Usually there are data on vehicle flows, but not on pedestrians (or on bikes, but this is another story). This does not mean there is no available methodology, but that they are rather used just in some cases. There are several systems for footfall counting (without being exhaustive), each with advantages and problems, and their technologies evolve really fast:
- Put someone on the street to count the pedestrians. As primitive as it may seem, it does not need a specialized personnel, but just a trustworthy one. Besides, a person recognizes much better than a machine other data (sex, age…).
- Count on video. This has had problems by night or with bad weather. There are image recognition algorithms, but they are not always accurate. Human supervision can help.
- Laser sensors. They avoid the bad weather issues, allow to describe the speed of the person and other elements on the public space, but their range is limited and they do not have an easy time discriminating individuals from groups.
- Infrared systems, mainly indoors, which eliminate some video issues.
Besides, there are method problems:
- When and how often to count? Once a year? Some days of the week? All the time?
- Where to install the counters? In traffic networks it is common to use a high number of counters, allowing the description of citywide grids, but the experience counting footfall is often in more restricted areas.
So it is not easy to have comprehensive data on this issue, and it is even more complex to have them describe an entire city. But there are experiences, even with continuous coverage, for retail malls and business districts, so data is available to face the evolution of the buyers behavoiur.
Footfall counting is relevant as many cities want to encourage non motorized mobility and must prioritize investments; usually it is rather easy to know which streets are the busiest, but understanding their network effects is less evident. It is also relevant for cities wishing to have streets with wider sidewalks in detriment of the traffic lanes.