CityGML translation is about making a three-dimensional city model understandable in another language while protecting the geographic and semantic evidence that lets software treat a building as a building, a room as a room and a roof as a roof. The OGC CityGML 3.0 model brings together urban objects, geometry, meaning and relationships. Human-readable names and descriptions coexist with coordinates, identifiers, code lists, thematic surface classes and Levels of Detail. Translating all visible words indiscriminately can break the model even when the 3D scene still looks impressive.
If you need to localize CityGML 3.0 building attributes, translate CityGML room names, distinguish building function and usage, preserve semantic surfaces or validate CityGML GML encodings, this guide explains the relevant boundaries step by step. It is for urban planners, geospatial engineers, architects, heritage teams, city governments, educators, smart-city researchers and professional translators across countries. Its scope is specifically 3D urban information models; it complements rather than repeats general GML XML guidance or ordinary map-label translation.
We will examine CityGML’s semantic structure, the separation between conceptual model and exchange encoding, the revised Level of Detail approach, addresses, linked geometry, code lists, dynamic data and practical multilingual delivery. The principle remains consistent: translate approved human communication but preserve the geometric representation and the machine-readable identity of each city object. That protects simulations, analysis and public understanding while allowing residents and visitors to read the same city through different languages.
CityGML is an urban information model
OGC CityGML defines a semantic model for representing and exchanging virtual 3D cities with urban objects and their relationships. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A planning department shares a city model with researchers who need to distinguish buildings, bridges and vegetation. A translator treats the model as a collection of decorative objects and rewrites technical classes as ordinary narrative nouns. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Classify names and explanations separately from object types and technical references. Confirm that every translated edition preserves the same modeled city objects and feature identities. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
CityGML 3.0 differs from earlier editions
The version 3.0 model develops the representation of urban semantics and harmonizes geometry through common space and space-boundary concepts. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: An agency receives a 2.0 city archive and a newer 3.0 conceptual-model export from a different vendor. An implementation assumes all legacy 2.0 element structures and Level of Detail rules apply unchanged to 3.0. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Record model and encoding versions explicitly before transforming any language field. Test against the actual schema and consumer profile rather than a generic CityGML label. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
Conceptual model versus encoding
CityGML 3.0 distinguishes the underlying city information model from particular encodings such as GML/XML. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: A digital twin is exported in a GML encoding while another application uses a database representation. A developer assumes that every CityGML-compatible system uses exactly the same XML tags and extensions. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Identify the chosen encoding and its mapping to the conceptual model. Validate semantic objects alongside syntax for the specific delivery technology. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Urban objects have persistent identity
A building or road object may connect to municipal records, simulations and historical versions using stable identifiers. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A government asset registry links an office building to inspection and energy-performance reports. Translating those identifiers into readable words breaks cross-system correspondence. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Preserve the canonical object ID and supply localized display names through authorized attributes. Check identifiers, foreign references and related records after every language export. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Names are not classes
A building can have a public name, but its machine-readable thematic class is a different kind of information. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A university library has a culturally distinctive public name while remaining a building object in the urban model. Replacing a Building class token with a translated phrase can make the object unrecognizable to software. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Translate only designated name properties and leave the core object class stable. Verify that the building is still treated as the same class in downstream analysis. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
Spaces and boundaries
CityGML 3.0 uses concepts of spaces and space boundaries to organize the geometry and meaning of urban objects. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: A modeled interior room is separated from another by a wall surface represented in the model. An editor interprets SpaceBoundary as an English phrase to rewrite, accidentally changing semantic structure. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Preserve semantic class identities and explain them in localized documentation. Check that space and boundary relationships remain valid in a conforming viewer. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
Buildings and building parts
Urban models can represent a whole building and relevant constituent parts with separate identities or geometry. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: A complex public facility has multiple wings with their own entrances and local names. A translation that conflates a building part with the whole property can mislead maintenance and planning users. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Keep part-whole relationships and assign translations to the correct object levels. Inspect the model tree and compare localized labels to the authoritative part identifiers. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Building rooms
A room may be an urban semantic object with a specific function, location and enclosing interior geometry. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A museum digital twin contains galleries, equipment rooms, accessible toilets and visitor corridors. A translated room label can confuse a storage room with a public meeting room when cultural vocabulary differs. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Use a controlled room-name glossary, retaining functional codes and real access limitations. Review room-to-geometry associations and actual signage with facility staff. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Interior and exterior spaces
CityGML 3.0 can describe both indoor and outdoor structures within its Level of Detail approach. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A station model has a simplified outer form and comparatively detailed internal circulation areas. A translator assumes an indoor representation always implies the maximum geometric detail. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Describe what is actually modeled rather than claiming detail that the data lacks. Check each object representation and the limits of the viewing application. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
The revised Level of Detail system
CityGML 3.0 uses Levels of Detail 0 through 3; older LOD4 treatment is not carried forward in the same way. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: A city model offers several building representations for map overview, energy analysis and interior navigation. A manual labels all interior features as LOD4 and misdescribes a modern 3.0 dataset. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Retain LOD codes and explain their version-specific meaning accurately. Compare LOD claims with the official 3.0 model and the actual geometry present. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
LOD does not equal semantic richness
The newer model separates geometric detail from whether thematic surfaces and semantic classes are present. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: A simple building representation still carries a meaningful classification of its exterior surfaces. A translator assumes a wall surface can exist only at the same detail level used in older CityGML versions. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Keep geometric LOD and semantic decomposition as distinct concepts in explanatory text. Inspect object classes and available geometry rather than inferring one property from the other. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Wall surfaces are semantic types
A wall surface can be a categorized boundary of a building, not simply a text label printed on a wall. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A building-energy model differentiates walls from roofs and grounds for simulation. Replacing technical surface terminology with a colloquial word can obscure engineering and energy analysis meaning. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Preserve the technical class and translate its human explanation in a glossary or popup. Verify surface counts, class assignments and geometry associations. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Roof surfaces and orientation
Roof surfaces may carry geometries and thematic classification used for solar potential, drainage or energy analyses. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A municipal tool estimates rooftop solar exposure across several neighborhoods. A translated caption could confuse roof orientation with an administrative direction label. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Keep roof geometry, classification and measured orientation unchanged while localizing descriptions. Compare calculated orientation and surface classification against the original model. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
Ground and floor surfaces
Ground surfaces, exterior floors and interior floor surfaces have different semantic roles depending on the object modeled. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: A museum model distinguishes the ground-facing shell from interior walking surfaces. A casual translation of ‘floor’ can erase whether it means a level, a physical slab or a surface classification. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Specify the intended technical sense in the glossary and preserve class codes. Check multilingual terminology against actual navigation and engineering use. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
Windows and doors
Openings such as windows and doors may be separate semantic objects related to surfaces and spaces. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: A station interior model represents entrances, glazed openings and evacuation doors. A translation that swaps an exit label with a window identifier can create an accessibility or safety error. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Keep opening geometry and identity intact while translating verified public-facing annotations. Check object associations, sign text and safety direction with the facility owner. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Building installations versus furniture
Installations are typically permanent components affecting a building, while modeled furniture may be movable. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A hospital digital twin distinguishes fixed ventilation infrastructure from movable desks and seating. Using one translated term for both may confuse maintenance responsibilities and asset inventories. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Maintain the model distinction and use industry-approved target-language labels. Verify asset categories and equipment IDs with facilities management. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Function and usage are not identical
CityGML can distinguish the intended function of an object from how it is actually used, subject to the model and code lists. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A building classified for education temporarily houses administrative offices. A translator renders a category as a claim about current public access when it only states an intended function. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Keep original controlled values and explain the distinction in public copy. Check function, usage and access facts separately with current data. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
Classifications and code lists
Thematic attributes can use controlled values or externally defined lists rather than unrestricted language strings. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: Two municipalities share datasets classified under a common urban object vocabulary. Translating the code value itself may break cross-city comparisons and software rules. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Preserve the authoritative code and localize the display label or definition through a mapping table. Check code-list versions, URI references and source-to-target category equivalence. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
Addresses need an authoritative source
Building addresses may include local spellings, house numbers, postal codes and jurisdiction-specific formatting rules. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: An international visitor uses a digital twin to navigate to a named cultural center. A fluent street translation can become unfindable if the official postal address uses another form. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Keep address identifiers and official names, adapting display ordering only under an approved postal standard. Test geocoding, routing and signage matching for the selected locale. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Coordinate systems remain structural
3D city models depend on a spatial reference system and meaningful coordinate axes, including elevation treatment. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A city model combines buildings with terrain data surveyed in an established local coordinate reference. Changing CRS references or interpreting elevations as translated text can move assets or distort heights. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Preserve numeric coordinates and authoritative spatial reference metadata during localization. Compare control points and visible alignment between buildings, roads and terrain. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Vertical datum matters
Elevation values may depend on a vertical reference that is distinct from everyday notions of building height above the street. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A flood simulation relies on terrain elevations and ground-floor thresholds near a coastal district. A translated description can wrongly equate elevation with the number of building stories. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Keep height quantities, vertical datum information and semantic distinctions explicit. Check elevation fields, units and reference assumptions with geographic specialists. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
Geometry must not move for labels
Three-dimensional meshes, solids and surfaces encode geometry, not text that should be rearranged for line breaks. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: A landmark’s translated name crowds the viewport above a detailed roof surface. A translated place label may be longer, but shifting a facade vertex is a different physical model. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Adjust label placement in the viewer rather than altering model geometry. Compare geometry and topology for matched objects before and after the language edition. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
Topology and adjacency matter
Relationships between surfaces, rooms and spaces can support building analysis and indoor navigation. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: An evacuation simulation connects corridors, stairwells and exits through structured relations. Losing a link between two spaces may change connectivity even when the rendered model still appears continuous. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Preserve relationship identifiers and object references alongside geometric invariants. Test affected adjacency queries and routes using the original and localized model. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Appearances are a separate layer
Visual appearance may involve colors, textures and material representations rather than an object’s essential geographic identity. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A heritage model uses facade textures plus multilingual interpretive descriptions. A translated material name should not require replacing a texture asset or altering how the surface is assigned. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Keep appearance resources and links intact while localizing approved captions. Check that texture files and related assets still resolve after packaging. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
External texture references
Some renderings point to images or material data held outside the core geometry document. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A partner downloads a model but sees blank walls because the texture directory was omitted. A localized export may break relative file paths if assets are renamed or moved casually. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Deliver the complete asset package with stable links and clear language-independent resource IDs. Reopen the export from a clean location to verify every reference resolves. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
Transportation objects
The city model may represent roads, railways and other transport spaces with particular semantic roles. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: A metropolitan model shows a tram corridor, pedestrian paths and public road connections. Replacing operational road classifications with local-language adjectives can break simulations or thematic styling. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Keep transport structure and class identities fixed while translating public explanations. Test route and spatial queries as well as translated map labels. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
Bridges and tunnels
Bridge and tunnel modules describe infrastructure objects whose identity, topology and geometry are critical. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: A regional digital twin supports maintenance planning for elevated roads and underground crossings. An imprecise target term can conflate an underpass with a tunnel or a pedestrian bridge with a road bridge. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Preserve structural classes and IDs while reviewing domain terms with engineers. Check feature type, connectivity, level relationships and explanatory text together. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Vegetation and green spaces
Urban vegetation can represent trees, plant cover and related properties relevant to planning or simulation. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A city climate project studies street trees and surface cover across districts. A translated tree classification might be confused with a generic park name. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Keep species or class identifiers and measurements while translating educational captions. Compare planted objects, measured attributes and local terminology against the source. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Land use classifications
Land-use objects can carry formal planning categories whose meaning depends on local regulations. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A city shares development zones with overseas researchers and local residents. A dictionary-equivalent category label may imply different permitted uses in another jurisdiction. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Retain official land-use codes and provide carefully qualified target-language explanations. Verify categories with planning authorities and preserve effective dates. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
Water bodies and terrain
Urban models may include water and terrain features that provide environmental context for buildings and infrastructure. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: A flood resilience model contains canals, river surfaces and detailed ground relief. A text change that accidentally modifies terrain elevation or surface classification corrupts simulation inputs. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Protect geometric and measurement data and localize only approved descriptive annotations. Run basic surface continuity and location checks against the original model. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
City furniture and small assets
CityGML can represent urban equipment and furniture that are easy to misname across languages. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: A pedestrian-accessibility model contains benches, bollards, signs and street lighting. The phrase ‘city furniture’ may be mistranslated as indoor furnishings rather than street-level installations. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Define the technical category before choosing everyday target-language words. Check identity, installation position and functional descriptions with public works staff. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Generics and extensions
CityGML allows extension mechanisms and application-specific information beyond the core thematic classes. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A municipality adds an extension that records flood barriers and maintenance states. Unknown fields may hold critical local utility or cultural data rather than irrelevant translator noise. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Preserve extension namespaces and technical keys until their owner authorizes editable strings. Validate with the supplied extension schema and consumer application. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Application Domain Extensions
ADEs can add domain-specific concepts to the CityGML information model using defined extension mechanisms. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: An energy-analysis platform uses extra attributes for building systems and thermal calculations. Rewriting extension names or controlled values as prose can sever integration with specialist tools. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Treat ADE data as versioned technical schema rather than an unrestricted note field. Compare model classes, extension IDs and values before and after localization. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
Dynamic data and observations
CityGML 3.0 supports integration with time-varying information, including sensors and simulations through suitable mechanisms. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: A digital twin shows temperature measurements and projected energy use under different scenarios. A translator may treat one simulated reading as a current confirmed physical condition. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Translate captions with clear observation dates, units and uncertainty while protecting data values. Check timestamps, source identity and scenario labels with the data owner. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
Time and scenario qualifiers
A city model can describe planned, existing, demolished or simulated conditions depending on its data source and temporal framework. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: A public engagement model includes a new bridge that has not yet been constructed. Removing a qualifier such as ‘proposed’ turns planning information into an apparent fact about the current city. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Keep status and scenario metadata stable and retain qualified wording in every language. Compare project phase, reference time and release provenance after translation. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Indoor public navigation
Localized indoor descriptions can help visitors find galleries, stairs, accessible routes and services within complex spaces. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A museum navigation application shows rooms and corridors modeled at varying detail. A translated exit instruction can be dangerous if it implies a door is accessible when the operational record says otherwise. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Use facility-approved directions and keep connectivity, room IDs and accessibility codes unchanged. Test actual visitor routes with on-site staff and target-language readers. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Public digital twin interfaces
Many users experience CityGML data through an application that chooses labels, panels, legends and layer names. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A city dashboard displays a building name, function and energy score in several locales. A correct translated attribute can remain invisible if the interface still reads the source-language field. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Coordinate field selection and UI localization with the data contract without changing core city objects. Check search, filtering, display and accessibility on deployed platforms. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
Bilingual naming needs context
A single site may require a registered name, a romanized form and a visitor-friendly translation. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: A historic market square is known by different names across local languages and tourism maps. Replacing the authoritative record with a translated brand name can undermine matching with official signage. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Store name roles distinctly and document which authority or community recognizes each form. Review maps, street signs and public records together before releasing localized labels. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
Heritage and archaeology use
3D heritage models combine physical reconstructions with interpretation, uncertainty and source documentation. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: An old town digital reconstruction includes hypothetical roofs and measured surviving foundations. Translation that removes ‘reconstructed’ can wrongly imply every wall is an observed archaeological fact. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Preserve evidential status and distinguish interpretation from surveyed geometry. Check provenance and uncertainty statements with heritage specialists. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
Energy performance use
A building-energy application can rely on semantic surfaces, orientation, area, materials and simulation parameters. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: An international city project estimates heat loss for a range of building archetypes. Translating a field that actually stores a code or unit can alter how energy inputs are interpreted. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Preserve numeric and controlled values while localizing descriptions of assumptions and results. Compare simulation inputs and representative outputs after translation. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Emergency planning use
Emergency teams may rely on building layouts, connectivity and operational information represented or associated with the city model. Remember that a city model is both a representation of physical space and an organized description of what its objects mean. A localization choice must respect both roles, even when only a short label is visible.
Consider a practical case: A campus model shows emergency stairs, assembly areas and closed sections. A polished target-language caption does not prove a route is accessible or an exit is operational today. A failure might become visible only after analysis or an export to another system. The absence of an obvious rendering defect should not be treated as proof that the intended urban meaning survived.
Keep asset identities and topology stable while obtaining current operational approval for public instructions. Test routes and timestamp-sensitive notices independently of geometry checks. Keep evidence with the release showing the source object, allowed edits, invariant values and reviewer. That allows later users to reproduce the acceptance decision without relying on memory.
Multilingual extraction needs object paths
A translation handoff should identify the object, property path, source text, locale and relevant naming policy. The practical question is what the reader needs in another language and what the technical consumer requires to remain identical. Those requirements overlap, but they are not interchangeable.
Consider a practical case: A vendor translates hundreds of building and room labels exported from a city model. A plain spreadsheet without IDs can place a correct room description on the wrong building part. In a multilingual public service, this can mislead residents or downstream engineers even when the translated sentence sounds plausible. The best prevention is to preserve the original source and make each intended change auditable.
Use immutable object keys and field-specific allowlists for extraction and reinsertion. Reconcile every localized string with the exact semantic object that owns it. Ask a model specialist and a target-language reviewer to evaluate different aspects of the same example. Both perspectives are needed because technical conformance alone does not guarantee honest communication.
Round-trip model validation
A localized model can be structurally valid but still change meaning through mistaken label-to-object mapping or unsupported code replacement. A city model can look correct in a viewer while carrying a changed classification, relationship or measurement. Therefore linguistic improvement must be evaluated against semantic and geometric integrity together.
Consider a practical case: A partner imports a translated model into a building-energy tool and an indoor viewer. A screenshot of the city scene cannot prove that semantic classes, relationships and quantitative attributes survived. Another user may make a consequential decision from the affected model, so the error is not just editorial. Structural, scientific and public-communication checks need separate attention.
Compare identities, classes, geometry, units, topology and intentional language differences in parsed data. Run recipient-side analytical and search tasks with representative objects. Document which CityGML version and receiving application were used in the test. Cross-organization interoperability requires precise assumptions rather than a claim that everything is universally compatible.
A controlled release sequence
An effective workflow records versions, freezes the city model, inventories editable fields, translates with context, reconstructs a locale edition and tests it. This is not merely a wording preference. Different software may use the same object for urban planning, indoor navigation, engineering or historical interpretation, so the field’s role matters before translation begins.
Consider a practical case: A public agency releases an accessible multilingual urban model for residents and research partners. Skipping model-version identification or a domain reviewer can make a translation subtly wrong despite passing an XML parser. The model’s semantic relationships are often more important than its appearance in a screenshot. Test what the object is and how it connects to other objects, not just what its caption says.
Keep source model, schema, glossary, translation approvals, validation and rollback packages versioned separately. Require geographic, semantic and linguistic sign-off before publishing each language edition. Record unexpected differences as issues to investigate, not as harmless side effects of translation. This keeps a language update from becoming an unapproved change to a digital representation of the city.
Frequently asked questions
Is CityGML only an XML file format?
No. CityGML 3.0 distinguishes a conceptual urban information model from particular encodings. A GML/XML encoding is important, but the same concepts can inform other technical representations. Confirm the actual implementation before choosing extraction rules. Use the responsible urban data authority’s definition where the technical and everyday senses differ. A trusted release should explain the distinction rather than hiding it behind a smooth but potentially misleading translation.
Can building IDs be translated?
No. Stable object identifiers are for cross-record and cross-system reference. Translate approved names, descriptions and interface labels without replacing IDs or related object keys. Use the responsible urban data authority’s definition where the technical and everyday senses differ. A trusted release should explain the distinction rather than hiding it behind a smooth but potentially misleading translation.
Does CityGML 3.0 use LOD4?
CityGML 3.0 revises the earlier Level of Detail approach and uses LOD0 through LOD3, allowing indoor and outdoor representations within that framework. Do not carry older LOD4 assumptions into modern documentation without checking the model version. Use the responsible urban data authority’s definition where the technical and everyday senses differ. A trusted release should explain the distinction rather than hiding it behind a smooth but potentially misleading translation.
Are WallSurface and RoofSurface ordinary words?
They are semantic class concepts that can affect analysis. Their class identities should remain stable, while user-facing explanations may be localized according to an approved terminology map. Use the responsible urban data authority’s definition where the technical and everyday senses differ. A trusted release should explain the distinction rather than hiding it behind a smooth but potentially misleading translation.
Can I translate every room function?
Only when the room-function value is intended for human display or an authorized multilingual mapping. Controlled codes and formal classifications may be consumed by software. Preserve them and provide translated labels separately. Use the responsible urban data authority’s definition where the technical and everyday senses differ. A trusted release should explain the distinction rather than hiding it behind a smooth but potentially misleading translation.
How can I verify that the 3D city is unchanged?
Compare feature IDs, classes, code-list references, geometry, topology, coordinates, dimensions and relationships before and after the edition. Then independently review the translated descriptions and test the real receiving digital twin. Use the responsible urban data authority’s definition where the technical and everyday senses differ. A trusted release should explain the distinction rather than hiding it behind a smooth but potentially misleading translation.
Conclusion: the same city, made readable to more people
The strongest multilingual CityGML edition gives people understandable building names, addresses, room labels and explanations without changing the represented physical or semantic city. Keep object identities, class systems, coordinates, Level of Detail meanings, code lists and relationships stable. Review translated narrative text with domain experts, and test technical invariants with the actual consuming systems. A city model becomes more accessible only when the new language expands understanding without pretending that the geometry, operational state or evidential certainty has changed.
Official standards and companion reading
The OGC CityGML standard, conceptual model and user guide are the authoritative references for the 3.0 approach. The adjacent GML localization guide addresses XML syntax and spatial feature encoding in general; it should not be mistaken for the full CityGML semantic model.
OGC CityGML standard; OGC CityGML 3.0 Conceptual Model User Guide; OGC CityGML 3.0 Conceptual Model Standard; Companion: GML 3.2 XML localization