Synopsis

The challenges resulting from global urbanization have prompted researchers to seek ways in which to better understand the behavior of existing and emerging cities. A proliferation of research perspectives on the economic, social and technical aspects of cities has led to broad and diverse spatial and temporal scales of urban studies. An enduring challenge for rigorous analysis is the generalization of urban resource consumption behavior over widespread geographic, climatic, development and technological contexts. To date, detailed studies for a relatively small set of cities have been accomplished while the vast majority remains unstudied. We present a typology of cities based on urban resource and socioeconomic data and information. We use a variety of statistical methods to arrive at the rational initiation for the construction of a typology that captures an extensive range of cities globally.  Employing scaled data to establish groupings of like resource intensity reveals significant commonalities that approach the construction of a working typology. While the results are provisional and will require additional investigation, the typology attempts to establish a working framework for further development. 

 

Methodology

This study develops a typology of urban metabolic (or resource consumption) profiles for 155 globally representative cities. Classification tree analysis is used to develop a model for determining how certain predictor (or independent) variables are related to levels of resource consumption. These predictor variables are: climate, city GDP, population, and population density.

Classification trees and their corresponding decision rules are produced for the following major categories of material and energy resources: Total Energy, Electricity, Fossil fuels, Industrial Minerals & Ores, Construction Minerals, Biomass, Water, and Total Domestic Material

Consumption. A tree is also generated for carbon dioxide emissions. Data at the city level was insufficient to include municipal solid waste generation in the analysis. Beyond just providing insight into the effects of the predictor variables on the consumption of different types of resources, the classification trees can also be used to predict consumption levels for cities that were not used in the model training data set.

Urban metabolic profiles were also developed for each of the 155 cities, resulting in 15 metabolic types containing cities with identical or almost identical levels of consumption for all of the 8 resources and identical levels of carbon dioxide emissions. The important drivers of the differences in profile for each type include the dominant industries in the cities, as well as the presence of abundant natural resources in the countries in which the cities are the main economic centers.