August 19, 2026

Generated lab notebook from daily_summary, plan_journal, plan_delivery_log, and setpoint audit data. It is intentionally chronological and may include in-progress cycles before validation.

Planner Execution Ledger

Planner checkpoints can acknowledge that the active plan is still suitable without writing a new public plan ID. Those no-change decisions are part of the audit trail.

🌅 Morning Cycle (12:18 AM) — iris-20260819-0016

Reflection

Validating previous cycle: iris-20260818-0617

Previous hypothesis: SUNRISE plan for Tuesday 2026-08-18. Monday closed at score 70.9, above the 62.3 seven-day average, but VPD was the bottleneck at 37.2% versus 50.6% temperature compliance. Heat narrowly dominated at 9.52 h, followed by VPD-high at 9.18 h and VPD-low at 3.94 h; cold was 0. Dew safety held at a 7.8F minimum margin with 0 risk hours. Reliable electric, gas, and total-cost fields remain unavailable, so utility cost cannot be attributed to heating versus cooling; water was 382 gal and mister water 93 gal versus 171 gal seven-day average total water. The completed Monday SUNRISE plan iris-20260817-0615 was graded 3/10 versus deterministic anchor 2. Its forecast captured the 93.6F heat but undercalled the 1193 W/m2 solar peak, and although scoped VPD-high stayed below 6 h with safe dew conditions, 101 of 109 guardrail events were held and 15 writes were missed… Result: Evaluated the completed 00:16-06:21 MDT MIDNIGHT window using its plan-scoped scorecard. Scoped planner score/compliance was 63.5; temperature compliance was 51.1% and VPD compliance 50.5%. Stress was 2.03 h heat, 1.40 h VPD-high, 0.61 h VPD-low, and 0 h cold, with 0 dew-risk hours. The plan correctly preserved dew safety and avoided cold stress, but its hypothesis overweighted pre-dawn VPD-low risk: high-side VPD and heat were the larger scoped burdens. The 09:30 hot/dry fallback never came under this plan because SUNRISE replaced it at 06:21, so it receives no credit. Reliable cost and provenance-gated resource terms were unavailable. Grade 5/10: safe but only moderate corridor performance and a miss on the dominant overnight stress direction. Score: 5/10

Hypothesis

Testing: Use an observed-state overnight wet-off posture, then reopen band-coupled moisture only for the forecast dry ramp; judge success by corridor outcome and clean materialization rather than guardrail repairs. Expected outcome: From 00:15-08:30 MDT, keep dew-risk hours at 0, cold stress at 0, and VPD-low stress below 0.5 h without unnecessary wetting. If the SUNRISE plan is delayed, the 09:30 fallback should have live-edge +0.05/+0.25 kPa moisture authority, 30% duty, 0.18 kPa fog escalation, 45-minute lead, and 300 gal bounded authority in place before corrected VPD rises above the served corridor; target less than 1.5 h controller-attributable VPD-high stress through noon, while accepting residual physics-limited heat.

Setpoints

Wednesday August 19

Primary crop-band changes:

Tactical tunable changes:

Changed secondary parameters:

🌅 Morning Cycle (6:21 AM) — iris-20260819-0618

Reflection

Validating previous cycle: iris-20260819-0016

Previous hypothesis: Tuesday closed at legacy planner score/compliance 64.0, with VPD the weaker axis at 43.8% versus 51.5% temperature compliance. Stress totaled 21.60 h: 10.91 h VPD-high, 9.75 h heat, 0.94 h VPD-low, and 0 h cold; dew safety held at an 8.6F minimum with 0 risk hours. Reliable resource terms are unavailable. The completed sunset plan iris-20260818-1954 is graded 7/10 versus deterministic anchor 5: I credit its excellent scoped 79.1 compliance, near-zero stress, safe 10.2F current dew margin, resolved moisture-gap alerts, and measured VPD recovery from 1.34 to 0.72 kPa. The anchor is appropriately lower because 313 VPD-high guardrail events included 272 held and only 41 dispatched, with three transition writes missed; the outcome was good but requested materialization was not clean… Result: Final SUNRISE evaluation for the governed 06:21-19:57 MDT window, using the plan-scoped scorecard rather than the whole-day summary. Scoped planner score/compliance was 55.6; temperature compliance was 47.1% and VPD compliance 46.2%. Stress was 5.23 h heat, 3.75 h VPD-high, 1.20 h VPD-low, 0 h cold, and 0 dew-risk hours. The predicted high heat and VPD-high windows were directionally correct, and the explicit VPD-high-under-6-hour goal succeeded at 3.75 h. Outdoor peak was close (88.1F actual versus 91F predicted), but solar was materially underforecast (1224 versus 782 W/m2), increasing structural heat pressure. Band-coupled engagement/all-zone moisture and 0.15 fog authority preserved dew safety, but VPD compliance remained weak and the 20 s pulse-gap posture coincided with bidirectional VPD stress. Delivery was not clean: 5 of 9 VPD-high guardrail events were held and only 4 dispatched. Grade 5/10: credit the measurable VPD-high bound and safety under harsher-than-forecast solar, but penalize weak corridor compliance, substantial heat, VPD-low overshoot, and majority-held guardrail delivery. Reliable resource cost terms were unavailable. Score: 5/10

Hypothesis

Testing: Hold one stable, guardrail-aligned +0.05/+0.25 kPa moisture posture through the confirmed hot/dry period after a dry dawn ramp; test whether this reduces held/missed materialization while keeping VPD-high stress below 6 hours and dew-risk at zero. Expected outcome: Plan-window VPD-high stress below 6 hours, VPD-low stress below 0.75 hours, zero dew-risk hours, and controller-attributable heat stress below 2 hours while accepting additional physically-unachievable heat. During dual-axis misses, action proof should show wet assist available rather than blocked; by midnight measured VPD should be at or within 0.2 kPa above the active high edge. Audit should materially improve from the prior 28 held of 38 guardrail events.

Conditions

Cool saturated dawn followed by a variable-cloud hot/dry ramp. Raw VPD peaks near 3.88 kPa; bias correction still leaves severe dry pressure. Cloud cover ranges from 0% to 100%, so escalation is demand-coupled rather than based on solar alone.

Expected stress windows

Parameter rationale

Setpoints

Wednesday August 19

Primary crop-band changes:

Tactical tunable changes:

Changed secondary parameters:

🌆 Evening Cycle (7:57 PM) — iris-20260819-1953

Reflection

Validating previous cycle: iris-20260819-0618

Previous hypothesis: SUNRISE plan for Wednesday 2026-08-19. Tuesday closed at planner score/compliance 64.0, slightly above the 62.4 seven-day average. VPD was the bottleneck at 43.8% versus 51.5% temperature compliance. VPD-high dominated at 10.91 h, narrowly above 9.75 h heat; VPD-low was 0.94 h and cold was 0 h. Dew safety held at an 8.6F minimum margin with 0 risk hours. Reliable cost, electric, gas, and mister-only water fields are unavailable; total water was about 297 gal versus a 201 gal seven-day average, so cooling/misting demand was elevated but utility-cost attribution is not possible. The completed 2026-08-18 SUNRISE plan iris-20260818-0617 was graded 4/10 versus deterministic anchor 2. I gave credit because actual outdoor peak temperature was close to forecast, the predicted heat/VPD-high pattern occurred, VPD-high stayed under the stated 7-hour plan-window bound at 5.35 h, and dew safety held… Result: Final MIDNIGHT evaluation of the elapsed 19:57-00:17 MDT SUNSET window. Scoped score was 33.9, temperature compliance 40.8%, VPD compliance 28.3%, with 0.01 h heat, 0.02 h VPD-high, 0 h VPD-low/cold, and 0 dew-risk hours. Current VPD remained 1.19 kPa, about 0.25 kPa above the active 0.94 kPa high edge, so the required evening recovery target was missed and an open VENTILATE moisture-gap alert remained. The pre-dawn and morning fallback goals were unobserved. Delivery was overwhelmingly guardrail-dependent: 250 of 276 VPD-high events were held and only 26 dispatched; conservative clock-based requests did not cleanly materialize. Final grade 3/10 versus anchor 1, with limited credit only for dew/cold safety and absence of low-side overshoot. Score: 3/10

New finding: At MIDNIGHT, if measured VPD is still above the active high edge and dew margin is healthy, do not jump directly to a conservative zero-wet or long-gap pre-dawn posture merely because a humid trough is forecast. First use a short, band-coupled live-edge recovery bridge, then taper only after measured VPD clears the high edge; otherwise sticky VPD-high guardrails will hold most writes and the plan will miss recovery despite safe dew conditions. → Added to Lessons Learned

Hypothesis

Testing: Use one stable two-step evening taper, then a zero-wet pre-dawn posture; test whether this limits overnight VPD-low stress without extending the current VPD-high miss or triggering guardrail-held churn. Expected outcome: By 00:30, measured house VPD is inside or no more than 0.15 kPa above the active high edge. From 00:30-07:30, VPD-low stress stays below 1.0 h, VPD-high stress below 0.75 h, cold stress at zero, dew-risk at zero, and no heater-vent whipsaw is observed. If SUNRISE replacement is delayed, wet assist becomes available on live morning demand before the corrected severe dry ramp, without pre-wetting at dawn.

Conditions

Dry sunset, then a 59F/72% RH pre-dawn trough, followed by a rapid clear extreme hot-dry ramp. Bias correction reduces but does not remove the severe daytime pressure.

Expected stress windows

Parameter rationale

Setpoints

Wednesday August 19

Primary crop-band changes:

Tactical tunable changes:

Thursday August 20

Primary crop-band changes:

Tactical tunable changes:

Changed secondary parameters:

End-of-Day Summary

Climate

Interior Light Evidence

• Crop DLI: unavailable
• Reason: source_reading_missing
• Provenance: legacy_invalid_exterior_proxy_plus_fixture_estimate
• Validity: dli-validity-v1; 2024-01-01T00:00:00+00:00 → open
• Qualified-light minutes and photoperiod remain independent of unavailable DLI.

Stress Hours

• Heat stress (>85°F): 8.0h
• VPD stress (>2.0 kPa): 11.6h
• Cold stress (<55°F): 7.3h

Economics

Equipment Runtimes

Crop Health (Gemini Vision)

Hourly Pattern

7-Day Stress Context