The Digital Twin of the City of Zurich for Urban Planning

Gerhard Schrotter, Christian Hürzeler•View original
HighlightsBalancedballad voice
Zurich is growing fast, and that growth has to fit inside the city. The Canton expects about 280,000 more residents by 2040, with 80 percent of that increase steered into urban areas to protect open land. If Zurich absorbs its share, you're looking at roughly 520,000 people inside the city limits. That means denser neighborhoods, tighter land-use tradeoffs, and a constant balancing act: add homes and jobs, but keep the character of beloved quarters and avoid the noise and friction that come with crowding. Traditional planning tools were built for slower times and smaller audiences. Schrotter and colleagues argue you need something more legible and shared. Enter the digital twin. Think of the twin as a trustworthy common map you can measure, question, and update. It's anchored in the Swiss national reference frame, EPSG 2056, and managed as a living system in the lifecycle sense that Michael Grieves described: data, models, and metadata that can be renewed and, when it helps, linked to real-time sensors. The core has three layers. Terrain, called Level of Detail 0, comes from a 2014 laser scan at 50-centimeter resolution with a typical height error of 30 to 40 centimeters. Blocks, Level 1, are extruded from official cadastral floor plans that carry roof-edge attributes like eaves and gable heights; their positions are accurate to about 10 to 15 centimeters, with height error around 50 centimeters. Roof shapes, Level 2, are refined using semi-automatic stereo aerial photogrammetry. It's detailed enough to trust, and disciplined enough to evolve. That discipline extends to governance. The twin sits under Switzerland's GeoIG 2007 framework with INSPIRE-aligned rules and Geocat as the national metadata catalog. Just as important, Zurich made "open by default" real. Since 2012, administrative data have been free to access and machine-readable, and by the end of 2018, the terrain, block, and roof models were public. In a ten-month window, 11,722 datasets were downloaded from the city's portal, with five three-dimensional datasets accounting for about a third—making three-dimensional data the most sought-after on the site. Coordination across 25 city departments keeps the pipes open and the updates flowing. Openness only matters if people can actually use it. So Zurich built low-threshold tools: a WebViewer and a browser three-dimensional map, using Esri's JavaScript framework, where anyone can sketch buildings, draw green space, drop trees, and submit proposals directly to planners. They tested it with people who had no planning background. It clicked. For younger residents, they turned parts of the twin into a Minecraft world. That experiment drew 100,000 downloads by early 2019 and racked up 24,500 hours of play—participation disguised as play. On site, an augmented reality app on HoloLens projects designs at a one to one scale, simulates shadows, and lets teams slice cross-sections. In an architectural competition, a toolkit of about 100 digital components rebuilt the old town, from Grossmünster to Fraumünster, so juries could compare variants on equal footing. This changes the conversation. Instead of opaque tradeoffs made by a few, debates move into transparent, climate-aware scenarios many can see and test. Zurich links geographic information system analyses to the three-dimensional twin to assess urban heat and airflow, then uses those results in real decisions on public construction. Under the Swiss CH2018 projections, the city expects hot days to jump from about six now to twenty-one by 2060. You can feel what that means when a shaded courtyard goes from a drawing to a place you can walk through virtually. There are limits. Switzerland lacks clear cantonal or federal guidelines for three-dimensional geodata. Questions remain about how building information modeling should fuse with city-scale geographic systems. And the practical stuff—keeping models current, modeling the subsurface, moving big volumes of data—never really ends. But the payoff is tangible: broader participation, clearer tradeoffs, and decisions that rest on a shared, testable model rather than static plans. In Zurich's case, as Schrotter and colleagues show, the twin isn't just a model of the city. It's a model for how a city decides.

Zurich is growing fast, and that growth has to fit inside the city. The Canton expects about 280,000 more residents by 2040, with 80 percent of that increase steered into urban areas to protect open land. If Zurich absorbs its share, you're looking at roughly 520,000 people inside the city limits.

That means denser neighborhoods, tighter land-use tradeoffs, and a constant balancing act: add homes and jobs, but keep the character of beloved quarters and avoid the noise and friction that come with crowding. Traditional planning tools were built for slower times and smaller audiences. Schrotter and colleagues argue you need something more legible and shared. Enter the digital twin.

Think of the twin as a trustworthy common map you can measure, question, and update. It's anchored in the Swiss national reference frame, EPSG 2056, and managed as a living system in the lifecycle sense that Michael Grieves described: data, models, and metadata that can be renewed and, when it helps, linked to real-time sensors. The core has three layers.

Terrain, called Level of Detail 0, comes from a 2014 laser scan at 50-centimeter resolution with a typical height error of 30 to 40 centimeters. Blocks, Level 1, are extruded from official cadastral floor plans that carry roof-edge attributes like eaves and gable heights; their positions are accurate to about 10 to 15 centimeters, with height error around 50 centimeters. Roof shapes, Level 2, are refined using semi-automatic stereo aerial photogrammetry. It's detailed enough to trust, and disciplined enough to evolve.

That discipline extends to governance. The twin sits under Switzerland's GeoIG 2007 framework with INSPIRE-aligned rules and Geocat as the national metadata catalog. Just as important, Zurich made "open by default" real.

Since 2012, administrative data have been free to access and machine-readable, and by the end of 2018, the terrain, block, and roof models were public. In a ten-month window, 11,722 datasets were downloaded from the city's portal, with five three-dimensional datasets accounting for about a third—making three-dimensional data the most sought-after on the site. Coordination across 25 city departments keeps the pipes open and the updates flowing.

Openness only matters if people can actually use it. So Zurich built low-threshold tools: a WebViewer and a browser three-dimensional map, using Esri's JavaScript framework, where anyone can sketch buildings, draw green space, drop trees, and submit proposals directly to planners. They tested it with people who had no planning background.

It clicked. For younger residents, they turned parts of the twin into a Minecraft world. That experiment drew 100,000 downloads by early 2019 and racked up 24,500 hours of play—participation disguised as play.

On site, an augmented reality app on HoloLens projects designs at a one to one scale, simulates shadows, and lets teams slice cross-sections. In an architectural competition, a toolkit of about 100 digital components rebuilt the old town, from Grossmünster to Fraumünster, so juries could compare variants on equal footing.

This changes the conversation. Instead of opaque tradeoffs made by a few, debates move into transparent, climate-aware scenarios many can see and test. Zurich links geographic information system analyses to the three-dimensional twin to assess urban heat and airflow, then uses those results in real decisions on public construction.

Under the Swiss CH2018 projections, the city expects hot days to jump from about six now to twenty-one by 2060. You can feel what that means when a shaded courtyard goes from a drawing to a place you can walk through virtually.

There are limits. Switzerland lacks clear cantonal or federal guidelines for three-dimensional geodata. Questions remain about how building information modeling should fuse with city-scale geographic systems.

And the practical stuff—keeping models current, modeling the subsurface, moving big volumes of data—never really ends. But the payoff is tangible: broader participation, clearer tradeoffs, and decisions that rest on a shared, testable model rather than static plans. In Zurich's case, as Schrotter and colleagues show, the twin isn't just a model of the city. It's a model for how a city decides.