A digital twin is a virtual counterpart of a physical asset, system or city — connected to it. The connection is the definition: sensors and operational systems feed the twin current state; models simulate behaviour; insights flow back as decisions. A detailed 3D city model is scenery; it becomes a twin when live data moves through it and someone manages the real world differently because of it.
The anatomy of a twin
Four layers recur. A geometric-semantic base: terrain, buildings, networks with identities and attributes (CityGML/3D Tiles for cities, BIM for facilities). A data spine: IoT feeds, SCADA, operations systems streaming state onto that base. Models: flood, traffic, energy, structural — simulation that turns state into foresight. And interfaces: dashboards and APIs where humans and systems act.
Maturity climbs the same ladder: visualise, monitor, simulate, and eventually optimise in closed loops.
City twins and asset twins
City twins federate: many custodians, many systems, one spatial frame — their hard problems are governance, identity management and data agreements more than graphics. Asset twins (a dam, a plant, a rail line) integrate deeply on one owner’s estate, often pairing BIM detail with GIS context.
Both stand on unglamorous geospatial foundations: accurate reality capture, well-modelled networks, and identifiers that let a sensor reading find its exact place.
Building one without the hype
Successful programmes start from decisions, not demos: which recurring choices will the twin inform, at what cadence, for whom? That scopes the geometry level of detail, the feeds worth integrating and the models worth calibrating. Standards (CityGML, 3D Tiles, sensor APIs) keep components exchangeable.
The anti-pattern is the cinematic model with no data behind it — beautiful, expensive, and abandoned after the launch video.