S-Map Digital Twin — Seoul's 605 km² 3D Urban Model for Planning and Simulation

What S-Map Is and Why It Exists

S-Map — officially branded Virtual Seoul — is the Seoul Metropolitan Government’s digital twin: a three-dimensional, data-rich replica of the entire city built for urban planning, infrastructure management, disaster simulation, and citizen engagement. Initiated in 2018, the project maps 605.23 square kilometers of metropolitan terrain and more than 600,000 above-ground structures into a navigable virtual environment that city officials, engineers, and — through an open-source laboratory — external researchers can use to test policy interventions before committing real-world resources. The ambition is not merely visualization. S-Map’s three development stages move the platform from a static 3D model toward a living simulation engine that ingests real-time data from the S-DoT sensor network, the TOPIS transport management hub, and dozens of other municipal data feeds.

Digital twins are not new to city government — Singapore’s Virtual Singapore and Helsinki’s 3D model preceded S-Map — but Seoul’s implementation is distinctive in two respects. First, it includes subsurface infrastructure: waterworks piping, natural-gas distribution networks, telecommunications conduits, and district-heating lines are mapped alongside above-ground buildings. Second, the Seoul Metropolitan Government has committed to open-sourcing the analysis layer, creating a digital laboratory where qualified experts can run their own experiments on the virtual city without navigating procurement processes or data-sharing agreements.

The Three Development Stages

S-Map’s buildout follows a phased architecture, each stage adding computational capability on top of the previous one.

Stage 1 — Physical Replication. The foundation layer is a geometrically accurate 3D model of Seoul’s built environment. Building footprints, facades, roof geometries, road surfaces, bridges, elevated highways, and major vegetation canopies are rendered at Level of Detail 3 (LoD3), meaning individual architectural features — window patterns, balcony protrusions, rooftop equipment — are visible. Terrain elevation uses a 1-meter-resolution digital elevation model derived from airborne LiDAR surveys. Data sources for Stage 1 include cadastral records from the Ministry of Land, Infrastructure and Transport, building-permit databases maintained by each of Seoul’s 25 autonomous districts, and the city’s own GIS repository.

Stage 2 — Spatial Information Collection and Visualization. Stage 2 layered sensor-derived and AI-analyzed data onto the Stage 1 geometry. The core input was a fleet of fixed-wing survey aircraft and rotary drones capturing 25,000 high-resolution aerial photographs across the metropolitan area. AI models — primarily convolutional neural networks trained on labeled urban imagery — processed these photographs to extract building heights, facade materials, vegetation health indices (NDVI), solar irradiance potential on rooftop surfaces, and shadow-casting patterns at hourly intervals throughout the year. LiDAR point clouds were fused with the photographic data to generate centimeter-accurate surface models, and photogrammetric mesh reconstruction produced the textured 3D surfaces users navigate in the S-Map viewer.

Underground infrastructure was modeled separately, using as-built drawings from utilities (Seoul Waterworks, Seoul City Gas, KT, SK Broadband) cross-referenced with ground-penetrating radar (GPR) surveys at 2,400 sample points across the city. The resulting subsurface layer maps more than 48,000 kilometers of buried piping, conduit, and cable runs, color-coded by utility type and accessible as a toggleable overlay in the S-Map interface.

Stage 3 — City Analysis and Simulation. The current and most ambitious phase transforms S-Map from a viewer into a simulation engine. Urban planners can define scenarios — a new high-rise development in Gangnam, a bus-lane expansion along Sejong-daero, a 100-year flood event on the Jungnangcheon tributary — and the platform computes projected impacts on traffic flow, pedestrian density, air quality, shadow patterns, noise propagation, and stormwater runoff. The simulation models draw on live feeds from the S-DoT IoT network (environmental conditions), TOPIS (traffic volumes), and the Korea Meteorological Administration (weather forecasts) to anchor projections in current conditions rather than static historical averages.

Technical Infrastructure Behind the Platform

Running a 605 km² digital twin at interactive frame rates demands substantial compute. S-Map’s backend is hosted across two Seoul Metropolitan Government data centers (primary in Guro Digital Complex, disaster-recovery in Sangam-dong) on a hybrid architecture: CPU-intensive GIS processing and database queries run on bare-metal servers; GPU-intensive 3D rendering and AI inference run on NVIDIA A100 clusters provisioned through a private-cloud orchestration layer. Storage exceeds 8 petabytes, dominated by the LiDAR point-cloud archive, aerial-photo library, and time-series simulation outputs.

The viewer application supports three access tiers. A web-based viewer built on CesiumJS and Three.js handles light-duty exploration and is available to any city employee with SSO credentials. A desktop client built on Unreal Engine 5 provides high-fidelity visualization for planning teams working on major development proposals. And the open-source laboratory — discussed in detail below — provides Python and R APIs for programmatic access to the data and simulation engines.

Network connectivity between the data centers and the TOPIS control room runs over dedicated dark-fiber links at 100 Gbps, ensuring that live traffic feeds arrive at the S-Map simulation engine with sub-10-millisecond latency. This is critical for real-time disaster-response simulations, where a 30-second delay in traffic data could render an evacuation-route model useless.

Subsurface Mapping — The Hidden Half of the City

Most city digital twins stop at the street surface. S-Map’s inclusion of underground infrastructure is both technically challenging and operationally valuable. Seoul’s subsurface is dense: the city operates 13,000 kilometers of water-supply piping, 6,200 kilometers of sewer mains, 3,800 kilometers of natural-gas distribution lines, and an estimated 25,000 kilometers of telecommunications conduit — all compressed into a footprint where excavation for one utility routinely damages another.

The subsurface model enables conflict analysis before any shovel breaks ground. When a district office submits a road-resurfacing plan, the system automatically checks whether the proposed excavation depth intersects any mapped utility. If a conflict is detected, the planner receives a visual overlay showing the offending pipe’s diameter, material, installation date, and operating pressure, along with the contact information for the responsible utility. This workflow has reduced accidental utility strikes during Seoul roadworks by an estimated 40 percent since Stage 2 went live, based on incident reports compiled by the Seoul Infrastructure Safety Division.

The Open Lab — Democratizing Urban Simulation

S-Map’s Open Lab is an open-source-based digital laboratory where accredited researchers, urban planners, and technology developers can run experiments on the virtual city. Access is granted through an application process managed by the Seoul Smart City Bureau; successful applicants receive API credentials, documentation, and a sandbox environment pre-loaded with anonymized datasets.

The lab exposes three primary APIs. A Spatial Query API returns geometric and attribute data for any object in the model — query a building by address and receive its footprint polygon, floor count, construction year, facade material, and current energy rating. A Simulation API allows users to submit scenario definitions (JSON-formatted parameter sets describing proposed interventions) and receive simulation outputs as GeoJSON layers or tabular data. A Streaming API provides real-time feeds from S-DoT and TOPIS, enabling researchers to build live-updating dashboards or trigger experiments based on current conditions.

Use cases from the first two years of Open Lab operation include:

• A Seoul National University research team modeling the urban heat-island effect of replacing asphalt parking lots with permeable pavement across Jongno-gu, finding a projected 1.2 °C reduction in peak summer surface temperature.
• A KAIST team simulating the noise-propagation impact of a proposed elevated rail extension in Songpa-gu, producing contour maps that the district office used to justify a revised alignment.
• A private-sector smart-building startup testing a predictive HVAC control algorithm against S-Map’s solar-irradiance and shadow data, reducing simulated cooling energy consumption by 18 percent in a 40-story mixed-use tower.

Integration With Seoul’s Smart-City Ecosystem

S-Map is not a standalone visualization tool. It is a platform that other smart-city systems use as a shared spatial reference.

• S-DoT sensors. Environmental data from the 1,100-unit IoT mesh populates real-time atmospheric layers in S-Map, enabling planners to see PM2.5 concentrations overlaid on building geometry and traffic patterns simultaneously.
• TOPIS transport hub. Live bus and taxi GPS positions from TOPIS are rendered as moving icons on the S-Map street network, providing a God’s-eye view of transit flow that the TOPIS control room uses during major events and incident response.
• AI traffic management. The AI traffic-signal optimization system uses S-Map’s intersection geometry to model signal-phase impacts on queue lengths and spillback, supplementing its camera-based detection with geometric constraints the cameras cannot capture.
• Smart waste management. The RFID food-waste bin locations are mapped in S-Map, and collection-route optimization algorithms use the platform’s road-network model — including turn restrictions, one-way streets, and time-of-day access rules — to compute shortest-path routes.
• Public safety. During disaster simulations, S-Map ingests flood-inundation models and overlays them on the subsurface infrastructure map to predict which utility segments will be submerged, enabling preemptive shutoffs of gas lines in flood zones — a capability directly relevant to the CCTV-based public safety network.
• Digital government. The digital government services portal links to S-Map’s public-facing viewer, allowing citizens to explore planned developments in their neighborhood and submit comments through an integrated feedback form.

How Seoul’s Digital Twin Compares Globally

Digital-twin programs are proliferating across major cities, but they differ substantially in scope, data freshness, and openness.

Seoul’s combination of full-city coverage, subsurface mapping, real-time multi-source data integration, and an externally accessible open laboratory places it among the most advanced municipal digital twins in production anywhere. The Stage 3 simulation capability — tested against real-time conditions rather than historical snapshots — is a differentiator that few peers have achieved at metropolitan scale.

Challenges and Future Directions

Data currency is the persistent challenge. Seoul’s built environment changes continuously: roughly 12,000 building permits are issued annually across the 25 districts, each potentially altering the physical geometry that S-Map represents. The current update cycle re-flies aerial photography annually and processes LiDAR refreshes for priority districts (those with the most active construction) semi-annually. The lag between a building’s completion and its appearance in S-Map can reach six to twelve months — acceptable for strategic planning but problematic for applications like emergency routing that depend on current conditions.

The SMG is exploring satellite-based change detection as a faster update mechanism. Commercial synthetic-aperture radar (SAR) satellites can detect new construction at roughly one-meter resolution on a weekly cadence, triggering targeted drone flights only where changes are identified. This approach could cut the average update lag to under eight weeks while reducing aerial-survey costs by an estimated 60 percent.

Interoperability with private-sector digital twins is another frontier. Samsung C&T, Hyundai Engineering & Construction, and several major Korean developers maintain their own BIM (Building Information Modeling) environments for projects under construction. Linking these BIM models to S-Map would provide near-real-time construction progress tracking, but differences in coordinate systems, level of detail, and data schemas have so far prevented seamless integration. The SMG is working with the Korean BIM Standards Committee to define a bridging specification that would allow BIM-to-CityGML data exchange with minimal manual intervention.

Finally, the growing volume of simulation outputs raises questions about archival policy. Every scenario simulation generates gigabytes of output data — flood maps, traffic heatmaps, noise contours — that may be needed for regulatory review, public consultation, or litigation years after the fact. The current archival policy retains all outputs for five years, but legal counsel has recommended extending retention to match the 30-year lifecycle of major infrastructure projects. At current simulation volumes, that would require roughly 200 terabytes of additional cold storage per year.

S-Map’s Role in Seoul Vision 2030

S-Map is not a showcase project. It is foundational infrastructure for the decisions that will shape Seoul through the next decade. The smart parking systems being deployed across Gangnam and Jongno were sited using S-Map’s pedestrian-flow simulations. The digital inclusion programs targeting senior residents use S-Map’s demographic overlays to identify neighborhoods where the over-65 population exceeds district averages. And the autonomous-driving corridors planned under Vision 2030 were validated in S-Map long before physical road modifications began, testing edge cases — a delivery truck double-parked in a narrow Bukchon alley, a sudden monsoon downpour flooding a Gangnam underpass — that would be dangerous or impossible to stage in the real city.