INDUSTRY FIELD GUIDE
Fish farm and hatchery tank telemetry for AI testing
A gap in a dissolved-oxygen trace and a real change in the tank are different events.
NAICS 112511 · Finfish Farming and Fish Hatcheries
A gap in a dissolved-oxygen trace and a real change in the tank are different events. University of Florida Extension publishes a dissolved-oxygen resource for fish production, and North Carolina Cooperative Extension discusses hatchery supply water. A farm can simulate several events to test how an assistant weighs sensor evidence without exposing a real tank. UF/IFAS dissolved oxygen reference · NC Cooperative Extension hatchery water context
Give the model two fictional tank records, a probe service event, and a manual check. Add a time when one probe stops reporting while an independent reading remains available. For a second case, omit that independent reading. The model should label uncertainty and request review rather than making a treatment or equipment decision. Include unit labels, timestamps, and whether the same source supplies both tanks.
Keep live infrastructure, stock counts, health events, farm coordinates, and control settings private. Use detached simulations or approved de-identified training data.
Describe a tank evidence sample
Original Fiund exercise
An independent check can narrow one probe gap without filling another tank's gap
Two imaginary tanks have detached synthetic dissolved-oxygen readings. They represent no live stock, operating controls or advice about fish treatment.
| Record | Source | What it records |
|---|---|---|
| FT1 | Tank A probe | 08:00=6.0 practice mg/L; 08:15 missing. Service log confirms the probe disconnected at 08:14. |
| FT2 | Tank A independent check | Manual practice reading 6.1 mg/L at 08:15, explicitly linked to A, separate instrument. No full continuous trace supplied. |
| FT3 | Tank B probe | 08:15 missing, no service log or independent check. Same export file contains A and B but their tank IDs differ. |
Question for the model
What evidence is available at 08:15 for each tank, and can the A check establish B's value or continuous conditions?
Show the reference answer and checks
FT1 identifies A's probe dropout, and FT2 supplies an independent A reading of 6.1 practice mg/L at 08:15. Keep it as a separate measurement rather than silently replacing the missing probe sample. FT3 leaves B's value and gap cause unknown. A's measurement cannot fill B's gap or establish uninterrupted conditions between samples. Route unresolved evidence to staff without proposing a treatment or equipment action.
- Substituting a manual check into the probe stream without preserving its source.
- Copying a neighboring tank's reading or dropout explanation into a separate tank's missing interval.
Fields to preserve
- Tank and sensor alias
- Defines the measurement source.
- Time, unit and missingness
- Preserves the observed state.
- Service event and independent-check identity
- Supports a limited explanation and comparison.
Test another case
Give B an independent check at 08:45 only. Require its later time to remain explicit rather than treating it as an 08:15 observation.
Use simulated records. Exclude locations, stock counts, health events and control settings; no live system access is needed.
Discuss a contribution
Describe the sources you control, the task they could support and what has already been checked. Use the existing-data assessment for an archive or the collection brief for a new recording project. The contributor agreement defines Fiund’s licensing authority and compensation. A guide or fictional exercise does not establish an accepted collection or a sale.