🌾 Victory Farm & Animal Farm — FMIS

Two adjoining holdings in Berlin-Tiergarten, run as one operation.

Victory Farm (Old MacDonald, Agricultural Manager) — arable. Parcels: Wheatfield (buckwheat, seeded, clay/clay loam, soil moisture 50%, air 15.4 °C) and Cornfield (corn, growing, loam, 60%). Buildings: farmhouse, Triton Tower (water tower), Big Red Barn.

Animal Farm — also "Manor Farm" (Napoleon the Pig). Four grass parcels, all maturing: South Pasture (loam), East Pasture (silt loam), Water Meadow (sandy loam, wettest at 70%), The Paddock (loam). Building: farmhouse. Squealer and Snowball are on the books with no land or stock.

Livestock — 73 head: 43 dairy cattle and 30 pigs. Cattle graze the four Animal Farm pastures (two held in the Big Red Barn); the pigs carry no parcel and sit near the buildings. Every animal wears a collar reporting heart rate, location and activity.

Weather: mild and dry across five stations — 15.3–16.2 °C, ~50% humidity, negligible rain, light south-easterly. Forecast overcast, 11–16 °C.

Flags: Goliath's collar (pig032) reads NOT OK with a flat battery. Also note the observation timestamps run ahead of today's date — the feed clock is fast.

Two bulls — Beany (752 kg, East Pasture) and Ferdinand (757 kg, South Pasture), both Old MacDonald's. There is also one bull calf, Bumble (118 kg, still with the cows on South Pasture). The three big pigs — Giant, Mammoth, Colossus — are boars, not bulls.

Ferdinand is on South Pasture, which is loam — an even mix of sand, silt and clay (texture class "Loam").

No — 41 of the 43 cattle are out on the four pastures. Two are held in the Big Red Barn: Twilight (flagged "needs to see the vet") and Buttercup (a calf born today).

Her record carries a single note — "Needs to see the vet." — with no diagnosis attached, so she is being held in the barn pending that check. Nothing else in her data points to a problem: health status still reads "healthy", heart rate 52 bpm, weight flat at ~434 kg for months. She is also on oats rather than grass, consistent with being kept in.

Twilight isn't sick — she calved today. Buttercup, the newborn in the barn with her, has calvedBy pointing to Twilight (cow006). Everything else lines up with a fresh cow:

• In the Big Red Barn (calving pen), alone but for her calf.
• Reproductive condition noStatus — the only adult cow not active/inHeat/inCalf.
• Moved onto oats; status AT REST; heart rate 52 bpm.
• Weight dipped 436→427 kg over the winter and recovered to 434 — a gestation body-condition swing.

So the "see the vet" note is the routine fresh-cow / newborn check, not an emergency. Worth watching anyway: she is 8.4 years old (the milk-fever risk group) and her HR of 52 is the lowest in the herd; the calf is light at 20 kg, so prioritise the colostrum check. Her collar is healthy, so the low HR is a real reading.

Pulled 12–25 months of weight history on 20 head across every age group. The herd is mostly sound — mature cows holding steady, growth curves on track, both bulls fine, nursing calves Bumble (0.79 kg/day) and Blossom (0.70 kg/day) on target.

Three individuals to investigate:

• Tulip (cow010) — rose to 355 kg by May, then 355→342→328→315→315: lost 40 kg (11%) over four months, now stalled. Abnormal for a young adult.
• Bug (cow025) — 413→360 kg steady decline over ~14 months, 8-year-old cow. Chronic-disease pattern.
• Sun (cow005) — weight fine, but resting heart rate 70 bpm against a herd average of 53. Sole cardiac outlier.

Data oddity: Spots (cow040) shows +63 kg in one month against a history of +9–20 — physiologically impossible, so a mis-weigh. Re-weigh.

Breed isn't recorded — every animal just carries species: "dairy cattle", no breed attribute. ("Fresian" and "Jersey" appear only as the names of two individual cows.)

From the weights it behaves like Jersey or a small breed: mature cows plateau at ~360–490 kg and the bulls sit at ~750 kg, well under Holstein-Friesian. If that's right, the absolute-weight worries soften — mature cows holding at 360–490 kg are at legitimate size, and the 2025-born heifers running ~0.5 kg/day are low-normal rather than stunted. The trend flags — Tulip, Bug, Sun — don't depend on breed and stand either way. Worth adding breed to the records so growth can be benchmarked properly.

Confirming Jersey clears the whole absolute-weight column — mature cows and the 2025 heifers are all normal for the breed, and Buttercup's 20 kg birthweight is only mildly light. What's left:

• Tulip (cow010) — top priority. −11% over four months, now stalled. Parasites, Johne's, or feed-access competition.
• Bug (cow025) — 53 kg decline over 14 months. Run a Johne's screen; check her teeth.
• Sun (cow005) — resting HR 70 vs herd 53. Look for pain, fever, or stress.
• Spots (cow040) — data fix: re-weigh, likely a mis-weigh or tag swap.
• Twilight (cow006) — routine post-calving check; watch the low HR given her age.

Everything else in the sample reads healthy against Jersey norms.

There is no North Pasture on this farm. The six parcels are Wheatfield and Cornfield (arable) and South Pasture, East Pasture, Water Meadow, The Paddock (grazing).

And parcel size can't be read from the system for any of them: AgriParcel records carry crop, soil, sensors and current temperature/humidity, but no boundary geometry and no area attribute. So there's no stored acreage. Collar GPS is the only way to estimate extent — name a real pasture and I'll do it.

Inferred from collar GPS — a single fix per animal, all stamped 06:36 on 2026-09-06, no track history.

9 fixes (Trotter and the calf Bumble on the identical point), centred just south-west of the Big Red Barn:

• Spread ~120 m east–west by ~80 m north–south
• Bounding box ≈ 0.97 ha (2.4 acres)
• Convex hull of the fixes ≈ 0.79 ha (1.95 acres)

The hull is a hard lower bound — stock bunch and don't walk the fence line — so realistic field size is around 1.2–2.5 ha. Eight grazing head on that hull is ~4 LSU/ha, above the ~2–3 typical for set-stocked grazing; it may be a rotational cell. Tulip is in this mob, at the south-west edge.

The pigs carry no parcel and no boundary record — this is from collar GPS, one snapshot at 06:36.

The 30 fixes split into two groups: a main piggery of 27 sows/gilts, and three boars — Giant, Mammoth, Colossus — clustered ~200 m north-west in a separate pen (normal husbandry).

Main piggery: ~63 m east–west by ~98 m north–south; bounding box ≈ 0.62 ha, occupied hull ≈ 0.50 ha. Treat as a lower bound — the real pen is plausibly 0.6–1.0 ha. At ~45–55 pigs/ha this reads as a bedded yard or hard-standing paddock, not extensive free-range. Boar pen: spans ~25 m between fixes, so on the order of 20–30 m across.

None. Every pig in the system is an adult — 27 sows/gilts and 3 boars, all 200–343 kg, none tagged as young or nursing.

Litters are on the way: Peach and Tango are inCalf (in-pig), and Magenta and Bruce are inHeat. No piglets have been born or registered yet.

Working from 27 breeding females, 3 boars and the reproductive flags, with standard pig figures (~10–12 born alive per litter, 115-day gestation, up to ~2 litters/sow/year):

• Confirmed pipeline now — 2 sows inCalf → ~20–25 piglets, due within about four months.
• Next round — add the 2 sows inHeat if served → another ~20–25; combined ~40–50 over the next 5–6 months.
• Steady-state capacity — ~24 productive sows → ~390/year (traditional rate) to ~530/year (pushed harder).

Only 2 of 27 showing inCalf is low — a continuously bred herd would have 5–7 pregnant at any moment — so current breeding activity looks like the bottom of that range. Boar ratio 1:9 is adequate, so output isn't sire-limited. No litters are registered as entities, so nothing here counts actual piglets on the ground.

Likely, yes. The main risk isn't square-metres — it's that the data shows one undifferentiated pig group with no farrowing sub-pens, and two sows due to farrow into that shared space within ~4 months.

• Crushing and savaging — group farrowing without protected creeps drives pre-weaning death; expect 15–25%+ litter mortality rather than ~10–15%.
• Chilling — newborns need a 30–34 °C micro-climate; ambient is ~15 °C and cooling, with no creep provision recorded.
• Overcrowding as litters grow — the sow group already sits at ~45–55 pigs/ha; piglets at 25–30 kg by ten weeks push bedding and fouling load hard.
• Neonatal disease — higher density plus young stock builds up scour pathogens, joint ill, greasy pig disease.
• Sow condition — several females already light (Blush 200 kg, Clifford 203 kg); group feeding lets dominant sows push lactating sows off feed.

From the current data: Goliath's collar is down — a sow with no location/activity/heart-rate telemetry as farrowing nears. And three mature boars share a pen next to in-heat sows (fighting risk).

Caveat: if farrowing arcs or a farrowing house exist but simply aren't recorded as entities, the first three points largely go away. Confirm on the ground before the first sow is due.

Inputs: 27 sow/gilt fixes and 3 boar fixes from the 06:36 snapshot, projected to metres. Bounding box of the pig data = 273.7 m × 193.7 m = 5.30 ha (the holding itself has no recorded boundary).

Voronoi over-allocates — most of its area is empty ground between the groups. NN circles measure spacing, not paddock. The convex-hull footprint (~0.5 ha sows) is the most defensible parcel size, and it's a lower bound — call the sow paddock 0.5–1.0 ha, the boar pen 0.05–0.1 ha.

Stocking check (27 sows ~283 kg; benchmark 9–15 sows/ha outdoor, ~1,000–1,500 kg LW/ha sustainable):

• On the footprint: ~50 sows/ha, ~14,000 kg/ha — 3–5× the density guideline, ~10× the loading. Yard, not pasture.
• On NN r = NN (1.40 ha): ~19 sows/ha — still over the 9–15 range.
• Only the Voronoi allocation (~3 ha, ~8–9/ha) looks acceptable, and that's inflated by dead land.

Boars: footprint gives ~75 m²/boar, above the EU 6–10 m² minimum. Their welfare risk is being housed together next to in-heat sows, not area.

Six-way partition: the four pastures plus the sow and boar enclosures. Animal Farm bounding box (71 fixes): 612 m × 970 m = 59.4 ha — mostly empty ground; the clusters sit 250–500 m apart.

* East Pasture is distorted by one animal (Beany) sitting 250 m from the rest.

Which to trust: Voronoi hands each pasture 9–15 ha of mostly dead ground — unusable for stocking. Footprint is the honest occupied area but a lower bound (herds bunched at 06:36). Best estimate: 2–5 ha per pasture.

Stocking (permanent pasture ~2–2.5 LSU/ha; a 400 kg Jersey ≈ 0.67 LSU): on the realistic middle, the four pastures run at to modestly above the guideline — tolerable under rotation with rest. The sow yard is the real exception at ~24 LSU/ha on its footprint. Boars have adequate space.

Anomaly the geometry exposes: Beany (bull, registered to East Pasture, flagged "check hooves") is physically in the South Pasture cluster, right alongside the South Pasture bull Ferdinand — two mature bulls together, one in the wrong paddock per the records. Check fences and the location tag.

Applied to all 43 head (cow001–cow040, cow101–cow103) — breed: "Jersey" on every one.

East Pasture (AgriParcel:004) — grass, silt loam. His locatedAt is stamped 11:32 today.

Note: Beany is a bull (legalId M-bull001), not a cow — despite the cowNNN entity ID.

Can't confirm — and nothing suggests it. His collar still reads locatedAt East Pasture, last GPS fix [13.34973, 52.51139], same timestamp.

The parcels carry no boundary geometry, so there is no fence line to test the point against, and only one location sample in history — no movement trail.

Done. Set Beany's location to [13.35359, 52.51361] — the centroid of the other 9 cows on East Pasture, about 300 m north-east of her last fix.

This overwrites a collar reading, so Device:cow001's next GPS report will replace it.

Boars Mammoth (pig031), Giant (pig035) and Colossus (pig037) cluster near [13.34713, 52.51973].

Created a 32-vertex polygon approximating a 20 m-radius circle on that centroid — all three sit within ~15 m — stored it as a new entity urn:ngsi-ld:Building:BoarPen (type Building), and set each boar's locatedAt to it. Read "20 m" as radius; a 20 m diameter would not enclose the spread.

Deleted urn:ngsi-ld:Building:BoarPen and rebuilt it as urn:ngsi-ld:Pen:BoarPen (type Pen, same circular polygon), then repointed all three boars' locatedAt at it.

Yes. locatedAt → Pen:BoarPen for all three, and their last collar fixes fall 12–15 m from centre, inside the 20 m radius — the polygon was drawn around those points.

Caveats: the collars keep reporting live positions, so locatedAt will not auto-update if a boar wanders out; the broker is not enforcing polygon filters, so containment was checked by hand.

Right now it carries only name, description, lastUpdate and location (the polygon). There is a Pen schema — the Smart Data Models model at ontology://agrifood/Pen — which only forces lastUpdate but defines plenty more:

• Siting — farmId → the farm, buildingId / compartmentId if indoors
• Occupancy — numAnimals, sex (the farm uses Male/Female, as on the animals), arrivalTimestamp, deadAnimalsSinceDateOfArrival
• Performance — avgWeight, weightStDev, avgGrowth, feedConsumption, waterConsumption
• Environment — temperature, humidity, co2, luminosity (all observedAt)

The model has no species, numeric area or occupancy status field, so those stay as extensions. Minimum worth adding now: farmId, numAnimals, sex, arrivalTimestamp, avgWeight, plus extension area / status / hasAnimal.

Written to Pen:BoarPen on the Pen schema:

• farmId → AgriFarm:farm002 (Animal Farm) · numAnimals 3 · sex Male
• arrivalTimestamp 2026-09-07T11:46Z · avgWeight 302 kg · weightStDev 44 kg (Mammoth 325, Giant 340, Colossus 241)
• extensions kept: species pig, area 1249 m², status occupied, hasAnimal → the three boars (inverse of their locatedAt)

lastUpdate bumped to 12:30. (An earlier pass had used ad-hoc category / containedInPlace names; those were re-aligned to sex / farmId.)

Via the temporal API:

• Adult cattle (cow001–cow040) — monthly readings, Sep 2024 → today, ~25 points each.
• Calves (cow101–cow103) — weekly since May 2026, ~17 points.
• Pigs — current value only, no series.

So cattle weights trend over ~2 years and the calves over ~4 months; pigs cannot be trended.

Sampled ~15 head across age classes against Jersey norms (mature cow 400–500 kg, bull 600–820 kg, healthy heifer gain ~0.5–0.7 kg/day).

Sound: both bulls stable, mature cows flat within range, calves Bumble (0.79 kg/day) and Blossom (0.71 kg/day) on target; 2025-born heifers ~0.47–0.55 kg/day — low-normal for Jersey.

Flags:

• Tulip (cow010) — lost ~40 kg (11%) May–Aug, now stalled.
• Bug (cow025) — steady decline 406→360 kg since Jan, 8-year-old.
• Spots (cow040) — impossible +63 kg in one month; re-weigh.
• Chalky (cow020) — tracking ~15% under age target; monitor.

healthCondition is recorded for only 2 of 43 animals — everything else is inferred from weight alone.

Bug recheck confirms a real problem: flat at ~405–413 kg to Jan 2026, then a steady monotonic decline to 360 kg (−13%). Sick, not a data blip.

Set healthCondition on all 43 cattle — 41 healthy, 2 sick:

• Tulip (cow010) — sick + comment: ~11% loss May–Aug; likely parasitism or subclinical disease — BCS and faecal egg count.
• Bug (cow025) — sick + comment: ~13% loss since Jan in an 8-year-old; likely parasitism, dental wear, or Johne's — vet exam.

Spots (cow040) — weight amended 349 → 300 kg, left healthy.

Added. Chalky (cow020) stays healthy with a note: average daily gain ~0.41 kg/day, roughly 15% below the Jersey age target, growth continuous with no stall — review ration and worm burden, re-weigh monthly.

From the healthCondition and comment fields on the animals:

• Gastrointestinal parasitism — Tulip (cow010), Bug (cow025), and Chalky (cow020) as a watch item.
• Johne's disease (paratuberculosis) — Bug: chronic steady wasting in an 8-year-old.
• Subclinical / chronic disease, unspecified — Tulip, pending tests.
• Dental disease — Bug, age-related.
• Lameness / hoof lesion — Beany (cow001), from a "check hooves" note.
• Cardiac / pain / fever — Sun (cow005), resting heart rate 70 bpm against a herd mean of ~53.

Pigs carry no health flags — only forecast farrowing risks and a dead collar battery on Goliath.

Created 5 AnimalDisease records on the Smart Data Models schema (disease, animals, date, diagnosticTest), each linked to its animal (animals ↔ hasDisease) and parcel, with extension fields for clinical signs, differentials, severity, zoonotic/notifiable flags and treatment notes:

• Tulip — parasitic gastroenteritis / chronic wasting (suspected, moderate)
• Bug — Johne's disease (suspected, high)
• Chalky — subclinical PGE / ill-thrift (under-observation, mild)
• Beany — lameness / claw lesion, unspecified (suspected)
• Sun — resting tachycardia, undiagnosed (suspected, moderate)

treatmentStatus is pending diagnosis throughout — treatment options recorded per condition (anthelmintics ± flukicide, hoof trim + footbath, the traumatic-reticuloperitonitis protocol, etc.), nothing to be given blind.

Added 5 more, all forecast since no pig is clinically ill:

• Piglet scour complex — litters due from Tango & Peach (anticipated, high)
• Joint ill / navel infection — same litters (anticipated; S. suis is zoonotic)
• Greasy pig disease (S. hyicus) — piglets and weaners (anticipated)
• Mastitis-metritis-agalactia — Tango, Peach and the thin sows (risk)
• Boar fight trauma — Mammoth, Giant, Colossus (risk)

Ten AnimalDisease records total — 5 cattle, 5 pig — plus hasDisease back-links on the ten animals involved.