A new framework for very large-scale urban modelling

Michael Batty,Richard Milton

doi:10.1177/0042098020982252

Abstract

The generation of ever-bigger data sets pertaining to the distribution of activities in cities is paralleled by massive increases in computer power and memory that are enabling very large-scale urban models to be constructed. Here we present an effort to extend traditional land use–transport interaction (LUTI) models to extensive spatial systems so that they are able to track increasingly wide repercussions on the location of population, employment and related distributions of spatial interactions. The prototype model framework we propose and implement called QUANT is available anywhere, at any time, at any place, and is open to any user. It is characterised as a set of web-based services within which simulation, visualisation and scenario generation are configured. We begin by presenting the core spatial interaction model built around the journey to work, and extend this to deal with many sectors. We detail the computational environment, with a focus on the size of the problem which is an application to a 8436 zone system comprising England, Scotland and Wales generating matrices of around 71 million cells. We detail the data and spatial system, showing how we extend the model to visualise spatial interactions as vector fields and accessibility indicators. We briefly demonstrate the implementation of the model and outline how we can generate the impact of changes in employment and changes in travel costs that enable transport modes to compete for travellers. We conclude by indicating that the power of the new framework consists of running hundreds of ‘what if?’ scenarios which let the user immediately evaluate their impacts and then evolve new and better ones.

Highlights

The generation of ever-bigger data sets pertaining to the distribution of activities in cities is paralleled by massive increases in computer power and memory that are enabling very large-scale urban models to be constructed
The digital computer was invented in several places in the decade spanning the Second World War, and as soon as it emerged from its scientific and philosophical origins, it was applied to large-scale problems involving predicting human futures as well as accounting and transactions processing in business
We are moving to an era where hundreds if not thousands of solutions might be explored in much the same way as online search systems operate, enabling massive numbers of queries to be handled which can only be absorbed by users in real time

Summary

Introduction

The generation of ever-bigger data sets pertaining to the distribution of activities in cities is paralleled by massive increases in computer power and memory that are enabling very large-scale urban models to be constructed. The model framework proposed here is different from most other previous applications in that the focus is on building spatially extensive urban land-use transportation models that can be run almost immediately, providing the user with rapid feedback with respect to the evaluation of scenarios.

Objectives

Results

Conclusion