As a form of public transportation, bikeshare systems have one major catch: The bikes seldom circulate themselves in quite the way planners would like. If users traveled around town in all directions, at all times of day, in relatively equal numbers, docks would empty and refill naturally. None would ever be totally empty. None would ever be completely full.
Of course, this is not how people travel in the real world (and it is not how cities are built). In Washington, commuters flood out of residential neighborhoods in the morning (emptying docks there), many aiming for the same few blocks downtown (where the docks are invariably full). In New York, riders descend on Penn Station during rush hour; they congregate around Union Square at night.
As a result, cities have to "rebalance" bikes by, well, truck. Someone must come along periodically and rearrange the supply to meet the shifting demand. This is a significant expense for bikeshare systems (not to mention an asterisk against their reputation for low-carbon transportation). And it's been a particular obstacle for New York's year-old and cash-strapped system.
To grasp what the challenge looks like there, the Spatial Information Design Lab at Columbia University has created several good data visualizations using Citi Bike data averaged over weekdays during October of last year. In this GIF, parts of town with the greatest concentration of activity – this includes people taking bikes and dropping them off – light up over the course of the day:
The hot spots at 11 p.m. are very different from those at 5 a.m. The same picture, sped up:
All this activity reflects both arrivals and departures. This next GIF separates the two. Areas in blue are parts of town that are overwhelmingly the source of trip origins. Areas in orange are predominantly destinations:
Here is another way of viewing the impact of demand that shifts over the course of the day on supply, by individual station and hour (click here for the larger version):
In that matrix, the more gray-colored stations are those that have roughly equal arrivals and departures – and thus probably rebalance themselves naturally. The blue stations (heavy on departures) are invariably going to be your empty stations, where it's hard to find a bike; the orange ones (heavy on arrivals) are probably the stations where it's hard to find a place to park your bike. Across the system, two periods of the greatest imbalance are clear: morning and evening rush hour.
These graphics don't point to any particular answers, but they highlight durable underlying patterns in commuter movement – and the potential of data to help cities design transportation systems attuned to them.
Washington's Capital Bikeshare releases similar data; here's an older and slightly different take on visualizing it.