This is an edited extract from Decoding the City, a collection of writing on data-driven urbanism from MIT’s SENSEable City Lab. This piece examines how data can be used to map residents’ activity in cities, and how the resulting analysis can be used to guide transport policy and even map residents’ cultural, religious and social activities.
In early 2007, a group of Google Earth users made a curious discovery in San Diego. Panning across the newly available satellite imagery, the armchair Magellans noticed a set of structures that formed an inexplicable shape when viewed from above; a Nazi swastika.
The find went viral – well before the concept of going viral existed – and was picked up by major news outlets. It was quickly discovered that the complex, whose surrounding roadways are coincidentally named after WWII-related sites, was actually built in 1967 by the US Navy.
Now visible to anyone with an Internet connection, the base’s plan view caused great public outcry, and the resulting political pressure led to a $600,000 reconstruction project to unmake the abhorrent shape. “We don’t want to be associated with something as symbolic and hateful as a swastika,” a spokesperson said.
The Navy claims the exact form and orientation of the structure was wholly unintentional, and simply the consequence of a humiliating planning oversight. But whether deliberate or not, it’s clear the project’s planners, architects, and builders were not anticipating a god’s-eye-view perspective on the finished project.
As ridiculous as the whole episode was, it highlights a particularly powerful idea: New ways of seeing can drastically refigure our understanding of, and relationship to, place.
Of course, not all urban data sets are as intrinsically visual as satellite imagery. Many require the context of urban spatiality to understand and operationalize them – i.e., they gain salience and power when seen through the city. The de facto example of this power – thanks in large part to Edward Tufte’s proselytization– is on display in John Snow’s 1854 cholera outbreak map :
Basic germ theory had not yet been accepted by the medical field, but by mapping cholera deaths across Soho, Snow was able to tangibly communicate the idea that cholera was transmitted not through infected air, but rather through contaminated water and food.
Now, in an age where data is everywhere, the prevailing notion is that important stories sit somewhere within all data, and consequently, the task of analysis and representation is to simply uncover these stories. And thus, the march toward data-absolutism continues, instilling a tendency to cast meaning where it simply doesn’t exist – to identify or construct false patterns in the great static that is big data.
What does this all mean in the context of the city? On a fundamental level it underscores the fact that urban datasets are powerful and capricious; they encapsulate countless dimensions at myriad physical and temporal scales, and a wide gulf exists between possible relationships and actionable results. Our struggle is now in reclaiming a sense of legitimate, verifiable meaning from the morass, i.e., reorienting our processes of modeling, simulation, and representation to distill value while also keeping validity in check.
The following project, in which we attempted to understand transport patterns in the Saudi Arabian city of Riyadh, illustrates our approach to achieving this delicate balance.
An urban traffic system for the city of Riyadh
Rapid economic and demographic changes throughout Saudi Arabia are posing new challenges and opportunities for the Kingdom. Of particular concern is the explosive growth of the nation’s capital, Riyadh, where development is quickly outpacing transportation infrastructure – between 1987 and 1995, automobile trips increased at the rate of 9 percent per year.
The Urban Traffic System project aims to develop an innovative, highly dynamic urban traffic system to address the mobility challenges specific to the city. To this end, the project is based on creating an alternative to traditional intelligent transportation systems by taking advantage of the digital traces of our everyday lives – specifically, mobile phone usage – to create models for mobility analysis, intervention, and planning, for policy makers, planners, and development professionals, as well as for the citizens of Riyadh themselves.
We partnered with telecom companies in the region to collect roughly one month of total phone activity across the country, where mobile phone penetration is, astonishingly, over 198 per cent. We aggregated nearly 100 million daily network connections, assigned to more than 10,000 unique cell towers.
Seeing Riyadh through Data
This image shows mobile phone activity through color, transparency, and height across Riyadh. This visualisation projects strong portrait of the social character of the city. With the inclusion of satellite imagery as a base map, we arrive at a unique view of how the social rhythm of the city is expressed over built form.
Watching the oscillations of the activity landscape, a unique character emerges – we see that the city really doesn’t come alive before noon, and peaks in aggregate activity around 6.15 pm. With a careful eye, we can begin to pick out subtle regional delineation: the residential neighborhoods to the south-west and northeast of the downtown core activate well before the rest of the city, and experience the strongest interhour fluctuations throughout the course of the day.
By overlaying our results on the geography of the city, a number of interesting relationships are revealed. Most strikingly, the clusters correlate very closely with the main arteries of the city. Mobility communities seem to be heavily reliant on the street network itself, underscoring the city’s overall dependence on highway infrastructure.
From the planning perspective, one of the most meaningful stories we can glean from this data is an how an individual moves through the city, which, at an aggregate level, describes one of the most vital components of urban analysis: origin-destination matrices. Traditionally, O-Ds, which are vital for transport network optimisation, are constructed through onerous census surveys that are conducted every five to ten years. The process is long and costly, and when completed, only provides a rudimentary snapshot of travel demand.
By collecting and filtering each user’s mobile activity, however, we are able to estimate a population’s travel demand in terms of origins and destinations of individual trips. We’ve shown that these approximated O-D flows hold a strong correlation to census estimates; however, this approach includes the added benefit of capturing travel demand which includes seasonal variations and hourly fluctuations.
Returning to John Snow’s cholera map, are we able to reveal hidden facets of social life by affixing our data to the structure of the city? And consequently, what can this transposition further teach us about the character and composition of urban form in Saudi Arabia?
One cultural phenomenon that is unique to the Arab world is the daily call to prayer. While collecting our data, we found an intriguing pattern in mobile activity distributions that was unlike any other country or city we’ve analyzed before: at various points in the day activity would simply drop off for around thirty to forty minutes before picking back up to its typical trend. These inactivity “valleys” were actually the result of these daily prayer times.
Millions of Muslims across the country put down their phones to turn and face the holy city of Mecca to give prayer five times a day. Shops and businesses essentially close down for roughly twenty to thirty minutes while the religious police – the Mutaween – surveil the streets in the hopes of sending all loiterers to the nearest mosques. To our surprise, our activity distributions very closely capture this behaviour.
The timing of these calls to prayer depend on the position of the sun in the sky, and thus, by looking at western, central, and eastern regions, we are able to see the prayer times moving across the country.. This presents another series of intriguing questions.
This sudden dip in cellular activity is identifiable when applied to the geography, but can we quantify and map the intensity of the disruption and show which areas are most affected by calls to prayer? Does it follow the density distribution of mosques? Leading from this, can we detect and illustrate how prayer time disruptions are expressed through mobility? Do average trip lengths shorten during prayer windows? Lastly, can the intensity of disruption serve as a proxy for regional religiosity?
We will continue forward with all of these questions in mind. And through this careful, back and forth negotiation into and away from the spatial frame, we hope to arrive at a collection of representations that capture new ways of seeing both the city and the social forces operating there.
Decoding the City is edited by Dietmar Offenhuber and Carlo Ratti and was released by Birkhauser in August.
This article is from the CityMetric archive: some formatting and images may not be present.