One of the arguments often made for rail over bus is that rail gives passengers a feeling of certainty: those two strips of steel are a promise that a train will be along soon, going exactly where you think it will go.
Do the 24 light rail systems in the United States actually fulfill that promise? They offer dramatically different levels of service. At 7:30 am on a weekday, some lines operate every 30 minutes, and some every 5. That difference is huge for a passenger: at a 5 minute frequency, the average wait for a train is only 2 1/2 minutes, and waiting will be an insignificant part of your trip; at a 30 minute frequency, your day can be ruined if you miss a train. In the transit industry, 15 minutes is generally considered “frequent”: at that level of service you can just show up to a station without consulting a schedule. Remarkably, there are three light rail segments in the United States that don’t meet that level of service even during weekday rush hour. On a Sunday morning at 7:30, more than two thirds of the systems have segments that aren’t frequent, and three systems don’t offer service at all.
There’s a clear pattern among the systems below, ordered by how many passengers they carry per mile of line. The most successful systems, at the top, have the highest frequencies. That’s actually a reflection of two things. The first is that as ridership goes up, higher frequency is needed to carry the loads. Every system has a limit on train length; if a system can accommodate only 3 car trains, and a three car train every 15 minutes is crammed full, the way to fit more people is to run three cars trains every 10 minutes. The second is that more frequent services are more convenient and will therefore draw more passengers. Thus, increasing frequency will actually increase total ridership.
The contrast between frequencies, especially Sunday morning frequencies, often reflects the type of destinations a systems serves. These maps are all to the same scale, and the geographic scope of these systems varies widely. Many of the newer systems, such as Denver, Salt Lake, Sacramento, and Dallas, are designed to connect suburban park-and-ride lots to jobs in the urban core. These systems tend to carry mainly home to work trips for office workers, so ridership on these systems tends to peak heavily at rush hour. Low off-peak demand leads to lower off-peak frequencies. Older systems in Boston and San Francisco, as well as new systems in Houston, Buffalo, and Jersey City, stay within the urban core, connecting walkable neighborhoods to a variety of destinations, including universities, hospitals, shopping, and entertainment. These systems attract riders all day, every day, increasing overall ridership and justifying more offpeak frequency.
Ultimately, frequency is as much a reflection of philosophy as of raw numbers. Some cities treat frequency as a basic promise, running trains every 15 minutes or less even if they are full. Some cities — notably Los Angeles and Minneapolis — extend this philosophy to bus as well as rail, specifically promoting frequent networks. Other try to minimize operating cost, running longer trains less often to reduce the labor costs. In other words, some cities start by providing all-day service and building on that for the peaks; others focus on peak service and provide some minimal service offpeak. Neither decision is correct — it’s a matter of attitude to what the purpose of transit is. (Jarrett Walker at Human Transit discusses this very well.)
The vast variation between these systems comes despite the fact that all of use the same technology, and some even the same exact trains. When we speak of transit only in terms of mode (heavy rail, light rail, streetcar, BRT, bus) and not level of service or destinations served, we’re missing the most import aspects of what makes transit useful.