Year rehearsal report

Synthetic practice, not a completed field study. Actual students: 0. Real tree observations: 0. The weather is recorded; plant states and practice records are generated. The model is uncalibrated, and this rehearsal does not demonstrate learning gains. Status: proposed. This worked practice covers 365 days of 2025 and 20 generated practice dates.

Change temperature. Keep the comparison fair.

Add 2 °C to each day’s low and high, including the prior autumn’s initialization. Keep precipitation, dates and teaching parameters fixed.

Results from the same proposed teaching model, not measurements of a red maple.
Model result	Historical baseline	+2 °C
Bud-opening threshold	Mar 16	Mar 11
Autumn short-day trigger	Sep 21	Sep 21
Year-average root water	68.41 mm	66.63 mm

The bud threshold arrives five days earlier. The autumn trigger stays on the same date because day length did not change. A warmer year does not move every process in the same way.

Why return after rain?

On Aug 16, modeled root water is 9.22 mm. After 36.3 mm of recorded rain on Aug 17, it rises to 35.48 mm, but the delayed stress index falls only from 0.769 to 0.679. On Aug 20 it is still 0.529.

The model’s canopy remains full across those days. Internal modeled change can precede an obvious appearance change. This is a question to investigate; no real tree recovery or water status was measured.

Keep dates. Keep the gaps.

Twelve monthly examples and eight event dates retain three predictions and five explanation revisions. July includes a deliberate summer handover gap: “not observed” stays missing, rather than becoming “no.”

Weather columns are monthly station summaries. Model columns show the 15th of each month; all plant records are synthetic practice.
Month	Temperature midpoint	Precipitation	Model canopy	Model root water baseline → +2 °C
Jan	-0.9 °C	23.7 mm	0%	119.4 → 119.3 mm
Feb	2.2 °C	52.2 mm	0%	119.8 → 119.8 mm
Mar	9.7 °C	136.2 mm	0%	115.1 → 113.1 mm
Apr	13.7 °C	72.0 mm	100%	107.9 → 107.1 mm
May	18.5 °C	158.1 mm	100%	50.7 → 48.2 mm
Jun	24.1 °C	56.0 mm	100%	38.8 → 34.0 mm
JulPractice gap	27.7 °C	117.0 mm	100%	38.5 → 36.2 mm
Aug	23.8 °C	43.9 mm	100%	10.1 → 9.2 mm
Sep	22.8 °C	45.5 mm	100%	12.0 → 11.1 mm
Oct	15.7 °C	51.4 mm	94%	24.0 → 22.9 mm
Nov	8.9 °C	58.5 mm	0%	58.5 → 55.5 mm
Dec	1.7 °C	85.2 mm	0%	118.6 → 119.8 mm

Keep four types of evidence separate

Direct tree observation: None collected. Synthetic cues do not fill this category.
External source: Pinned NOAA station weather with provenance.
Model output: Reproducible spring, root-water and autumn trajectories with stated assumptions.
Untested hypothesis: Labeled synthetic predictions; no real learner reasoning was collected.

A single year cannot establish a climate trend. A real visit in a different year cannot validate this 2025 counterfactual.

Ready to explore. Fieldwork comes next.

This can support a teacher-led prototype review: inspect slow change, compare one driver and preserve an explanation. It is not yet a complete year-long classroom system or evidence that students learned more.

Real field basis: one teacher-approved tree, same-view photographs and matching-year records.
Continuity: a durable class archive, revisit reminders and an actual summer handover.
Shared reasoning: peer review and cards that distinguish observation, source, model and hypothesis.
Validation: species parameters, multiple years and an independent new case without AI.

See all 12 readiness checks

Teacher-approved tree · missing
Teacher verifies one safe local tree and a persistent alias before field use.
Real photos and a protected site reference · missing
2–3 real, consented same-view photos and teacher-approved site reference; protect precise locations.
Matching-year field evidence · missing
Use real observations with matching calendar weather; 2026 visits do not validate a 2025 counterfactual.
Species and site validation · missing
Fit and held-out validate a stated species/site event; flowering/fruit are currently omitted.
Revisit reminders · missing
Opt-in reminders and teacher scheduling with a demonstrated delivery path.
Durable classroom archive · partial
Browser JSON works, but annual classroom needs durable archive, backup owner and cross-device recovery.
Peer review and evidence cards · missing
Persistent card with four evidence types, actual peer comment and append-only revision.
Questions after learner reasoning · partial
Existing separate inquiry AI does not yet scaffold annual records after their own reasoning; no calls were made.
Multiple years and independent transfer · missing
A comparable second year and independent unfamiliar case without AI.
Current and generated weather · missing
Pinned historical replay exists; current-weather service and validated random generator do not.
Weekly visits and summer handover · partial
This rehearsal covers monthly/event examples; actual weekly visits and summer handover require people and records.
Real learner evaluation · not collected
Real learner reasoning, usability, pre/post or delayed transfer evidence; not clicks or synthetic revisions.

Flowering, fruit, wind injury and frost injury are not implemented. Browser downloads support portability; they are not automatic classroom sync. Card presentation and its learning benefits still need testing.

Try the example in a separate notebook.

Downloading does not save or import anything. Import only in the separate practice notebook. Do not use these generated visits as real tree evidence; no photos, GPS records or actual learner data are included.

Download practice JSON Open the practice notebook at Restore from a file

Data and reproduction details

NOAA GHCN-Daily source ↗ (opens in a new tab)

Philadelphia International Airport, USW00013739. A regional reference, not a tree-site measurement. Temperature midpoint is derived from the daily low and high, not measured daily mean temperature.

Model 6139-year-0.1.0-proposed. 92 warm-up days. No LLM calls. The downloaded file retains weather, assumptions and trajectories for offline reproduction.

Download SHA-256: ae107bf0a93d947e06d3e1fc23ddccbd368e59d809fd9f2d7c677b5cf23a6364