_p$trend_p <- round(trn_p$trend_p, 3)
@@ -580,17 +588,18 @@ cmp_combined <- cmp |>
       NA_real_,
       round(100 * (mean_a - mean_b) / mean_b, 1)
     ),
-    mean_a   = round(mean_a, 2),
-    mean_b   = round(mean_b, 2),
-    abs_diff = round(difference, 2)
+    mean_a    = round(mean_a, 2),
+    mean_b    = round(mean_b, 2),
+    abs_diff  = round(difference, 2),
+    window_p  = round(p_value, 3)
   ) |>
-  dplyr::select(variable, period, mean_a, mean_b, abs_diff, pct_change) |>
+  dplyr::select(variable, period, mean_a, mean_b, abs_diff, pct_change, window_p) |>
   merge(trn_p, by = c("variable", "period"), all.x = TRUE)
 
 names(cmp_combined) <- c(
   "Variable", "Period",
   "Recent (2015–2025)", "Pre-warming (1951–1980)",
-  "Δ absolute", "Δ %", "Trend p (75-yr)"
+  "Δ absolute", "Δ %", "Δ p (windows)", "Trend p (75-yr)"
 )
 
 kableExtra::kable_styling(
diff --git a/vignettes/peace-fwcp.Rmd b/vignettes/peace-fwcp.Rmd
index 88ade99..ac11e84 100644
--- a/vignettes/peace-fwcp.Rmd
+++ b/vignettes/peace-fwcp.Rmd
@@ -517,24 +517,31 @@ The table below compares two windows directly — the recent decade
 absolute is the difference of the two means in the variable's native
 units. Δ % is shown only for variables where it is meaningful
 (precipitation, soil moisture, vapour pressure deficit, relative
-humidity). The trend p column is the Mann-Kendall p-value of the
-75-year (1951–present) trend on the same variable and period; it
-tests for a steady year-on-year ramp, which is a related but distinct
-question from "do these two windows differ".
+humidity). Two p-values are reported, answering different
+questions. **Δ p (windows)** is the Welch t-test on the annual
+values within each window — does the recent decade differ from
+the pre-warming reference? **Trend p (75-yr)** is the Mann-Kendall
+test on the full 1951–present series — is there a steady
+year-on-year ramp? Step changes show up as significant Δ p with
+non-significant trend p; gradual ramps show up as both significant.
 
 The recent decade was 1.7 to 2.4 °C warmer than the pre-warming
 reference for annual mean, daytime maximum, and overnight minimum,
-with Mann-Kendall trend p-values below 0.001. Vapour pressure deficit
-is up significantly. Annual precipitation was about 3 to 4 % higher
-in the recent decade; the long-term trend test does not confirm a
-steady year-on-year ramp (p ≈ 0.20). Soil moisture and relative
-humidity show no meaningful change.
+with both window and trend p-values below 0.001. Vapour pressure
+deficit is up significantly on both tests. Annual precipitation
+was about 3 to 4 % higher in the recent decade; neither test
+confirms the shift (Δ p ≈ 0.4, trend p ≈ 0.20). Soil moisture and
+relative humidity show no meaningful change.
 
 ```{r compare-recipe, eval = FALSE}
-cmp <- cd::cd_compare(ts, window_a = 2015:2025, window_b = 1951:1980)
+cmp <- cd::cd_compare(ts)   # defaults: 2015–2025 vs 1951–1980, Welch t-test
 ```
 
 ```{r compare-table, echo = FALSE}
+# Recompute cmp from the cached ts so the p_value column reflects the
+# current cd_compare() defaults — the precomputed rds was generated
+# before #43 added the test argument.
+cmp <- cd::cd_compare(ts)
 trn_p <- trn[trn$trend_start == 1951, c("variable", "period", "mk_pvalue")]
 names(trn_p)[3] <- "trend_p"
 trn_p$trend_p <- round(trn_p$trend_p, 3)
@@ -553,17 +560,18 @@ cmp_combined <- cmp |>
       NA_real_,
       round(100 * (mean_a - mean_b) / mean_b, 1)
     ),
-    mean_a   = round(mean_a, 2),
-    mean_b   = round(mean_b, 2),
-    abs_diff = round(difference, 2)
+    mean_a    = round(mean_a, 2),
+    mean_b    = round(mean_b, 2),
+    abs_diff  = round(difference, 2),
+    window_p  = round(p_value, 3)
   ) |>
-  dplyr::select(variable, period, mean_a, mean_b, abs_diff, pct_change) |>
+  dplyr::select(variable, period, mean_a, mean_b, abs_diff, pct_change, window_p) |>
   merge(trn_p, by = c("variable", "period"), all.x = TRUE)
 
 names(cmp_combined) <- c(
   "Variable", "Period",
   "Recent (2015–2025)", "Pre-warming (1951–1980)",
-  "Δ absolute", "Δ %", "Trend p (75-yr)"
+  "Δ absolute", "Δ %", "Δ p (windows)", "Trend p (75-yr)"
 )
 
 kableExtra::kable_styling(

From 88e697ef74716b760f77f107c86911a5acb1570a Mon Sep 17 00:00:00 2001
From: almac2022 <al@newgraphenvironment.com>
Date: Sat, 9 May 2026 14:04:28 -0700
Subject: [PATCH 6/7] Mark Phase 4 release-flow notes (PR + post-merge bump)

Repo pattern: NEWS + DESCRIPTION bump lands as a separate
'Release v0.X.Y' commit on main post-merge, driven by
/gh-pr-merge. The branch itself stays release-prep clean.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 planning/active/task_plan.md | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/planning/active/task_plan.md b/planning/active/task_plan.md
index 3450f57..d5b1472 100644
--- a/planning/active/task_plan.md
+++ b/planning/active/task_plan.md
@@ -59,11 +59,12 @@ together" — single branch, single PR.
 
 ## Phase 4 — Release
 
-- [ ] Update `NEWS.md` — single section bullet covering both
-- [ ] Bump `DESCRIPTION` Version: `0.2.8` → `0.3.0` (minor — new
-      arg with default value, output schema gains `p_value` column
-      by default)
-- [ ] PR via `/gh-pr-push`; merge via `/gh-pr-merge`
+- [ ] PR via `/gh-pr-push` (NEWS + DESCRIPTION bump handled
+      post-merge by `/gh-pr-merge`, per repo convention — recent
+      pattern: `Release v0.X.Y` is a separate post-merge commit
+      on main, e.g. `298cce4 Release v0.2.8`)
+- [ ] Bump on merge: `0.2.8` → `0.3.0` (minor — output schema
+      gains `p_value` column by default; new `test` argument)
 - [ ] `/planning-archive` after merge
 
 ## Validation

From 4a193798109d5277da264592af112d99fc18ee35 Mon Sep 17 00:00:00 2001
From: almac2022 <al@newgraphenvironment.com>
Date: Sun, 10 May 2026 00:02:20 -0700
Subject: [PATCH 7/7] Move Recent Decade vs Pre-Warming Reference up + rename
 for prominence
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The cumulative-impact comparison (mean(2015-2025) - mean(1951-1980))
is the headline number this package produces — it was buried 300+
lines down between Snowpack and Spatial Pattern, and titled with
methodological framing ('Recent vs Pre-warming') rather than
something that telegraphs the answer.

Two changes in both vignettes:

- Rename: 'Recent vs Pre-warming' -> 'Recent Decade vs Pre-Warming
  Reference'. Adds 'decade' + 'reference' so the heading reads as
  a comparison label rather than a methods stub.
- Reorder: place directly after Trends, before Daytime Highs and
  Overnight Lows. Trends still introduces the 1951-1980 reference
  and the two-test framing the table relies on, then the headline
  cumulative number lands; DTR / Snowpack / Spatial Pattern read
  as drill-down on the headline rather than a long detour before it.

Both vignettes render clean; new heading visible in HTML, old
heading absent.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 vignettes/kootenay-lake.Rmd | 154 ++++++++++++++++++------------------
 vignettes/peace-fwcp.Rmd    | 150 +++++++++++++++++------------------
 2 files changed, 152 insertions(+), 152 deletions(-)

diff --git a/vignettes/kootenay-lake.Rmd b/vignettes/kootenay-lake.Rmd
index 6282c00..ecaaed0 100644
--- a/vignettes/kootenay-lake.Rmd
+++ b/vignettes/kootenay-lake.Rmd
@@ -261,6 +261,83 @@ cd::cd_plot_timeseries(
 )
 ```
 
+## Recent Decade vs Pre-Warming Reference
+
+The table below compares two windows directly — the recent decade
+(2015–2025) against the pre-warming reference (1951–1980). Δ
+absolute is the difference of the two means in the variable's native
+units. Δ % is shown only for variables where it is meaningful
+(precipitation, soil moisture, vapour pressure deficit, relative
+humidity). Two p-values are reported, answering different
+questions. **Δ p (windows)** is the Welch t-test on the annual
+values within each window — does the recent decade differ from
+the pre-warming reference? **Trend p (75-yr)** is the Mann-Kendall
+test on the full 1951–present series — is there a steady
+year-on-year ramp? Step changes show up as significant Δ p with
+non-significant trend p; gradual ramps show up as both significant.
+
+The recent decade was 1.6 to 1.8 °C warmer than the pre-warming
+reference for annual mean, daytime maximum, and overnight minimum,
+with both window and trend p-values below 0.001. Vapour pressure
+deficit is up significantly on both tests. Annual precipitation
+was about 6 to 7 % *lower* in the recent decade — and unlike the
+FWCP Peace just to the north, the long-term Mann-Kendall trend
+test does confirm a steady year-on-year decline (trend p ≈ 0.02).
+Soil moisture is roughly flat. Relative humidity shows a small
+significant decline.
+
+```{r compare-recipe, eval = FALSE}
+cmp <- cd::cd_compare(ts)   # defaults: 2015–2025 vs 1951–1980, Welch t-test
+```
+
+```{r compare-table, echo = FALSE}
+# Recompute cmp from the cached ts so the p_value column reflects the
+# current cd_compare() defaults — the precomputed rds was generated
+# before #43 added the test argument.
+cmp <- cd::cd_compare(ts)
+trn_p <- trn[trn$trend_start == 1951, c("variable", "period", "mk_pvalue")]
+names(trn_p)[3] <- "trend_p"
+trn_p$trend_p <- round(trn_p$trend_p, 3)
+
+no_pct_vars <- c(
+  "tmean", "tmax", "tmin",
+  # snow_cover and snowfall_fraction are already in % (pct_point_diff
+  # anomaly type) — pct-of-baseline mixes units. snowmelt_doy_50 is a
+  # day-of-year — pct-change of a date is meaningless.
+  "snow_cover", "snowfall_fraction", "snowmelt_doy_50"
+)
+cmp_combined <- cmp |>
+  dplyr::mutate(
+    pct_change = ifelse(
+      variable %in% no_pct_vars,
+      NA_real_,
+      round(100 * (mean_a - mean_b) / mean_b, 1)
+    ),
+    mean_a    = round(mean_a, 2),
+    mean_b    = round(mean_b, 2),
+    abs_diff  = round(difference, 2),
+    window_p  = round(p_value, 3)
+  ) |>
+  dplyr::select(variable, period, mean_a, mean_b, abs_diff, pct_change, window_p) |>
+  merge(trn_p, by = c("variable", "period"), all.x = TRUE)
+
+names(cmp_combined) <- c(
+  "Variable", "Period",
+  "Recent (2015–2025)", "Pre-warming (1951–1980)",
+  "Δ absolute", "Δ %", "Δ p (windows)", "Trend p (75-yr)"
+)
+
+kableExtra::kable_styling(
+  knitr::kable(cmp_combined, label = NA,
+    caption = "Recent decade (2015-2025 mean) compared to pre-warming reference (1951-1980 mean) for the Kootenay Lake Region.",
+    row.names = FALSE),
+  bootstrap_options = c("striped", "hover", "condensed")
+) |>
+  kableExtra::scroll_box(height = "360px")
+```
+
+<br>
+
 ## Daytime Highs and Overnight Lows
 
 The cd package ships daytime maximum (tmax) and overnight minimum (tmin)
@@ -536,83 +613,6 @@ decade. For aquatic ecosystems downstream, this is a loss of
 late-season cold-water input to streams during the warmest, most
 thermally stressful weeks of the year.
 
-## Recent vs Pre-warming
-
-The table below compares two windows directly — the recent decade
-(2015–2025) against the pre-warming reference (1951–1980). Δ
-absolute is the difference of the two means in the variable's native
-units. Δ % is shown only for variables where it is meaningful
-(precipitation, soil moisture, vapour pressure deficit, relative
-humidity). Two p-values are reported, answering different
-questions. **Δ p (windows)** is the Welch t-test on the annual
-values within each window — does the recent decade differ from
-the pre-warming reference? **Trend p (75-yr)** is the Mann-Kendall
-test on the full 1951–present series — is there a steady
-year-on-year ramp? Step changes show up as significant Δ p with
-non-significant trend p; gradual ramps show up as both significant.
-
-The recent decade was 1.6 to 1.8 °C warmer than the pre-warming
-reference for annual mean, daytime maximum, and overnight minimum,
-with both window and trend p-values below 0.001. Vapour pressure
-deficit is up significantly on both tests. Annual precipitation
-was about 6 to 7 % *lower* in the recent decade — and unlike the
-FWCP Peace just to the north, the long-term Mann-Kendall trend
-test does confirm a steady year-on-year decline (trend p ≈ 0.02).
-Soil moisture is roughly flat. Relative humidity shows a small
-significant decline.
-
-```{r compare-recipe, eval = FALSE}
-cmp <- cd::cd_compare(ts)   # defaults: 2015–2025 vs 1951–1980, Welch t-test
-```
-
-```{r compare-table, echo = FALSE}
-# Recompute cmp from the cached ts so the p_value column reflects the
-# current cd_compare() defaults — the precomputed rds was generated
-# before #43 added the test argument.
-cmp <- cd::cd_compare(ts)
-trn_p <- trn[trn$trend_start == 1951, c("variable", "period", "mk_pvalue")]
-names(trn_p)[3] <- "trend_p"
-trn_p$trend_p <- round(trn_p$trend_p, 3)
-
-no_pct_vars <- c(
-  "tmean", "tmax", "tmin",
-  # snow_cover and snowfall_fraction are already in % (pct_point_diff
-  # anomaly type) — pct-of-baseline mixes units. snowmelt_doy_50 is a
-  # day-of-year — pct-change of a date is meaningless.
-  "snow_cover", "snowfall_fraction", "snowmelt_doy_50"
-)
-cmp_combined <- cmp |>
-  dplyr::mutate(
-    pct_change = ifelse(
-      variable %in% no_pct_vars,
-      NA_real_,
-      round(100 * (mean_a - mean_b) / mean_b, 1)
-    ),
-    mean_a    = round(mean_a, 2),
-    mean_b    = round(mean_b, 2),
-    abs_diff  = round(difference, 2),
-    window_p  = round(p_value, 3)
-  ) |>
-  dplyr::select(variable, period, mean_a, mean_b, abs_diff, pct_change, window_p) |>
-  merge(trn_p, by = c("variable", "period"), all.x = TRUE)
-
-names(cmp_combined) <- c(
-  "Variable", "Period",
-  "Recent (2015–2025)", "Pre-warming (1951–1980)",
-  "Δ absolute", "Δ %", "Δ p (windows)", "Trend p (75-yr)"
-)
-
-kableExtra::kable_styling(
-  knitr::kable(cmp_combined, label = NA,
-    caption = "Recent decade (2015-2025 mean) compared to pre-warming reference (1951-1980 mean) for the Kootenay Lake Region.",
-    row.names = FALSE),
-  bootstrap_options = c("striped", "hover", "condensed")
-) |>
-  kableExtra::scroll_box(height = "360px")
-```
-
-<br>
-
 ## Spatial Pattern
 
 The zonal mean reduces a region this size to a single number; the spatial
diff --git a/vignettes/peace-fwcp.Rmd b/vignettes/peace-fwcp.Rmd
index ac11e84..6bb5381 100644
--- a/vignettes/peace-fwcp.Rmd
+++ b/vignettes/peace-fwcp.Rmd
@@ -253,6 +253,81 @@ cd::cd_plot_timeseries(
 )
 ```
 
+## Recent Decade vs Pre-Warming Reference
+
+The table below compares two windows directly — the recent decade
+(2015–2025) against the pre-warming reference (1951–1980). Δ
+absolute is the difference of the two means in the variable's native
+units. Δ % is shown only for variables where it is meaningful
+(precipitation, soil moisture, vapour pressure deficit, relative
+humidity). Two p-values are reported, answering different
+questions. **Δ p (windows)** is the Welch t-test on the annual
+values within each window — does the recent decade differ from
+the pre-warming reference? **Trend p (75-yr)** is the Mann-Kendall
+test on the full 1951–present series — is there a steady
+year-on-year ramp? Step changes show up as significant Δ p with
+non-significant trend p; gradual ramps show up as both significant.
+
+The recent decade was 1.7 to 2.4 °C warmer than the pre-warming
+reference for annual mean, daytime maximum, and overnight minimum,
+with both window and trend p-values below 0.001. Vapour pressure
+deficit is up significantly on both tests. Annual precipitation
+was about 3 to 4 % higher in the recent decade; neither test
+confirms the shift (Δ p ≈ 0.4, trend p ≈ 0.20). Soil moisture and
+relative humidity show no meaningful change.
+
+```{r compare-recipe, eval = FALSE}
+cmp <- cd::cd_compare(ts)   # defaults: 2015–2025 vs 1951–1980, Welch t-test
+```
+
+```{r compare-table, echo = FALSE}
+# Recompute cmp from the cached ts so the p_value column reflects the
+# current cd_compare() defaults — the precomputed rds was generated
+# before #43 added the test argument.
+cmp <- cd::cd_compare(ts)
+trn_p <- trn[trn$trend_start == 1951, c("variable", "period", "mk_pvalue")]
+names(trn_p)[3] <- "trend_p"
+trn_p$trend_p <- round(trn_p$trend_p, 3)
+
+no_pct_vars <- c(
+  "tmean", "tmax", "tmin",
+  # snow_cover and snowfall_fraction are already in % (pct_point_diff
+  # anomaly type) — pct-of-baseline mixes units. snowmelt_doy_50 is a
+  # day-of-year — pct-change of a date is meaningless.
+  "snow_cover", "snowfall_fraction", "snowmelt_doy_50"
+)
+cmp_combined <- cmp |>
+  dplyr::mutate(
+    pct_change = ifelse(
+      variable %in% no_pct_vars,
+      NA_real_,
+      round(100 * (mean_a - mean_b) / mean_b, 1)
+    ),
+    mean_a    = round(mean_a, 2),
+    mean_b    = round(mean_b, 2),
+    abs_diff  = round(difference, 2),
+    window_p  = round(p_value, 3)
+  ) |>
+  dplyr::select(variable, period, mean_a, mean_b, abs_diff, pct_change, window_p) |>
+  merge(trn_p, by = c("variable", "period"), all.x = TRUE)
+
+names(cmp_combined) <- c(
+  "Variable", "Period",
+  "Recent (2015–2025)", "Pre-warming (1951–1980)",
+  "Δ absolute", "Δ %", "Δ p (windows)", "Trend p (75-yr)"
+)
+
+kableExtra::kable_styling(
+  knitr::kable(cmp_combined, label = NA,
+    caption = "Recent decade (2015-2025 mean) compared to pre-warming reference (1951-1980 mean) for the FWCP Peace Region.",
+    row.names = FALSE),
+  bootstrap_options = c("striped", "hover", "condensed")
+) |>
+  kableExtra::scroll_box(height = "360px")
+```
+
+<br>
+
 ## Daytime Highs and Overnight Lows
 
 The cd package ships daytime maximum (tmax) and overnight minimum (tmin)
@@ -510,81 +585,6 @@ decade. For aquatic ecosystems downstream, this is a loss of late-
 season cold-water input to streams during the warmest part of the
 year.
 
-## Recent vs Pre-warming
-
-The table below compares two windows directly — the recent decade
-(2015–2025) against the pre-warming reference (1951–1980). Δ
-absolute is the difference of the two means in the variable's native
-units. Δ % is shown only for variables where it is meaningful
-(precipitation, soil moisture, vapour pressure deficit, relative
-humidity). Two p-values are reported, answering different
-questions. **Δ p (windows)** is the Welch t-test on the annual
-values within each window — does the recent decade differ from
-the pre-warming reference? **Trend p (75-yr)** is the Mann-Kendall
-test on the full 1951–present series — is there a steady
-year-on-year ramp? Step changes show up as significant Δ p with
-non-significant trend p; gradual ramps show up as both significant.
-
-The recent decade was 1.7 to 2.4 °C warmer than the pre-warming
-reference for annual mean, daytime maximum, and overnight minimum,
-with both window and trend p-values below 0.001. Vapour pressure
-deficit is up significantly on both tests. Annual precipitation
-was about 3 to 4 % higher in the recent decade; neither test
-confirms the shift (Δ p ≈ 0.4, trend p ≈ 0.20). Soil moisture and
-relative humidity show no meaningful change.
-
-```{r compare-recipe, eval = FALSE}
-cmp <- cd::cd_compare(ts)   # defaults: 2015–2025 vs 1951–1980, Welch t-test
-```
-
-```{r compare-table, echo = FALSE}
-# Recompute cmp from the cached ts so the p_value column reflects the
-# current cd_compare() defaults — the precomputed rds was generated
-# before #43 added the test argument.
-cmp <- cd::cd_compare(ts)
-trn_p <- trn[trn$trend_start == 1951, c("variable", "period", "mk_pvalue")]
-names(trn_p)[3] <- "trend_p"
-trn_p$trend_p <- round(trn_p$trend_p, 3)
-
-no_pct_vars <- c(
-  "tmean", "tmax", "tmin",
-  # snow_cover and snowfall_fraction are already in % (pct_point_diff
-  # anomaly type) — pct-of-baseline mixes units. snowmelt_doy_50 is a
-  # day-of-year — pct-change of a date is meaningless.
-  "snow_cover", "snowfall_fraction", "snowmelt_doy_50"
-)
-cmp_combined <- cmp |>
-  dplyr::mutate(
-    pct_change = ifelse(
-      variable %in% no_pct_vars,
-      NA_real_,
-      round(100 * (mean_a - mean_b) / mean_b, 1)
-    ),
-    mean_a    = round(mean_a, 2),
-    mean_b    = round(mean_b, 2),
-    abs_diff  = round(difference, 2),
-    window_p  = round(p_value, 3)
-  ) |>
-  dplyr::select(variable, period, mean_a, mean_b, abs_diff, pct_change, window_p) |>
-  merge(trn_p, by = c("variable", "period"), all.x = TRUE)
-
-names(cmp_combined) <- c(
-  "Variable", "Period",
-  "Recent (2015–2025)", "Pre-warming (1951–1980)",
-  "Δ absolute", "Δ %", "Δ p (windows)", "Trend p (75-yr)"
-)
-
-kableExtra::kable_styling(
-  knitr::kable(cmp_combined, label = NA,
-    caption = "Recent decade (2015-2025 mean) compared to pre-warming reference (1951-1980 mean) for the FWCP Peace Region.",
-    row.names = FALSE),
-  bootstrap_options = c("striped", "hover", "condensed")
-) |>
-  kableExtra::scroll_box(height = "360px")
-```
-
-<br>
-
 ## Spatial Pattern
 
 The zonal mean reduces a region this size to a single number; the spatial