Using Distance Profiles to Identify Peer Cities

Once you leave the Eastern Seaboard public transportation in the United States is a local affair. The type of service you find in a city depends on the interplay among several factors including total population, population density, geography of roads and terrain, demographic patterns, development patterns, local and state politics and values, and local agency funding structure.  In this post, I use a tool published by the US Census Bureau called Distance Profiles to analyze potential peer cities.  The Distance profile provides a cross section of Metropolitan Statistical Area (MSA) populations that allows us to visually identify cities with very similar population distributions, geographic size, and density distributions. Using this tool we can make an informed comparison of Louisville’s transit system and those of true peer cities.

Fig. 1a Portland vs Louisville Population Profile. PEER CITY? – NOT! (click for larger image)

Fig. 1b Portland vs Louisville Population Weighted Density (click for larger image)

Once we identify our peer cities we have to be careful. All of our peer cities’ transit systems are somewhat different and it doesn’t necessarily mean one city’s public transit system is more or less effective.  Measurements of “effectiveness” must be based on the public transit values that each local community chooses for itself. Cities always make tradeoffs between coverage and ridership within a fixed budget. A city that values coverage will design a network that tries to deliver public transit opportunities to every corner of the service area.  A city that values ridership will concentrate their resources on the high volume corridors to try and deliver maximize ridership. The former provides lots of access to service but with very low frequency.  The latter leaves many areas unserved but provides much more frequent service along selected corridors.  All cities try to balance these goals based on additional values including social service concerns, the desire to  influence development patterns, mitigation of pollution, and  congestion reduction, and fall somewhere along a continuum from maximum coverage to maximum ridership.    We can also look at a peer city’s budget and facilities which will tell us more about the community’s public transit values and perhaps operational efficiency.

We will address these ideas in a later post, but before we can do that, we need to identify our peer cities based on their Distance Profiles.  Then we can evaluate their transit systems and infer the community values illustrated by their transit system choices and decide what if any elements of that system we can emulate.

Population Profiles

The Distance Profile includes two metrics: population by distance from city hall and population weighted density by distance from city hall. The results for each MSA are plotted on line graphs where each point represents a circular band of land one mile wide with each successive band radiating outward from the city center.  The geometry of this metric results in each successive band having a greater area. The resulting graph has a peak at the band with the greatest population or density then falls off as the bands contain more and more rural areas. Take a look at Louisville’s graphs.

Fig. 2a Louisville KY Population Profile (Click for larger image)

Fig. 2b Louisville KY Population Weighted Density Profile (Click for larger image)

Louisville’s first population band (fig. 2a – Point 0) is a circle with a radius of 1 mile from city hall and has a population of about 9400. The point at mile 5 (x axis) represents a band between 4 and 5 miles from city hall that has a population of about 78,000. The largest population band occurs at point 9 or the band that is between 8 and 9 miles from city hall (beyond I-264 and inside I-265) and has a population of about 98,000.  Keep in mind that the entire population of Louisville MSA is “under the line.” The profile is a graphic that shows us the area where most people live and the geographic extent of the MSA. In the Louisville MSA, two thirds of the population lives within 12 miles of the city center.

The population weighted density chart tells a different story. Population weighted or “perceived” density is not standard density.  Standard density is calculated by dividing the number of people by the total area (acres, sq.miles, etc.) of the place that contains them. There are about 756,000 people in Jefferson County KY and the area is 380 square miles.  The Standard density is 756,000 / 380 = 1,989 people per square mile.  Standard density doesn’t consider large areas where nobody lives like the airport, industrial zones, malls or parks.  Population weighted density gives more weight to census tracts where more people live and the result is a metric that tells us more about how dense it “feels” to live in a place. Louisville’s average population weighted density is 2,477 people/sq. mi. (fig. 2a) For our purposes it gives us a clue about our market for transit in relation to other cities with a similar density in populated places. To identify peer cities we need to consider both the population and density profiles.

We often hear about the marvelous transit system in Portland, OR. But when you look at Portland’s profiles compared with Louisville’s you immediately see that comparing the two cities is really not reasonable because the profiles are quite different. Portland has a million more people than Louisville and substantially greater weighted density from the urban core to 22 miles out to the exurban communities. It is classic comparison of apples to oranges. We can model their public transit values but the profiles show that the market for our transit system and the solutions that fit are significantly different. (fig. 1a and Fig 1b above)

Peer Cities

To begin the process of identifying Louisville’s peer cities I first selected cities that had MSA populations between 75% and 125% of Louisville’s population of 1, 237, 851 (43^rd largest in the US).* The population range of the selections is from 843,746 (Tucson) to 1.58 million (Providence). The MSA is the broadest measure of a city’s population and is defined by a core city with a relatively dense population and the surrounding area that has significant economic ties to the core.  The MSA extends far beyond Jefferson County. This process resulted in 14 potential peer cities, 5 of greater population and 9 with lower populations. (table 1)

Table 1. Louisville’s Potential Peer Cities

(L) = larger Cities; (S) = Smaller Cities; Note: The central city names are used instead of the official MSA name.

*Based on 2010 Census numbers that differ slightly from the population on the charts.

The next step was to visually identify the cities whose population and density profiles were similar to Louisville’s.  It was easy to eliminate some. For example let’s compare Louisville and Milwaukee (pop. 1.55 million). (Fig. 3)

Fig. 3a Milwaukee, WI Population Profile (Click for larger image)

Fig. 3b Milwaukee, WI Density profile (Click for larger image)

You can see that the population profiles do not align well.  A much higher proportion of the population lives within the first 7 miles of the urban core in Milwaukee.  The density profile reinforces this.  There is substantially more density in the urban core than in Louisville.  Their transit system network requirements would therefore be substantially different.  There are similar contrasts when comparing Providence, Hartford, New Orleans, Buffalo, and Rochester.

A different result can be seen if we compare Louisville with Memphis (pop. 1.32 Million). (Fig 4) While not as extreme as the obvious differences in Milwaukee, it is clear that Louisville is somewhat denser in the urban core and less sprawling than Memphis even though their populations are very similar.

Fig. 4a Memphis, TN Population profile. (Click for larger image)

Fig. 4b Memphis, TN Population Weighted Density profile. (Click for larger image)

Ultimately I narrowed it down to 5 peer cities, three of which most closely resemble Louisville’s population and density profiles. To separate the best peer cities I calculated the average population weighted density for each of the cities out to mile 20 and the subtracted Louisville’s average density from the results.

The five cities whose profiles most resemble Louisville are Jacksonville, Richmond, Raleigh, Salt Lake City and Tucson.  But the three that match Louisville best both visually and arithmetically are Jacksonville, Richmond and Tucson. (Figs. 5, 6, and 7 below)  By looking at what is working in these three cities we can be more confident that it would work in Louisville because of the very similar density and population distribution.  There are other factors to be sure.  It is also true that Memphis, Raleigh, Salt Lake City and Oklahoma City are very similar to Louisville in many ways and it is prudent to pay attention to what they are doing but Jacksonville, Tucson, and Richmond are our colsest peers when looking at population and density profiles.

Fig. 5a Jacksonville, FL Population Profile.

Fig. 5b Jacksonville, FL Population Weighted Density profile.

Fig. 6a Richmond, VA Population profile.

Fig.6b Richmond, VA Population Weighted Density profile.

Fig. 7a Tucson, AZ Population profile.

Fig. 7b Tucson AZ Population Weighted Density profile.