Square-Mile Street Network Visualization

The heart of Allan Jacobs’ classic book on street-level urban form and design, Great Streets, features dozens of hand-drawn figure-ground diagrams in the style of Nolli maps. Each depicts one square mile of a city’s street network. Drawing these cities at the same scale provides a revealing spatial objectivity in visually comparing their street networks and urban forms.

We can recreate these visualizations automatically with Python and the OSMnx package, which I developed as part of my dissertation. With OSMnx we can download a street network from OpenStreetMap for anywhere in the world in just one line of code. Here are the square-mile diagrams of Portland, San Francisco, Irvine, and Rome created and plotted automatically by OSMnx:

OSMnx: Figure-ground diagrams of one square mile of Portland, San Francisco, Irvine, and Rome shows the street network, urban form, and urban design in these cities

The top row depicts the late 19th century orthogonal grids of Portland, Oregon and San Francisco, California. Portland’s famously compact walkable blocks are clearly visible but its grid is interrupted by the Interstate 405 which tore through the central city in the 1960s. In the bottom row, the business park in suburban Irvine, California demonstrates the coarse-grained, modernist, auto-centric urban form that characterized American urbanization in the latter half of the 20th century. In stark contrast, Rome has a fine-grained, complex, organic form evolved over millennia of self-organization and urban planning.

Remember these are all at the same scale: one square mile. Compare the block sizes and intersection density in Portland to those in Irvine. Compare the orthogonal grid in San Francisco and the modernist simplifications of Irvine to the complex mesh of pedestrian paths, passageways, and alleys comprising much of the circulation network in Rome.

OSMnx: Figure-ground diagrams of one square mile of Manhattan New York, Paris France, Tunis Tunisia, and Atlanta Georgia shows the street network, urban form, and urban design in these cities with Python in the style of Allan Jacobs Great Streets and Nolli maps

Above, we see New York, Paris, Tunis, and Atlanta. Manhattan’s rectangular grid originates from the New York Commissioners’ Plan of 1811. You can see Broadway weaving diagonally across it. At the center of the Paris square mile lies the Arc de Triomphe, from which Baron Haussmann’s streets radiate outward, remnants of his massive demolition and renovation of 19th century Paris. At the center of the Tunis square mile lies its Medina, with a complex urban fabric that evolved over the middle ages. Finally, Atlanta is typical of many American downtowns: fairly coarse-grained, disconnected, and surrounded by freeways.

One final example: square miles of central Boston, Jumeirah circle in Dubai, suburban northeastern Sacramento, and Osaka, Japan. What can you tell about these neighborhoods from their street patterns?

OSMnx: Figure-ground diagrams of one square mile of Boston Massachusetts, Dubai UAE, Sacramento California, and Osaka Japan shows the street network, urban form, and urban design in these cities with Python in the style of Allan Jacobs Great Streets and Nolli maps

To compare urban form in different kinds of places, these visualizations have depicted some downtowns, some business parks, and some suburban residential neighborhoods. These patterns also vary greatly within cities: Portland’s suburban east side looks very different than its downtown, and Sacramento’s grid-like downtown looks very different than its residential suburbs. These visualizations, rather, show us how different urbanization patterns and paradigms compare at the same scale.

Several more examples and all the code to reproduce these diagrams are in this notebook in OSMnx’s GitHub repo. You can adapt it to visualize any street network anywhere in the world. For more info, check out the paper about OSMnx.

34 thoughts on “Square-Mile Street Network Visualization”

    1. It’s a business park in the western part of Irvine, near the airport. Most of Irvine comprises the loops-and-lollipops style of suburban form. This Irvine business park is more coarse-grained and more orthogonal, and useful to compare to other types of urban form.

      1. This is very cool, but don’t you think controlling for land use is important? How can you compare an industrial development pattern to residential or downtown areas? The Irvine example isn’t apples to apples.

        1. Hi Marta. Yes indeed, hence my paragraph toward the end of the article: “To compare urban form in different kinds of places, these visualizations have depicted some downtowns, some business parks, and some suburban residential neighborhoods. These patterns also vary greatly within cities: Portland’s suburban east side looks very different than its downtown, and Sacramento’s grid-like downtown looks very different than its residential suburbs. These visualizations, rather, show us how different urbanization patterns and paradigms compare at the same scale.”

    1. Sid, nice visualizations. The code to set road widths is at the bottom of his notebook linked above. Your visualizations would be easier to read – esp. Valparaiso. It’s just one more line of code.

    2. Me encanta! My problem is I know nothing about coding though I would like to have two of these images of Toronto (area around first Canadian place) and Boston (area around Northeastern University)! I want to create a 3d printed version of these two cities/areas for an architectural installation. Can anyone help me out? Is there a video on how I could do this? Any help is greatly appreciated! Fantastic subject for a dissertation. Congrats!

  1. Thank you!
    This is a useful way to contemplate why cities work, or don’t. Wondering if you’ve developed understanding of how these organic vs gridlike forms have affected humans.

  2. I’m totally new to data visualization, but I was wondering if there was some way to show a USGS topographical behind the streets; perhaps to show how geographical features influence (sub)urban grids. I think the topo lines would have to be faint in order not to interfere with the most important message of the visual. But, strong enough for the viewer who wants to extract more information.

  3. Pingback: Milltir sgwâr

Leave a Reply

Your email address will not be published. Required fields are marked *