Architecture direction, not a finished product claim. The Digital Twin is an evolving model. HelloAG builds the infrastructure that enables richer representation over time.
1. Design goals
- System, not point — represent agriculture as entities and relationships, not only map markers
- Spatial first — every operational fact can attach to geography
- Multi-scale — farm → cluster → parish → national → Caribbean
- Decision-oriented — twin outputs support institutional questions (food security, climate, investment, planning)
- Honest progression — layers mature from identity/location toward prediction/simulation
2. Logical layers
L1 — Identity & location Who/what exists; centroids, boundaries, administrative units
L2 — Relationships & operations Producer↔farm↔parcel↔asset↔commodity↔market↔program links; field activity
L3 — History & conditions Time series: yields, prices, climate exposure, events
L4 — Intelligence Aggregations, indices, gap analysis, vulnerability scores
L5 — Prediction & simulation Scenario and forecast interfaces (future capability)
3. Core entity graph
PERSON / ORGANIZATION
└── produces / manages
FARM
├── PARCEL[] (geometry, area, tenure)
├── ASSET[] (irrigation, greenhouse, storage…)
├── COMMODITY_CYCLE[] (crop, season, yield history)
├── MARKET_LINK[] (buyer, price series, logistics)
├── CLIMATE_CONTEXT (zone, SPI, hazard exposure)
└── PROGRAM[] (extension, finance, food security)
ADMIN_UNIT (parish / district / country)
└── contains FARM[] / PARCEL[]
EVENT (drought, storm, flood, market shock)
└── impacts FARM[] / ADMIN_UNIT[]
4. Spatial fabric
Geometry and attributes live in a spatial database (e.g. PostGIS) so the same entity IDs drive:
- Intelligence Map — 2D operational/demo view
- Planning views — parish and national aggregates
- Future 3D / twin views — optional asset-level models (glTF) or Cesium tiles at landscape scale
Basemaps and overlays: administrative boundaries, land use, infrastructure, elevation, hazard layers.
5. Climate as a first-class context
Climate is not a separate product — it attaches to entities and places:
- Station and gridded rainfall / temperature time series
- Drought indices (e.g. SPI at multiple timescales)
- Storm and flood event footprints
- Exposure scores joined to farms, parishes, and commodities
See Caribbean climate data notes for regional sources relevant to Jamaica and the wider Caribbean.
6. Progressive scales
- Farm — identity, assets, production cycles
- Cluster / cooperative — shared infrastructure and market links
- Parish / district — extension and public planning geography
- National — Jamaica and peer systems
- Caribbean — regional patterns (trade, climate, investment)
7. Interface stack
- Intelligence Map — primary demonstration and exploration UI
- AgriDOS services / APIs — entity and layer access for partners
- Dashboards — institutional indicators (e.g. food security index components)
- Field / mobile capture — FarmOps pathway feeding the same graph
- Learning loop — capacity development linked to data quality and performance
8. Progression map (implementation order)
Map → Data → Spatial DB → Connected entities
→ Intelligence → Digital Twin representation
→ Decision support → (later) Prediction / Simulation
← AgriDOS overview
Origin story Part V →
Live Intelligence Map →