Developing Common Data Structures for Urban Modeling and 3D Visualization

Participants

This initiative is intended to provide a 'common ground' for projects that are interested in collaborating on a common data structure for Open Source modeling, visualization and community participation. Initial collaborating projects include:

• NSF: Behavioral and Geometric Modeling (Aliaga/Benes, and Waddell/Jordan)
• MTC: Sustainable Communities Outreach (Waddell, Aliaga)
• UCTC: CommunityBuilder? (Waddell)
• MRPI: Sustainable Transportation (Waddell, Walker, Rodier)
• FHWA: Urban Continuum (Pendyala, Chiu, Waddell)
• Swiss NRP 65: Sustainable Urban Patterns (Schmitt, many others)
• European Research Council: SustainCity (Axhausen, Waddell, many others)

Other projects are invited to collaborate in this as well.

Objectives

• Multiple projects share the need for a high-resolution representation of the urban landscape, from the level of individual buildings, to parcels, blocks, streets, up to neighborhoods, cities and regions.
• There is also a need for algorithms to read, manipulate, and write to these data structures for simulating changes to buildings, streets, and patterns of urban development and transportation and environmental conditions over time.
• We also need to adopt common APIs to make it easy to interface current and future models and visualization functionality in ways that are efficient (fast) and modular

Urban Hierarchies and Graphs

We propose to represent urban areas with connected, dual representations: one reflecting the hierarchy and adjacency relationships among buildings, parcels, blocks, and zones; the other representing the connected street and transit nodes and edges connecting them. The streets enclose blocks which enclose parcels which enclose buildings. Similarly, we can connect parcels and the buildings and occupants (households and businesses) to the street nodes in order to support transportation related queries such as which activities can be achieved at different locations, and how long would it take to reach those activities using different modes of travel? At the most microscopic level, and mostly for visualization, we also need to relate building facades to buildings, and vegetation, pedestrian and vehicle visualizations to the spatial landscape, including parcels, streets and sidewalks.

Three Views of the City: Existing, Planned and Future

We will need to design the data structures and algorithms around three views of the city:

• Existing: this view should represent as accurately as possible the existing buildings, parcels, streets and activities within an urban area. We may want to build on or use OpenStreetMap? as a point of departure for the data, since it is global in scale and open source, with community-based editing. This is a good model for how we want to proceed. We need to extend the data to include terrain, buildings, transportation network (roadway capacities and speeds and the transit network), and provide a user interface for anyone to be able to contribute to improving the data. A web-based map interface with data forms is a likely candidate.

• Planned: this view focuses on the data that describe the regulations governing entitlements to the land: what can be developed, at what density, where? Also, other policies that influence development, including financial tools such as taxes and subsidies. And finally, alternative plans for transportation and other infrastructure. For this task, we not only need to clarify the data structures, but also methods of handling user involvement that will generate many 'scenarios' containing different, possibly related versions of these data. For example, the adopted General Plan for City X might be the default for the land use regulation component. But a user may create a variant of this data that makes a reference to the default data (by inheritance, for example), and overrides certain elements in the data that make it different from the default. This needs to be stored as a new scenario, and may be done compactly by referring to the original as the 'parent' scenario and creating a new 'child' scenario.

• Future: this view focuses on the results from combined behavioral and geometric simulation that use the existing urban area data as starting conditions, and the planned scenario as a set of influences (particularly on constraints for development and transportation and other infrastructure that shape accessibility and opportunities for development), and generate a plausible future representation of the urban area. Here we need to blend the existing city with simulated changes, so the 3D visualization must contain both existing buildings and new, simulated buildings, as well as any changes to local streets, and assumptions embedded in the scenario for new arterial roads, highways and transit stations and service.

User Interfaces

The different views of the city (present, planned and future) suggest at least three different kinds of user interaction:

• loading existing data and interacting with it both to visualize it and to analyze it; including 2D maps, 3D visualization, indicators (e.g. walkability, emissions, housing prices...)
• interacting with the existing data and creating scenarios: land use regulations, changes to the transport system (and other infrastructure), public services (schools, parks, public safety, etc), and financial policies (how to pay for it all, and who pays)
• interacting with the future city, using indicators, maps, and 3D interactive visualizations; having ways to express preferences over alternative scenarios (voting methods)
• there may be 'official' scenarios and plans that are adopted by public agencies and which will be used to guide investment
• there may be 'unofficial' scenarios generated by citizens, students, researchers, and advocates
• one could also imagine a game-style interface allowing the user to rapidly experiment with the planned and future city, and a social networking approach to searching for consensus.

Open Questions

• How to store/manipulate raster data efficiently in a manner that is compatible with PostGIS raster data?

-- PaulWaddell - 10 Dec 2010

• streets_and_parcels.png: