Photo: James Llamas.

The notion that transit riders in large swaths of the city will be left high and dry is demonstrably false, and, of course, the vast majority of riders will see significantly improved service.

Two weeks ago, the Kinder Institute's Kyle Shelton profiled an intriguing study from the UT Center for Sustainable Development called "Identifying Transit Deserts in Texas Cities." In the study, researchers Junfeng Jiao and Aaron Nichols sought to compare estimates of transit demand to the amount of service supplied in five major Texas cities.

As a follow-up, the researchers ran their analysis again on the New Bus Network METRO developed through the Transit System Reimagining project, a complete overhaul of Houston's local bus system set to be implemented this August. They concluded that while the new system better balances transit supply and demand in most of the city, deficiencies will actually increase in some high-demand areas including Gulfton and Sharpstown in southwest Houston.

As a planner at Traffic Engineers, Inc. I had the privilege of working with METRO staff and industry experts on the reimagining plan. The project has received quite a bit of attention around the country, and it's encouraging to have well-respected researchers evaluating its effectiveness in different ways. Coincidentally, Reimagining was assailed early on by critics who alleged it would create "transit deserts," presumably large areas with no transit service whatsoever. As it turned out, we were able to maintain service within a half mile of over 99.95 percent of existing riders (their boarding location, at least, which is all we actually know), including most of those in very low-ridership and hard-to-reach locations.

The notion that transit riders in large swaths of the city will be left high and dry is demonstrably false, and, of course, the vast majority of riders will see significantly improved service.

Naturally, the return of the "transit desert" concept grabbed our attention. In this case, however, the study isn't looking at the complete absence of transit service but rather the gap between supply and demand as measured by adults without vehicles.

Nevertheless, it was troubling to read the conclusion that transit supply was actually being reduced in two of the densest residential areas in Houston. The goal adopted by the METRO board of directors to guide the System Reimagining Plan was to increase ridership by devoting a greater share of local bus hours to high-ridership service. Our strategy for doing that was to design straight, simple routes with buses coming frequently along the high-density corridors in the region. Did we completely miss the mark in Sharpstown and Gulfton?

Let's take a closer look at the study's methodology. As mentioned, transit demand was measured by subtracting the number of household vehicles in each Census block group from the number of adults living there. While carpooling and sharing of vehicles among family members could mean that not every adult without a vehicle will necessarily be looking to ride transit, this does seem like a reasonable way to estimate the magnitude of demand. The chosen language, "transit dependent" has some issues, but that's a topic well covered by one of TEI's partners on the project, Jarrett Walker.

How about transit supply? The study combined seven factors to generate the supply score per acre in each block group. The first four relate to transit accessibility rather than the service itself and wouldn't have changed between the existing system and the New Bus Network:

  • Total length of sidewalks (omitted in Houston due to lack of data)
  • Total length of bike routes
  • Total length of low-speed roads (measured for Houston as non-thoroughfare streets)
  • Intersection density

The other three did change with the network redesign:

  • Number of transit stops
  • Number of weekday transit trips
  • Number of transit routes

Reading through the study, the assumption on bus stops presents the first problem.

"Since information on the location of transit stops is not available yet," the authors say, "points were placed along the future routes roughly every quarter mile and multiplied times two to account for a stop on either side of the road."

Is this reasonable? In fact, all existing bus stops will be retained unless the street in front of them will no longer have service in the new network. This is a fate that will befall only a handful of stops each in Sharpstown and Gulfton, mostly located on end-of-line loops that will no longer exist. Practically all stops along major streets will remain.

So is the quarter mile spacing an accurate assumption? As many a weary bus rider can tell you, METRO local buses tend to stop much more often than that. The mile of Bellaire Boulevard centered on the intersection with Fondren Road in Sharpstown has seven bus stops in each direction. If the researchers placed four stops in each direction along this mile, they incorrectly assumed a 40 percent reduction in the number of stops. The typical spacing of stops along METRO's local routes is more like one-sixth to one-fifth of a mile, so the quarter-mile assumption would provide a significant undercount. This likely explains part of the supply reduction alleged by the study.

But, perhaps more significantly, is the number of bus stops even a good measure of transit service quality? Most people on a bus stopping six times each mile would say no. Certainly, a base number of stops need to exist to allow people along the route to access the bus, but after a certain point (about quarter-mile spacing) adding stops has diminishing returns. According to the assumptions underlying the study, however, adding more stops is always better. This is an aspect of the methodology that should be reevaluated.

The last two metrics - the number of transit routes and weekday trips - are the two that reimagining primarily addressed. Sure enough, Sharpstown in particular will see fewer bus routes and fewer weekday bus trips in the New Bus Network than it does today. But does this necessarily mean riders in that area will experience worse service?

Without getting deep into the theory of network planning, it's safe to say that the primary measure of a successful transportation system is to be able to reach as many destinations (jobs, schools, grocery stores, social, recreational, romantic partners, entertainment opportunities) as possible within a reasonable travel time. In transit, travel time is the sum of walking, waiting, and riding time aboard the first vehicle as well as any connecting vehicles required to make the trip. As a short-term plan constrained by existing infrastructure, Reimagining primarily sought to improve travel time by reducing wait time with more frequent service and reducing ride time with simpler, straighter routing and more logical connection points.

Did we succeed? Below is a matrix showing the change in travel time among 30 representative locations around the region. Coincidentally, the Sharpstown location (Bellaire & Fondren) and the Gulfton location (Bellaire & Renwick) cited in the study were both included in the matrix. The columns showing the changes in travel time from those points to the 29 others are in the purple boxes.

Both columns are overwhelmingly green, meaning travel times to most places in the region will be faster in the New Bus Network than they are today. This was achieved by consolidating the bus service into fewer routes that come more often. Bellaire & Fondren currently has six bus routes, none of which run with seven-day frequency better than 20 minutes. Most run more like every 30 minutes, and one disappears altogether on Sundays.

This means that riders in Sharpstown either live their lives in consultation with a bus schedule or spend significant time waiting at bus stops, especially when traffic inevitably holds up a bus. Strangely, three of the routes in Sharpstown are ultimately destined for Downtown Houston, but two pick up on the eastbound side of Bellaire Boulevard and one on the westbound side. More routes and even more buses are not necessarily better if they're hard or inconvenient to use.

In the New Bus Network, Bellaire & Fondren will only have four routes but two of them will run every 15 minutes or better, seven days a week, and a third will run with that frequency on weekdays. Trips to most major destinations in the region can be made with zero or one connection to another route that likely also runs frequently. Riders can show up to the stop with confidence that a bus is coming soon, even if they just missed one or a particular trip is running late.

The story in Gulfton isn't quite as rosy. One of the frequent routes that made the column in the travel time matrix so green, the 32 Renwick/San Felipe, had to be scaled back to 30-minute frequency when its schedule came in over budget. It will still connect to new destinations, however, and the area should see an overall increase in bus trips. The reduced supply of service the study identified in Gulfton must primarily be a result of the bus stop assumption.

If the metrics used in the UT study could lead to incorrect conclusions about transit service supply, what measures should be used instead? I would suggest that weekday bus trips be replaced by total weekly bus trips since most people - even we transit riders - have places to go on the weekends. Inadequate weekend service is a great way to force people into car ownership who would rather do other things with their household budget.

We saw how basing the index on the gross number of routes could be problematic. At the same time, number of routes can be an indicator of the number of destinations reachable from a particular area. Perhaps double-counting routes that meet a certain frequency threshold like 15 minutes would yield more accurate results.

All of this isn't to say that the parts of town called out in the study can't use more transit service. They undeniably can. The methods developed by Nichols and Jiao hold great promise in helping transit planners identify areas where improved service is most needed. For instance, on this basis the 32 Renwick/San Felipe should be prioritized for a frequency increase when the budget allows.

The study has also helped trigger long-overdue conversations about how we set goals for our transit system and measure whether it is achieving them. However, the before-and-after comparison of Houston METRO's Reimagining plan revealed some shortcomings in the methodology that will need to be cleaned up in future applications of this work if it is to best capture the important aspects of transit service supply.

James Llamas is a Senior Associate at Traffic Engineers Inc.