rationalising the city through data

Collaborator: Yun Kong Sung, Jordon Saunders

There are vast amounts of information available on the urban environment: suburb population densities; domestic, industrial and commercial land usage; soil type, depth and acidity; pollution indexes and air quality, to name but a few. 25 datasets were available on Christchurch covering a wide spectrum of environmental data from soil acidity, soil type and depth to pollution and wind direction. These datasets were attained through datasets such as high resolution maps and pattern recognition algorithms.
We categorized the data into two categories. Existing composition of all the visual and less dynamic data consisted of existing buildings, foliage, and roads. Also, existing field conditions consisted of ecological data such as wind intensity, soil acidity, and pollution. These were mapped based on a height map based on their location on a 156x139 grid, with their z axis corresponding to their amount.
When approaching computing for urban planning, our stance was not necessarily to optimize the city for a particular field condition, rather to negotiate a balance between a composition which was necessary and a field which needed to improve – in other words the fields and new composition were rationalized.
To start to utilize this data, an agent based system was deployed. Each agent had a set of operations that could be parameterized [slide 5] – thus separate agent types could be designed. We designed cluster agents which would form into groups and create built clusters, pheromone agents which could form new routes, and green agents which would network and form clusters of foliage.
Based on the amount of built environment in the city, a series of agents were deployed throughout the city. Agent parameters were designed to reinact three scenarios: pollution rationalized, wind rationalized, and flood rationalized. In this example [slide 6], the pollution optimized city paramatizes green networks to attract toward high amounts of the pollution field, and the built cluster agent and pheromone route to repel away from this field. To the right we see the resultant new composition of the city generated from the agent simulation. The next iteration [slide 7], agent types are split into their sub categories – for example built environment split into housing, industrial, and commercial - and run through a simulation with 4 times larger resolution producing an even more detailed map.
Moving into greater detail, the agents could be programmed to perform a number of three dimensional operations[slide 8]. This is guided by a set of rules which allow, for example adjacent building agents to move and shift to accommodate right to daylight, maximum cantilever for buildings and can allow maintain distances away and towards road networks. In processing this data into three dimensions, the rule set was applied into a live model – low, medium, and high densities of the simulation were sampled from the previous level of simulation as the base condition and the 3d agents were deployed. At this scale specifics such as emergent agents could also be deployed – consisting of public amenities and programs such as fire stations at certain proximities.
What resulted form these simulations were a series of maps of the city generated through computing and information which reorganized the city of Christchurch from the largest scale of the city down to the specifics mass models and program type for building.