Computational Design Laboratory

Urban Form Types: Jinmo Rhee’s PhD Proposal

PhD Proposal presentation announcement:

A Computational Method for the Identification and Comparative Analysis of Urban Form Types: A Case Study of Rust Belt Cities

Jinmo Rhee, PhD Candidate in Computational Design

Date: Dec 5, 2022

Time: 3-5 pm

Location: MMCH 107


For this research, I propose to develop and evaluate a new computational method for constructing and analyzing large urban datasets, enabling new insights into a city’s spatial, urban-economic, land use, and demographic dimensions. The method is predicated on an alternative way of representing cities that takes urban spaces — rather than buildings or blocks — as a city’s constitutive unit. In contrast to conventional data methods for urban form analysis, this method will preserve the morphological features of urban spaces at different scales in order to capture their spatial characteristics. A deep neural network model trained on the dataset will cluster form data by feature, yielding morphologically distinct urban types. By examining characteristics of each urban type one can visually and numerically address configurational characteristics of the urban space. When types are induced based on multiple cities, unique and common types can be identified by comparing the morphology features of individual urban spaces. Analyses of morphological traits derive from an inductive investigation of changes, trends, and structures of constituents for differing city situations, namely, emergence, growth, shrinkage, declination, and dissipation. Individual urban space-centered analysis enables the identification of unique and common urban types and their characteristics, providing spatial insights into restructuring strategies of city space via urban economy, land use, and demography analysis.

To evaluate this method, and the proposed restructuring of urban data as an aggregation of individual urban spaces surrounded by urban forms, I propose to analyze morphological traits of urban spaces across multiple cities. A case study is designed which includes four Rust Belt cities: Pittsburgh, Detroit, Buffalo, and Cleveland — cities that once experienced great growth, and now face a need for shrinkage and restructuring in response to a decreasing population and economic decline. The case study aims to 1) identify urban types and their distribution in Rust Belt cities; 2) examine morphological traits of unique urban types and fabrics that only appear in the cites; and 3) investigate potential roles for the types in redefining city boundaries and restructuring urban spaces in a shrinking city. From the study, I expect to develop a new framework to identify various urban form types and analyze their characteristics with an aspect of urban form data that represent individual urban spaces as constituent units of city space. I further expect that the findings from the framework could inform urban planners and designers about alternative morphological and spatial understandings of city space and provide for strategies to revamp city structures.

Advisory Committee

Dr. Daniel Cardoso Llach
Associate Professor
School of Architecture
Carnegie Mellon University

Dr. Ramesh Krishnamurti
Emeritus Professor
School of Architecture
Carnegie Mellon University

Dr. Bhiksha Raj Ramakrishnan
Language Technologies Institute
School of Computer Science
Carnegie Mellon University

Proposal draft

Author: Daniel Cardoso Llach
Category: News