Fraser Morgan and David O'Sullivan have recently written a paper entitled "Using binary space partitioning to generate urban spatial patterns" which was presented at the 4th International Conference on Computers in Urban Planning and Urban Management (CUPUM).
The abstract of their paper is:
Creating a realistic representation of urban spatial patterns is often problematic. Using existing development as a starting point for testing models of urban growth may introduce inherent biases based on the city chosen. To address this issue, we have used binary space partitioning (BSP) trees to quickly generate a representation of the cadastral spatial pattern seen in cities. This approach, which has links to quadtrees and binary trees used in computer science, includes the standard topologic elements (leaf nodes, children, root node, ancestors, descendants and levels) while also incorporating the spatial element of territory. The substantial flexibility in the resulting urban structure belies the small number of parameters used to control the creation of the tree.
This BSP tree approach was developed to create many similar and realistic urban spatial patterns on which an agent-based model of the purchase, subdivision, building and disposal behaviours of property developers would operate. The motivation for implementing a BSP tree approach was the ability for the developer agents to be able to understand, analyse and enact the mechanism of subdivision upon the urban environment. Through the formalised node structure, the BSP tree approach also enables the straight forward implementation of developer territoriality, neighbourhood attractiveness and spatial metrics for analysis."
Click here to read the full paper which is very interesting. To accompany the paper there is also a NetLogo model is just a small aspect of the wider ABM model that Fraser is working on which provides an interesting representation of an random urban cadastral pattern upon which the ABM does its work.
The abstract of their paper is:
Creating a realistic representation of urban spatial patterns is often problematic. Using existing development as a starting point for testing models of urban growth may introduce inherent biases based on the city chosen. To address this issue, we have used binary space partitioning (BSP) trees to quickly generate a representation of the cadastral spatial pattern seen in cities. This approach, which has links to quadtrees and binary trees used in computer science, includes the standard topologic elements (leaf nodes, children, root node, ancestors, descendants and levels) while also incorporating the spatial element of territory. The substantial flexibility in the resulting urban structure belies the small number of parameters used to control the creation of the tree.
This BSP tree approach was developed to create many similar and realistic urban spatial patterns on which an agent-based model of the purchase, subdivision, building and disposal behaviours of property developers would operate. The motivation for implementing a BSP tree approach was the ability for the developer agents to be able to understand, analyse and enact the mechanism of subdivision upon the urban environment. Through the formalised node structure, the BSP tree approach also enables the straight forward implementation of developer territoriality, neighbourhood attractiveness and spatial metrics for analysis."
Click here to read the full paper which is very interesting. To accompany the paper there is also a NetLogo model is just a small aspect of the wider ABM model that Fraser is working on which provides an interesting representation of an random urban cadastral pattern upon which the ABM does its work.
No comments:
Post a Comment