Experiments in Procedural Cartography
UC San Diego Parking Operations Study
A project I have been working on recently got me thinking about the interactions of cartography, procedural modeling, and 3D visualization. Procedural modeling helps planners to communicate and evaluate their ideas in a data driven, efficient, and scenario focused manner. It accelerates the iterative design and evaluation process that scenario planning demands and yields enriching visualizations that enhance the communication of planning scenarios.
Garsdale Design, for example, has developed a CityEngine ruleset for Abu Dhabi using their parcel data that allows their Urban Planning Council to leverage context-specific design criteria and zoning analytics. These procedural rules help to accelerate the iterative design process and enable planners to articulate design elements that would hard to communicate at this scale with a more manual modeling effort.
However, recently I got to think about how mapping, procedural modeling, and 3D visualization might work together to create a better understanding of parking demand. As a topic of study, parking has gone from a rather obscure assumption made by many planning processes to one of the most heavily analyzed elements of urban form. Traditionally, parking demand is typically communicated via 2D occupancy maps.
While these are useful, they do have their limitations:
Continuous data has to be broken down into bins that may not be that helpful. When you need to break a percentage occupancy into 5 logical classes, you can encounter trade-offs in what you can communicate about parking demand over time. You want all the time periods to have the same classification, but really each time period will have its own distribution that might not be best represented by your chosen breakpoints. In the map above, you know which lots are 70%-90%, but do you know which ones are closer to 70% or 90%? You can't really tell by looking at the map. This becomes even more irksome when some time periods have observed overflow parking you want to communicate to a client.
There is no easy way to communicate the number of spaces. We can label how many parking spaces are in each facility, but it can be difficult read such a map when there are many lots that need to be symbolized. Even more deceptive is the fact that surface lots can have larger footprints and fewer spaces relative to a parking structure: potentially drawing the viewer's eye away from key facilities.
Recently, while on a plane to San Diego (need something to do, right?), I developed a ruleset specifically to visualize parking demand for this project. Here is the same project at the same time period generated using this ruleset:
The rules that drove this visualization addresses the shortcoming of the traditional parking demand map by:
Communicating the number of spaces by extruding a parking facility's height so that it is relative the total number of spaces at the lot.
Use color and the "fill" of the extrusion to communicate percentage occupied on a continuous gradient that makes changes between time periods very apparent.
Provides equally spaced cuts in the geometry to provide an intuitive sense of the actual percentage occupancy being displayed without a legend.
Create a specialized symbology in the case of overflow parking where the observed parking occupancy exceeds 100% (requirement came from a future project).
In many ways this visualization method not only provides an interactive way to view the data, but helps address some of the limitations of the 2D parking occupancy map. In addition, the data preparation that is required for the 2D map is the exact data preparation required for this type of visualization. Since this rule's first application, it has been applied to multiple projects, including locations as grand as some of the United States' National Parks.
Parking Study for Yellow Stone National Park iterated on the first rule to communicate overflow in parking occupancy.
In the future, I hope to combine this type of visualization with transportation demand management scenarios and investigating different technologies to display the same concepts (three.js/geospatial frameworks). However, it got me thinking of different ways that 3D and precise control of geometry can be used to explore planning data. Much of this is not new, but provides another example of how procedural and 3D can help enhance planning communication.
Thank you to Elliot Hartley at Garsdale Design for contributing some supporting ideas and images. You can see more of Garsdale Design's work at their home page.
If you have any thoughts on other planning communication strategies I would love to hear from you.
