Wednesday, June 27, 2012

Citywide Syntactic Map

The citywide syntax map is coming along.  It measures network integration, which researchers have found a correlation (and likely causation) between degree of network integration (red = highly interconnected) and things like density, pedestrian counts, land values and, inversely, crime.  Immediately below is London

As with all mathematical measures, there is a good bit of abstraction going on.  However, I like the software program, and moreso the intellectual path the UCL researchers are on because it is more of a measure of underlying patterns that form cities and explain the things we see on the ground in a predictive manner rather than most simple statistical measures.  In other words, it is focused on the underlying processes and coda of cities rather than more superficial measures (which can then be compared for possible relationships).

 Above is Dallas citywide framework, which will all be filled in, measured for global interconnectivity.  Or, a heatmap showing areas best connected up to 4 turns away.  You'll notice the glowing red sections are drastically diminished compared to London, which is predictable.  Perhaps more interesting is where the darker sections are in Dallas.  Downtown, theoretically, shouldn't be the densest area of the city (and it isn't - thus the indirect relationship between building height / density / interconnectivity - a sign of a system in distress).  I suspect this is related as it goes a long way to explaining why downtown nearly died completely until DART linked it up with the surrounding region, downtown at the hub.

I've written and talked about this.  At one point in time, Dallas had a naturally emergent conical shape to its skyline that is directly related to interconnectivity.  Interconnectivity = availability of interactions.  That availability = demand, which is then responded to by the market with supply, building space usually in the form of height (as the small block size of interconnected networks forces that space upwards).

The height is greatest where demand is highest.  Ever decreasing sizes of buildings hug up to the next and so on to the edge.  There is a natural order to it.  You'll notice in places like London, Paris, and DC they tend to have extremely high costs in the center due to height restrictions, which manifests itself in height pushed outside of the core.  

At this point in the picture above, developers began building ever taller high-rises to respond to the growing demand.  At the very same time, Dallas was building a series of highways in and around downtown, sapping this demand while supply was being added.  Hence, downtown Dallas had far more supply than demand.

Above is the syntactic map from 1945 as put together by a UTA student and sent to me.  You can see how well stitched downtown and east Dallas are, resulting in the glowing red areas that brought about high-rises on Main and Ross.

It is also worth noting that I plan on adding DART lines to this, but the effort will take some time as it isn't merely an exercise in drawing, but also linking and unlinking as the train largely exists on a separate plane from the road network (except when it itself acts as a barrier, which is a corrosive effect except at immediate station areas).  

Back to today, above is when I change the settings of the calculations to scale for local interconnectivity, or how connected a place is to things two turns away.  The details are less important than if you squint and you begin to see little yellowish blobs, these show theoretically, walkable areas in terms of how interconnected small areas are within themselves, useful for identifying neighborhoods and neighborhood-scaled centers (or at least where they should be - not taking into account for identifiable larger factors).

As the graphic immediately below illustrates, there are two gradient hierarchies at work: global and local.  Mapping these seperately allows the differentiation, essentially pulling the two overlapping maps apart to examine individually.

I suspect it will begin looking a lot like the walkscore heat map, except in inverse.  Walkable (green) areas will correlate with highly integrated areas of the 2-dimensional road network map (yellow/orange/red).

When cropping for areas fully mapped we get the following maps:

A couple interesting things begin to emerge, particularly when examining the table with the integration values street by street.  The highest, or most integrated street yet mapped, is Preston Road.  Therefore, it isn't difficult to understand why so many real estate companies locate in Preston Center.  Downtown migrated north.  And it did so because the road network moved it there.

Here are the following scores of some of the primary, most integrated streets:

Preston:  8.09782
Mockingbird (I suspect the residential nature preserved by Highland Park pushes much of this energy south towards Knox, McKinney, and Oak Lawn):  7.12074
Greenville:  7.07212
Davis:  6.91496
Carroll (most surprising):  6.57049
Live Oak: 5.92579
Main: 5.46883
Commerce:  5.48214
Elm: 5.22184

It's also very interesting and telling that Ross scores higher in East Dallas than it does in downtown.  Between Main scoring lower than the "suburban" neighborhood scaled streets above and Ross's, further underscoring the effect the downtown loop had on vitality and land values in downtown.  

Furthermore, the overall strength of East Dallas isn't surprising, given its large contiguous area of historic grid uninterrupted by freeways between 75 and White Rock Lake.  It should come as no surprise that by the 2010 census the densest census tract in the city had hopped the east-west divide that is 75.

This also tells us how much latent potential there is on streets like Live Oak, Carroll, and Davis in comparison to what is currently on the ground.  The gap between potential and existing is great.  The next step is solving all or as many of those barriers causing the rift as possible.

Lastly, when we convert the map to emphasize local integration, the thing that jumps out is the area at Marsalis & Kiest in South Dallas.  By the looks of it, it is the most locally integrated area of the entire city.  No, really strong commercially viable roads like the way Greenville, Preston, and Lovers jump out, but an area that holds together well.  It may be isolated from its surroundings, but on further aerial inspection, it isn't a huge surprise that much of the housing stock is in pretty good shape.  Not many of the lots are vacant.

However, again on closer inspection you see how awful those roads are (Kiest/Marsalis) and badly overscaled they are given their place within the network.  They should be neighborhood scaled "main and mains" not unlike Greenville.  Instead, because we seem to think South Dallas needs massive infrastructure as a form of economic development we get 6 lane divided (emphasis on the DIVIDE) roads that carry 8,000 cars per day, or about a 1/4th of capacity.

You want economic development.  Give some of that excess Right-of-way back.  Particularly at the commercial intersections like Marsalis & Illinois.  Roll the recaptured right-of-way into an incentive for investment.  Narrow and calm the intersections to improve pedestrian connections in a manner that is context sensitive and conducive to safe, vibrant, desirable places.  Improve pedestrian and bike accessibility to these n'hood centers as well as DART to the east on Lancaster.

The median household income of each of the census blocks in the area are in the 20,000s.  Connections are opportunity and empowerment without making people not in cars feel like second class citizens or worse, roadkill.  As we know, this is the heart of the most dangerous US congressional district in the country for pedestrians.  To be safe, we force the citizenry into cars that often take up 40-50% of their income.  Money far better saved, invested, or put into entrepreneurial endeavors like new, local, small businesses.

That's one way you can start immediately "growing South."