Using Integrated Urban Models to analyse the built environment

My previous article discussed cities as complex systems and talked about the need to understand the constraints the built environment places on daily activities, this post explains how Integrated Urban Models (IUMs) can help do this.?

On the back of tech developments in the last decade, including better hardware, more (open) data, better (open) software, and wider prevalence of coding capabilities, it is possible to make bigger more complicated models and analyse them faster, but the real question is “Why do this?”.?

In a slightly counter-intuitive way, these more complicated models create outputs that are more easily understood by non-experts, such as: “How walkable is each street?”, “How many jobs can be reached within half an hour of every house?”, or “Where is there a good choice of schools within walking distance?”. Although the models may be more complicated to make, these more intuitive outputs make them well suited to supporting strategic decisions.????

IUMs combine data on the big physical systems in a city; the street and pedestrian networks, land uses, and public transport. They use the spatial network to connect uses, which means that from the point the data is combined and connected in the model, it starts to consider how people move through the actual city, rather than using abstract definitions of distance or arbitrary administrative boundaries. The model incorporates a topological representation of the public transport network, with all street, pedestrian and transport links assigned time costs and transport modes. This means that from all parts of a city, we can see how far it’s possible to move in a defined time using any mode of transport.??

?There is an important conversation about different data types, and it is vital to understand both what data represents and to have a theoretical framework to position it. In this case it’s useful to think of built environment systems as the variables in a science experiment, and the result of the experiment as the activities that occur in cities.??

There is an ever-growing amount of live, device and sensor-generated activity data, and this tells us “What” is happening, “Where” and “When”, for example; 1,000 people are at this location at 8:00 am. It is useful but it doesn’t explain “Why” it happens, it only gives us the result of the experiment, not the information on what the variables were, or how they were combined.??

Without the data to describe and explain the variables of the experiment - in this case the way the physical systems of the built environment work individually, and in combination - we can’t understand “Why” the result happens, nor the type of interventions that could affect activities and longer-term outcomes.??

?By dealing with the hard systems of the built environment, specifically the elements that influence where people might be going, and which routes they are more likely to take, IUMs allow us to explain “Why” activities happen in particular locations, for example; 1,000 people are in this location at 8:00am because it’s on the best-connected street linking x homes to y jobs.?

?In 2020 we built an IUM of Great Britain and used it to create a Walkability Index. This assigns land uses to categories and measures the number of (and distance to) categories, reached by moving through the street and pedestrian networks, from every street and property. This is not a measure that considers public realm quality, street lighting, active frontage, or street trees, but it considers how easily a person can move through the actual built environment, around urban blocks and what they can access. Good quality public realm encourages people to walk, but if the big physical systems aren’t in place, and it’s not possible or practical to walk, then improving the public realm has limited impact.?

?Before we made our model of Great Britain we developed and tested the Walkability Index on a smaller scale. In Exeter, as part of our work on the NHS Healthy New Towns, the Public Health department linked our Walkability index to health outcomes including Obesity. More recently, we combined components of the Walkability Index with socio-economic data to predict Car Ownership, Active Travel Mode Share and Public Transport Mode Share across the Thames Valley. What was really interesting in this example was that the association between income deprivation and lower levels of car use wasn’t as strong as against the number of jobs or retail services within 15 minutes’ walk of every property. So, while in areas with lower car use there are often lower levels of income, there are also people who live in these areas who could afford a car but choose not to. And this can only happen if the built environment makes it possible to walk to work, or the post office, or the bank.?

The IUM outputs such as the Walkability Index allow us to start to see how the built environment works, but it’s also important to consider that groups of people experience things differently or may have difficulty taking part in some activities. To understand better how people and place overlap, we coordinate the IUM with census data boundaries. This means that we can see where there might be a more vulnerable population in an area where you need a car to reach essential services. Or whether there’s likely to be an older, socially isolated population in a segregated part of the city. By including open data sets we can look not just at access to education and health facilities, but also at their capacity and quality of service.?

This ability to see place risk at the same time as demographic risk then allows specific responses to be developed. For example; there might be parts of a city where it’s expensive and difficult to intervene physically to reduce car dependence, but alternative mitigating interventions could be possible; subsidising taxis for low income residents of these areas, providing e-bike services, encouraging clean (e)car share clubs or providing charging infrastructure.??

As discussed in the previous article, as well as creating conditions that influence daily activities, the big built environment systems are essentially fixed for long periods of time. In the context of climate change and scarce resource (including funding), we need to make better use of what we have, and this means better understanding the opportunities and constraints of the built environment. Tools such as the IUM can help us tailor and retrofit interventions to our urban environments to support positive long-term outcomes.?

In my next article, I’ll show how we’re packaging the outputs of these models into tools that further remove technical barriers so that non-experts can use them more easily.?