Fixed Guideway system mileage in the US

Using multiple charts to show this, in a 'Powers of Ten' sort of way.

Chart 1:  Most of what we have is highways; which has been oddly steady. While miles keep going up, you can see it's pretty flat after 1980.

Chart 2: Class 1 railway  has seen big declines, while oil pipelines are up.

Chart 3: Navigable channels steady as a rock, Amtrak had a big decline in '98, but much less than in '80. (Not sure this is a bad thing--so many of their long distance routes are terrible).

Chart 4 shows transit:

Commuter Rail - Bit of a surprise there--I would never have guess so many miles of commuter rail were being built, or had grown so steadily. Retrospectively predictable--it's the most auto-compliant form of transit; as it's based on a combination of downtown stations and suburban park and ride, and so the most suited for most of America. I'd like to slander it as never generating TOD, but then I recognize my ignorance on the topic.

Heavy rail is flat -after the DC Metro, BART, and Atlanta's MARTA, I'm not sure who has build any (excepting tiny mileages in NYC and LA). Heavy rail may be a dead technology (barring existing networks). No need for it--it's utility lay in being able to share track with freight ROW, a niche now (largely) filled by commuter rail. Given the greater acceleration provided by EMU (Electric Multiple Units) I suspect we may see a convergence between 'heavy rail' and 'commuter rail' over time.

Light rail--we've generated about 52 miles/year over the last 30 years. If we look at post 1990, 62 miles/year; post 2000, 72 miles a year. But post 2010....62 miles a year. Light rail may be reaching its limits--all the metros suited for it have it, and all the corridors within the metros have it. I expect we'll continue to see organic growth in those networks, but the number of light rail 'New Starts' is going to be pretty limited. My rule of thumb is a metro needs about a 2 million population for light rail to be feasible.



*Data for all charts for the 80's and 00's straight line interpolated from the 5 year increments of the data source, which flatters the regression, but not likely much.