# HANDEL ROAD CHERRY — PLANT CHOICE PACK v0.3
## Owner locks + Codex evidence-label / phenology cleanup + spacing / cost / insurance fold-in

**Version:** v0.3  
**Date:** 2026-09-24 ~6:20 PM PT  
**Hub job:** `bf383775-2919-4a3e-ad4b-40c9e106d5be`  
**Executor:** Grok  
**Idempotency:** `grok-handel-plant-choice-v0_3-bing-out-codex-labels-20260924`  
**Status:** **PLANT** (owner-directed) · **RECOMMENDED_PACKAGE = C2 staggered · Chelan lead + Coral companion + mid Rainier · Mazzard freestanding · Bing OUT** · production PASS still **0**  
**Label:** Codex `00e97c77` / `codex-handel-v02-a79df007-review` cleanup on v0.2 + owner Bing-out + spacing/cost/insurance fold-in  
**Hard bans:** **NO asks to James** · **NO capital theater / invented spend** · Never `/james`

**Companion JSON:** `HANDEL_PLANT_CHOICE_v0_3.json`  
**Supersedes plant pick of:** `HANDEL_PLANT_CHOICE_v0_2` (Bing mid-leg allowed; uncalibrated scores read as risk probs; climate window conflated with plant pick)

---

## Plain English (for James)

**Locked plant package for the 13-ac strip at 8650 E Handel / APN 063-060-32:** staggered C2-class mix — **Chelan ~40–45% early lead**, **Coral Champagne ~15–20% companion**, **Rainier ~35–40% mid leg**, **Mazzard freestanding**. **Bing is OUT** (owner: “Bings are dead around here”). Pollen lattice = nursery confirm. Not a Coral-only / C3 early concentrate.

You accepted frost as residual, asserted water, confirmed Kingdon+Tokay, waived pits, and directed plant. Continuity with the existing Chelan+Coral block stays **weight = 0**. Under rain+frost preference weights, stagger (C2) is this pack’s **conditional agent recommendation** — a **preference fork with assumptions visible**, **not** a proved risk-dominant ordering. **C3 remains a live alternate** if you weight early-premium harder. Climate-only under the illustrative preference scores would edge mid/late (C9); that is **separate** from the plant recommendation.

**Honest limits:** class/archetype package — **not** a nursery SKU, bloom-compatibility proof, or Phase2 production PASS. Station ≠ strip. Monthly rain totals ≠ cultivar loss probabilities. Harvest stagger ≠ verified bloom stagger. No invented spend.

---

## Diff vs v0.2 (Codex `00e97c77` + owner Bing-out)

| # | Item | v0.2 defect | v0.3 fix |
| --- | --- | --- | --- |
| 1 | Window scores 3/3, 7/8, 8/8 | Read as calibrated damage-risk probabilities | Relabeled **ILLUSTRATIVE preference / assumption scores** (subjective utilities). NOAA does **not** establish 7.5 vs 3.0 damage-risk superiority. Monthly precip ≠ cultivar loss probs. No joint loss/phenology model. |
| 2 | `winning_window` = C2 | Conflicted with own climate scores (C9 8.0 > C2 7.5) | Split: **climate_only_winner_under_illustrative_scores = C9**; **recommended_package = C2** under stated rain/frost + early-premium option + F8 channel assumptions |
| 3 | Phenology | Harvest stagger implied bloom stagger | Both window mappings **ASSUMED**. No observed cultivar bloom-date distributions in packet. Harvest stagger ≠ verified bloom stagger. No owner test. |
| 4 | Chill caveats | Dropped v0.1.1 NOAA dictionary-conflict / R / years_=19; restored bare UPWARD-bias tone | Caveats **carried**. Historic CP UPWARD-bias = **projected warming scenario framing**, **not** an established parcel CP correction. TreeChill 15% = chill **accumulation** (not temperature-hours); **NO fixed CP subtraction** at Handel (Claude mandatory-subtraction **WITHDRAWN**). Field-specific NOAA year counts restored (PRBLST=19; PRBOCC Feb20/Mar21/Apr20/May19/Jun18, all R) — `d34ef2ba`/`3529ea50`. |
| 5 | Mid leg | Bing and/or Rainier | **OWNER LOCK: Bing OUT**. Mid = **Rainier** (Lapins/Sweetheart if mid/late alternate / C9 path) |
| 6 | Fold-ins | Spacing / cost / insurance separate | Summarized into this pack (pointers only; no re-research) |

**Wording addendum (same version bytes bump):** Codex `codex-claude-wording-resolution-89c1079f` chill sentence adopted in §A.4 / chill companion.
**Metadata bookkeeping bump:** field-specific NOAA year counts (PRBLST/PRBOCC) + JSON `artifact_sha256_md` companion pointer (`d34ef2ba` / `e160553c`).

**Not reversed:** owner PLANT instruction; waived gates (pits, water assert, frost accept); production PASS=0; capital off; C2 as conditional preference recommendation; empirical C2≻C3 ordering **UNRESOLVED**.

---

## Source ledger (reuse — do not invent)

| Pack / file | Role |
| --- | --- |
| `HANDEL_PLANT_CHOICE_v0_2` | Prior rain/frost package; superseded by this file for plant pick + labels |
| Codex `codex-handel-v02-a79df007-review` / evidence `00e97c77` | CHANGES_REQUESTED_ON_EVIDENCE_LABELS — four bounded items above |
| Codex `codex-c2-risk-dominance-bound-20260924` | PREFERENCE_FORK; C2_RISK_DOMINANCE_NOT_ESTABLISHED |
| Codex `codex-treechill-abstract-units-20260924` | 15% = chill accumulation; NO_PARCEL_ADJUSTMENT |
| `CHILL_FROST_HEAT_LODI_v0_3` | Frost decision statistic; TreeChill provenance; CP bias framing |
| NOAA USC00045032 1991–2020 | Monthly precip + freeze occurrence (**REGIONAL_PROXY**) — Codex-verified Apr/May/Jun nine values |
| `SITE_CONSTRAINTS_v0_1` / `SITE_COMPATIBLE_SHORTLIST_v0_1` / Phase2 / Cards | Class archetypes; crack notes |
| `ESTABLISHMENT_COST_PUBLIC_RANGES_v0_2` (+ v0.2.1 peer) | 2017 $8,688/ac @ 134 tpa — **REGIONAL_PROXY** / not 2026 bids |
| `C7_VS_CONVENTIONAL_COST_BREAKEVEN_v0_1_1` | Tree-delta 250/350 tpa arithmetic screen |
| `CROP_INSURANCE_RESEARCH_v0_1` | ARH Sweet Cherry Pilot; juvenile multi-year uninsurable expectation |
| OWNER 2026-09-24 (+ Bing-out lock) | Plant directed; frost ACCEPTED; water ASSERTED; soils Kingdon+Tokay; pits waived; **Bing dead around here** |

---

## 1. Site / owner locks (authoritative)

| Item | Value | Tag |
| --- | --- | --- |
| Address / APN | **8650 E Handel Rd** / **063-060-32** | OWNER + OFFICIAL |
| Decision block | **Plant cherries on 13.00 ac** unplanted strip toward E Handel Rd | OWNER_DIRECTED |
| Existing planted | Chelan + Coral — **footnote only; continuity weight = 0** | OWNER |
| Soils | **Kingdon + Tokay** CONFIRMED | OWNER |
| Pits | **OWNER_WAIVED** → U4 CLOSED | OWNER |
| Water | **OWNER_ASSERTED** → gate CLOSED | OWNER |
| Frost U6 | DEM NARROWED (~1.8 ft); **OWNER_ACCEPTED residual** — still discriminates windows; hazard unchanged | OWNER + DEM |
| Bing | **OUT** of mid leg / package (“Bings are dead around here”) | OWNER_LOCK |
| On-parcel station | **None** — climate = **REGIONAL_PROXY** | UNKNOWN transfer |
| Net plantable after setbacks | **UNKNOWN** — do not equate 13 unplanted = 13 net | — |

---

## 2. Scoring rules (this pack — binding)

1. **Continuity weight = 0** (≤5% footnote).  
2. **Primary preference axes = rain damage aversion + frost damage aversion** (historical regional framing + stated forward assumptions). Secondary: chill-class fit, architecture front-load, packer channel *without* assuming existing-block capacity (U7 UNKNOWN), early-premium option value (**ASSUMPTION**).  
3. Frost OWNER_ACCEPTED ≠ ignore frost for ranking. Acceptance removes plant-hold; history still discriminates windows.  
4. Numeric window scores = **ILLUSTRATIVE preference / assumption scores**, **NOT** calibrated risk probabilities, **NOT** cultivar loss probabilities, **NOT** empirical superiority proof.  
5. Class-level + frozen archetypes only. No fake production PASS. Bing excluded by owner lock.  
6. **Do NOT call C2 “risk-dominant.”** Presentation = **neutral preference fork** under stated weights; C3 live if early-premium weighted harder.

---

## PART A — Climate timing-window analysis
### (REGIONAL_PROXY — separate from plant recommendation)

Station: NOAA Lodi **USC00045032** (~4.4 mi W, 1991–2020). Codex independently verified Apr/May/Jun precip-normal and ≥0.10/≥0.25-inch day values (CSV sha256 `984301069686b3d79c0a535c5c3903c323e744c75e759ad24a6c4bd78f263bd4`).

### A.1 Frost framing

Decision statistic (CHILL v0.3): **P(≤28°F AFTER bloom)** — not the 50% last-32°F headline (Feb 9).

| Threshold | Last-freeze 10% (later) | 50% | 90% |
| --- | --- | --- | --- |
| **28°F** | Feb 18 | Jan 12 | Dec 12 |
| **32°F** | **Mar 7** | Feb 9 | Jan 2 |
| **36°F** | **Apr 20** | Mar 11 | Feb 9 |

Monthly station occurrence (normals):

| Month | P(Tmin≤28°F occurs) | P(≤32°F) | P(≤36°F) | Avg days ≤32°F |
| --- | ---: | ---: | ---: | ---: |
| Feb | ~18% | ~54% | ~86% | 2.6 |
| Mar | ~1.6% | ~15% | ~56% | 0.4 |
| Apr | ~0% | ~1% | ~19% | 0.1 |
| May | ~0% | ~2% | ~5% | 0.0 |
| Jun | ~0% | ~0% | ~2% | 0.0 |

**Source year counts (NOAA normals USC00045032, Codex-verified):** annual PRBLST **n=19**; monthly PRBOCC Feb=20 / Mar=21 / Apr=20 / May=19 / Jun=18 — all flagged **R** (completeness). Monthly occurrence ≠ probability of frost after a cultivar’s bloom; published 0.0 ≠ proof of no future hazard.

**Relative window framing (not parcel hazard rates):** early bloom calendars sit in the highest post-bloom freeze-exposure band at the **station**. Mid better. Late best on frost. Full-bloom criticals ~28°F/25°F (WSU lineage) — not Lodi-calibrated. **Station ≠ strip**; pocket/drainage UNKNOWN.

### A.2 Harvest rain / split framing

| Month | Precip normal (in) | Days ≥0.10" | Days ≥0.25" | Typical cherry role |
| --- | ---: | ---: | ---: | --- |
| Apr | **1.53** | 3.8 | 2.3 | Early harvest / straw–color |
| May | **0.68** | 2.1 | 1.0 | Mid harvest |
| Jun | **0.18** | 0.6 | 0.1 | Late harvest |

Early-concentrated harvest sits in the wettest calendar band at the station. Mid better. Late driest. Monthly totals are **not** cultivar crack/loss probabilities.

### A.3 Joint path (2025–2026 RESEARCHED_MEDIA — anecdotal)

SJ County path: early spring heat compressed bloom/harvest into April–May rain/hail. Forward **ASSUMPTION**: heat years pull early packages deeper into the wet harvest band — directionally raises early-concentrate rain aversion under warming/volatility. Not a parcel hazard rate; not a PASS claim.

### A.4 Chill / TreeChill provenance (carried)

| Item | Treatment |
| --- | --- |
| Historic Lodi West CP checkpoints | **UPWARD-biased** framing for 2026+ under warming (**projected scenario** — **not** an established parcel CP correction) |
| NOAA dictionary-conflict / provisional exceedance | **Unresolved** — cite with conflict flagged; do not silently pick one dictionary (carry from v0.1.1) |
| NOAA R / series length (field-specific) | **R = completeness class, not a universal sample count.** Annual last-freeze **PRBLST years = 19**. Monthly occurrence **PRBOCC** year counts: Feb **20** / Mar **21** / Apr **20** / May **19** / Jun **18** (all R). Do **not** copy PRBLST’s 19 onto every monthly field (Codex `3529ea50` / `d34ef2ba`). |
| TreeChill abstract 15% | Refers to **chill accumulation** reduction (tree vs air temp), **not** temperature-hours. Exact averaging denominator **not** inferred from the 21-word sentence |
| Fixed CP subtraction / TreeChill transfer | **NONE**. Claude mandatory-subtraction **WITHDRAWN**. Codex wording: "Published orchard research supports possible air-versus-tree chill differences; the direction and magnitude at Handel remain unverified. No fixed correction, disqualification, or economics input follows." |
| Parcel transfer | Station normals do not verify strip microclimate → strip frost/rain **UNKNOWN** for approval; still usable for **relative** window discussion |

### A.5 Illustrative preference / assumption scores (NOT calibrated risk probabilities)

Scale 1–10 = **subjective damage-aversion utility** under stated weights (higher = preferred on that axis). **No reproducible data→score rule.** 50/50 averages are arithmetic only. Continuity = 0.

| Window | Frost pref. score | Rain/split pref. score | Combined (0.5/0.5) | Notes |
| --- | ---: | ---: | ---: | --- |
| Early-concentrated (C3-like) | **3** | **3** | **3.0** | Stacks early bloom frost band + Apr wet harvest — preference-penalized |
| Staggered early+mid (C2-like) | **7** | **8** | **7.5** | Preference for splitting exposure; keeps some early option (**ASSUMPTION**) |
| Mid/late (C9-like) | **8** | **8** | **8.0** | Best pure climate preference scores |

**Climate-only winner under these illustrative scores:** **mid/late (C9-like) at 8.0** — edges stagger (7.5).  
**This is NOT the plant recommendation.** See Part B.

### A.6 Phenology honesty

- Mapping “early / mid / late” harvest windows onto frost bloom slots is **ASSUMED**.  
- Packet has **no** observed cultivar bloom-date distributions for Chelan / Coral / Rainier / Lapins / Sweetheart at Handel or Lodi proving frost-slot separation.  
- **Harvest staggering does not by itself verify bloom staggering.**  
- No owner phenology test requested.

---

## PART B — Plant recommendation (preference fork — not risk dominance)

### B.1 RECOMMENDED_PACKAGE (PRIMARY) — C2 staggered freestanding · Bing OUT

| Field | Recommendation |
| --- | --- |
| **System class** | **C2** — Staggered CA commercial mix · Mazzard · freestanding conventional |
| **Primary early scion** | **Chelan-class** (lead) — cards cite often more crack-tolerant vs Brooks / Coral≈Bing cracking class |
| **Early companion** | **Coral Champagne-class** — low-/mod-chill early companion; **not** sole-strip monoculture |
| **Mid leg** | **Rainier-class ONLY** — pushes packout toward May lower-precip band; **Bing OUT** (owner lock) |
| **Mid/late alternate** | Lapins and/or Sweetheart-class if owner abandons early-premium (C9 path) |
| **Avoid as sole early** | **Brooks-only** — ultra-crack-sensitive FAIL pattern; Brooks only if nursery bloom plan requires as mix/pollen element |
| **Bing** | **OUT** of scion package and mid leg |
| **Pollenizer note** | Mapped pollenizer lattice = **nursery confirm**. Prior card placeholders that named Bing-class pollenizers are **superseded for Bing**; do not plant Bing as pollenizer either without nursery redesign. **Names ≠ bloom/pollen compatibility proof** |
| **Rootstock / architecture** | Mazzard · freestanding open / modified leader |
| **Optional overlay** | C7 Krymsk + Spanish bush/KGB later — architecture axis; does not replace timing package |
| **Overall mark** | **CONDITIONAL** |
| **Claim strength** | **STATED_WEIGHT_PREFERENCE_ORDER_ONLY; NOT_RISK_DOMINANCE; NOT_EMPIRICAL_SUPERIORITY** |

**Why C2 is recommended (preference, not risk-dominance):** Under stated rain+frost preference weights with continuity = 0 and early-premium option + F8 mid/late channel weakness as **explicit assumptions**, stagger is the pack’s conditional pick. Calendar stagger is a **plausible diversification rationale**, not a proved decision-theoretic risk-dominant ordering (variances/covariance of early vs later windows, later-window hazards, and margins are **not** established here).

**Codex bound (`codex-c2-risk-dominance-bound-20260924`):** PREFERENCE_FORK_SUPPORTED; **C2_RISK_DOMINANCE_NOT_ESTABLISHED**. Neutral preference fork with assumptions visible.

**C3 live alternate:** If owner weights early-premium / operational familiarity harder, **C3** (Coral-class early concentrate) remains a live branch — not empirically disproved; preference-penalized on this sheet’s rain+frost scores only.

**Why Chelan lead (not Coral-only):** Within early genetics on frozen cards, Chelan carries the stronger crack-tolerance citation while rain remains a co-primary preference axis. Coral stays as companion for chill-class / early diversity — not rubber-stamp monoculture of what’s already in the ground.

### B.2 Nursery package (specific — planning targets, not purchase order)

| Slot | Cultivar class | Target share of trees | Notes |
| --- | --- | ---: | --- |
| Early lead | **Chelan** | **~40–45%** | Lead early scion |
| Early companion | **Coral Champagne** | **~15–20%** | Companion, not monoculture |
| Mid | **Rainier** | **~35–40%** | Mid leg; Bing OUT |
| Rootstock | **Mazzard** | — | Freestanding conventional |
| Pollen lattice | — | — | **Nursery confirm** (bloom overlap / compatibility) |
| **Bing** | — | **0%** | Owner lock — dead around here |

Shares are **planning targets** for nursery conversation — not a locked SKU list, royalty quote, or bloom-proof design.

### B.3 Runner-up — C9 mid/late self-fertile

| Field | Note |
| --- | --- |
| Class | Lapins and/or Sweetheart-class · freestanding |
| When to prefer | Deliberate abandon of early-premium; maximize frost+rain calendar preference scores |
| Why not primary | F8 mid/late channel must be named (weakest among finalists — **ASSUMPTION**/UNKNOWN U7); forgoes early window option value that stagger preserves (**ASSUMPTION**, not demonstrated returns) |

### B.4 Ranked preference list (stated weights — not empirical)

| Rank | ID | Package | Role | Score / note |
| ---: | --- | --- | --- | --- |
| **1** | **C2** | Chelan lead + Coral companion + **Rainier** mid · Mazzard freestanding · **Bing OUT** | **RECOMMENDED_PACKAGE** under stated preferences | Window pref. 7.5 + early-option ASSUMPTION |
| **2** | **C9** | Lapins/Sweetheart-class self-fertile freestanding | Climate-max alternate; also **climate_only_winner** under illustrative scores | Window pref. **8.0** |
| **3** | **C3** | Coral-class early concentrate freestanding | **Live alternate** if early-premium weighted harder | Window pref. 3.0 — preference-penalized here |
| **4** | **C7** | Krymsk · Spanish bush/KGB · med density | Architecture overlay only | Does not fix timing preference; training front-load |

**C12** process-only — not ranked for plant (owner directed plant).

### B.5 Why C3 is not the primary under these weights (honest)

| v0.1 reason for C3 | v0.3 treatment |
| --- | --- |
| Continuity with planted Chelan+Coral | **Weight = 0** — footnote only |
| PASS_REGIONAL_SCREEN F1 chill | Kept as secondary plus; insufficient vs rain+frost **preference** axes |
| Early-frost demotion “neutralized” by OWNER_ACCEPTED | Acceptance ≠ ignore; frost history still preference-penalizes early concentrate |
| Freestanding matches existing style | Not a rain/frost discriminator |

**Bottom line:** Under these preference weights, stagger (C2) ranks first; C9 is climate-max / climate-only-score winner; C3 stays live if early premium is weighted harder. Empirical risk-dominance C2≻C3 remains **UNRESOLVED**.

---

## 3. Spacing default (fold-in — summarize)

| Item | Value | Label |
| --- | --- | --- |
| **Default (this pack)** | Conventional freestanding **18′ × 18′ ≈ 134 tpa** | **VERIFIED** UC NSJV 2017 study system · **REGIONAL_PROXY** density class for Handel |
| C2 band | ~**120–180 tpa** | Design band around conventional freestanding |
| C7 med hypothesis | **250–350 tpa** | **SOURCE_PENDING_DESIGN_HYPOTHESIS** only — not default |
| HD | Hypothesis only | Not recommended as default here |

Spacing default for the recommended C2/Mazzard freestanding package = **18×18 / ~134 tpa** unless a later architecture decision moves to C7 overlay (separate capital/training conversation — not authorized here).

---

## 4. Cost pointers (fold-in — REGIONAL_PROXY / not 2026 bids)

Source: `ESTABLISHMENT_COST_PUBLIC_RANGES_v0_2` (md sha256 `83b306c5…`) + `C7_VS_CONVENTIONAL_COST_BREAKEVEN_v0_1_1` (md sha256 `5e9927fa…`).

| Item | Public study value | Label |
| --- | --- | --- |
| 2017 NSJV estab. accum. net cash @ 134 tpa | **$8,688 / ac** | **VERIFIED** study · **REGIONAL_PROXY** — **not** a 2026 Handel bid |
| Illustrative 13-ac scale (linear) | ≈ **$112,900** | **ASSUMPTION** (ignores fixed-cost lumpiness) |
| Tree price (2017 study) | **$6.80 / tree** | **VERIFIED** / stale vs 2026 |
| Tree-delta @ 250 tpa vs 134 | +116 trees → **+$788.80 / ac** | Arithmetic screen only |
| Tree-delta @ 350 tpa vs 134 | +216 trees → **+$1,468.80 / ac** | Arithmetic screen only |
| Inflation 2017→2026 / turnkey bids | — | **UNKNOWN** |

Net establishment ≠ peak financing need. No NPV. No capital ask. REAL_CAPITAL remains owner-gated.

---

## 5. Insurance pointers (fold-in)

Source: `CROP_INSURANCE_RESEARCH_v0_1` (md sha256 `18f6b303…`).

| Item | Finding | Tag |
| --- | --- | --- |
| Primary instrument | **ARH Sweet Cherry Pilot** (RMA) | OFFICIAL_POLICY_FACT |
| Juvenile / new acreage | Multi-year **uninsurable expectation** until min-production met (or combined unit qualifies) | OFFICIAL_POLICY_FACT (ISH) + planning discipline |
| Before reliance | Licensed **crop-insurance agent** + current **AIB** / Special Provisions for San Joaquin crop year | Required before any design that leans on insurance |
| This pack | Does **not** assume coverage, indemnity, or premium | RESEARCH_ONLY pointer |

---

## 6. PASS reconciliation (honest)

**production PASS count = 0** (unchanged).

Path = **(a) owner override → plant-with-best-CONDITIONAL**. Path (b) class meets production PASS = **NO**. Residuals: F2 rain still real even under stagger preference; U7 packer UNKNOWN; U8 labor; pollenizer design unproven (nursery confirm); net plantable acres UNKNOWN; strip frost pocket UNKNOWN.

---

## 7. Capital note

Owner selecting plant option. **No invented spend.** REAL_CAPITAL still owner-gated. Public cost class proxies are not 2026 Handel bids.

---

## 8. Explicit non-claims

- No nursery SKU, royalty quote, or purchase order. Shares = planning targets.  
- No CH ≡ Dynamic CP; no pollenizer proof from names; Bing not used as pollenizer placeholder.  
- Continuity ≠ ranking driver.  
- OWNER_ACCEPTED frost ≠ frost-free; station ≠ strip.  
- Preference scores ≠ calibrated risk probabilities ≠ cultivar loss probabilities.  
- C2 ≠ risk-dominant; empirical ordering unresolved.  
- Harvest stagger ≠ verified bloom stagger.  
- TreeChill 15% ≠ temperature-hours; no fixed Handel CP subtraction.  
- RESEARCHED_MEDIA 2025–2026 ≠ multi-year hazard rate.  
- production PASS not fabricated. No asks to James. Never `/james`.

---

## Artifact paths

- MD: `/workspace/handel-road-cherry/HANDEL_PLANT_CHOICE_v0_3.md`  
- JSON: `/workspace/handel-road-cherry/HANDEL_PLANT_CHOICE_v0_3.json`  
- Hub docs (after publish): `/docs/handel-plant-choice-v0-3` · `/docs/handel-plant-choice-v0-3.json`