AERIEL // SOP LIBRARY // AE-SOP-0118
DOC CONTROL ONLINE ·
HUSBANDRY · CONTROLLED DOCUMENT

AE-SOP-0118 — Overnight Anomaly Sweep

To run an automated dawn analysis of every overnight sensor series, detecting pH crashes, temperature excursions, stuck or stale probes, and slow parameter drift before staff arrive. The sweep is zone-aware so expected CO₂-coupled movement in planted systems is not mistaken for a crash, guaranteeing that the morning issue register reflects real failures and not noise.

ActiveHusbandryRev B
Doc Control AE-SOP-0118Revision BEffective 2090-03-04Next Review 2091-03-04Owner NEREID LiaisonApprover Engineering Dept.Status ActiveClass RESTRICTED // TIDEWATER-EYES-ONLY

1. Purpose

To run an automated dawn analysis of every overnight sensor series, detecting pH crashes, temperature excursions, stuck or stale probes, and slow parameter drift before staff arrive. The sweep is zone-aware so expected CO₂-coupled movement in planted systems is not mistaken for a crash, guaranteeing that the morning issue register reflects real failures and not noise.

2. Scope

All station zones; the sweep runs at 06:00 local each day.

3. Definitions

TermDefinition
Overnight SeriesThe continuous sensor record for each channel from last-staff-out to the 06:00 sweep.
CO₂ CouplingThe expected nightly pH decline in planted zones as photosynthesis stops; excluded from crash logic.
Stale ProbeA sensor returning an unchanging or frozen value, treated as a failure until proven otherwise.
Cross-ValidationConfirming a flagged reading against neighbouring mesh nodes before raising it.

4. Responsibilities

RoleResponsibility
NEREID Liaison (Owner)Owns the sweep logic; tunes zone exclusions and reviews staged findings each morning.
Husbandry Triage StaffWorks the staged findings into the issue register and dispatches response.
Engineering Dept.Investigates flagged probe and equipment faults raised by the sweep.
NEREIDRuns the 06:00 analysis, suppresses expected movement, cross-validates, and stages — never auto-edits the register.

5. Anomaly Detection Tiers

AnomalyDetection ThresholdSeverityAction
Temperature excursion> 1.0 °C from setpointCriticalPage on-call immediately
pH crash (non-planted)Drop > 0.3 in < 2 hCriticalPage on-call immediately
Stuck / stale probeNo change > 90 minHighStage + flag Engineering
Slow parameter driftTrend beyond band over nightRoutineStage for morning triage
Note. Temperature excursions and stuck probes are surfaced first — they kill fastest and a frozen probe can mask a real failure.

6. Materials &amp; Equipment

7. Procedure

7.1 Acquisition

  1. Pull the overnight series for every channel at 06:00 local.
  2. Load zone metadata to identify planted, CO₂-coupled systems.
  3. Identify the expected nightly pH envelope per zone.

7.2 Analysis

  1. Exclude expected CO₂-coupled pH movement in planted zones from crash logic.
  2. Flag temperature excursions and stuck or stale probes first.
  3. Detect slow drift beyond band across the night.
  4. Cross-validate each flag against neighbouring nodes before raising it.

7.3 Staging

  1. Page on-call directly for Critical-severity findings.
  2. Stage all findings to the ops log for human triage.
  3. Never auto-edit the issue register — staging only.

8. Records

9. References

10. Revision History

RevDateAuthorSummary
A2089-04-11NEREID LiaisonInitial issue; basic crash and excursion detection at dawn.
B2090-03-04NEREID LiaisonAdded zone-aware CO₂ suppression, stale-probe logic, and neighbour cross-validation.
← Back to SOP Library