How can we measure and map "Sicherheit"?

[philshem]

As a current bike-commuter in Stadt Zürich, this project will be useful for me to understand the fastest routes for crossing the city in the shortest time.

Optimizing "time" is good for current cyclists like me, who often skip the first coffee because the adrenaline is pumping through my veins when I get to my desk.

The barrier for would-be cyclists remains: unsafe infrastructure. For that reason, especially because the Velorouten-Initiative should (hopefully) improve infrastructure, how can we map "safe routes" between A and B? Is that something of interest to this group? (Or perhaps this map/data is what the Stadt Zürich must deliver?!)

An example, similarly for Paris, is Plan Vélo.

[rastrau]

I find both the path mapping and the corridor (straight-line) time mapping ideas interesting. With regards to your idea, @philshem: Do you have a suggestion or preference for an approach? The way I see it, one could look at this, for example:

• a) from an infrastructure perspective using the official open data on bicycle infrastucture (where "infrastructure" includes paint)
• b) from an infrastructure and safety evidence-perspective using the official open data on bicycle infrastucture as well as crowdsourced data from e.g. https://bikeable.ch on "Schwachstellen" (dangerous sites)
• c) from a usage perspective using data from tracking apps
• d) from an expert perspective focussing on assessed (not data-derived) safety
• e) from an expert perspective, taking into account all afore-mentioned perspectives, essentially manually curating safe and sensible routes (potentially very involved)

[philshem]

Thanks @rastrau for formalizing the possiblities

In my non-expert view, I'd think that the last option would be the most promising. The data at bikeable.ch are singular points, but perhaps with "community suggested routes" that take into account both (1) existing infrastructure via geodata, and (2) avoiding bikeable.ch's dangerous spots.

I don't see an automatic way of generating this map/routes. To makie it "community suggested" means we can curate the map (although this is a manual effort).

[rogerfischer]

We can combine the two, straight lines and best route between 2 points.

PS: I am currently outside the office, but can explain how we could do it later.

[philshem]

The city provides the mapping segments in their open data set: https://www.stadt-zuerich.ch/geodaten/download/Fuss__und_Velowegnetz?format=10009

The categories for bicycles are:

0=keine Route in der gedruckten Karte "Map Velo"
1=empfohlene Route
2=empfohlene Route Naturbelag
3=schnelle ergänzende Route
5=Fahrverbot
6=Biketrail

I've parsed each category into its own geojson file: https://github.com/swiss-bike-data/zurich_city_bike_paths_2020/

Building on Ralph's option (e)

e) from an expert perspective, taking into account all afore-mentioned perspectives, essentially manually curating safe and sensible routes (potentially very involved)

What if we created a data-pipeline:

1. download that existing dataset from the city
2. filter on relevant categories from _mapvelo (1,2,6)
3. enhance the dataset by adding nodes for the hub spots, defined in the columns "origin" and "destination" in this csv file: https://github.com/posmocoop/spatial_future/blob/master/bikemappingzrh.csv (e.g. Bucheggplatz)
4. enhance the dataset by adding properties for "community_suggested_route", perhaps with values {1, 2, 3} to indicate how "safe" the community considers them. These routes should aim to connect the hub points defined in step 3. This would be a crowd-sourced project, building on bikeable.ch but for routes instead of individual points. btw, the bikeable JSON endpoint is here: https://backend.bikeable.ch/api/v2/cachedlightentries

This would create an interactive map based on our enhanced geojson, routing people between hubs with a community suggested route:

If we add our assessment on top of the existing data, then it's easy then to later abstract the map to something easy to read, either a subway-style map,

(source)

or the hub-and-spokes that this project is building.

[rogerfischer]

This is the format, I would propose for the future: https://github.com/posmocoop/spatial_future/blob/master/walkmappingzrh_alternative.csv

It is based on Posmo Segments data (Posmo Segments is an app available on the App Store/Google Play Store): E.g.

The "City Mapper" (if you have a better name, ping us) would collect all "Posmo Segments" for a specific edge and than we could see what is the shortest, safest etc.

We have an idea to automate this in the future, but currently our resources are completely bound to bring out Posmo One early next year. However as shown here (https://github.com/posmocoop/spatial_future/blob/master/bikemappingzrh.csv), it is also not too difficult to do this by hand.

Automation idea:

• First we create the map of major locations in a city (and circumference and touch points around major locations, e.g. all 3 touch points of HB Zurich, Europallee, Head of HB, Sihlquai)
• Every user can load in Posmo Segments data and select files for City Mapper. For every direction (e.g. HB Zurich (Sihlquai entry) - Limmatplatz, Limmatplatz - HB Zurich (Sihlquai entry), and any files available, we create median and average time.
• On a map you can then check the whole map and detailed edges. This way you can single out the safest route (consensus based) and the shortest.

[rogerfischer]

@philshem @rastrau @thohug Maybe the best would be to do a video call and discuss options short, mid- and long term. There are other people thinking around this as well which we could invite.

[thohug]

I haven't really seen a lot of good studies on local bike safety and safe infrastructure; most of them are strongly biased (as they rarely take into account the amount of traffic - obviously because it's hard to measure properly...). The best we get are studies comparing cities, which in the end doesn't turn out to be accurate enough.

So I think Roger's suggestion on choosing the best route based on consensus seems like a fairly good approach. This way the impact of improved infrastructure could also be determined (if there is enough data tracked...). Nevertheless it won't tell us where it's safe to ride and where not.

[thohug]

One more thing: Overtaking-Distance of cars could be a good proxy for safety of infrastructure (not only though...). Really nice work here by the Tagesspiegel: https://interaktiv.tagesspiegel.de/radmesser/index.html

If anyone is keen on doing something similar in Zurich hit me up :)

[rogerfischer]

@thohug I would be happy to test distance as well, while recording segments. This would give us an objective idea of how dangerous a street is. Unfortunately Switzerland has no law to keep a safe distance of 1.5m when overtaking a bike.

We could then also check if there is a correlation between overtaking and accidents (located accidents are available as Open Data)

[rogerfischer]

PS: While I was recording bike segments, I found out that it is important to define some boundaries where a place begins. E.g. if we only define Bellevue as the Kiosk, you always have a) to add the additional time to wait till being able to cross the street and sometimes this means walking instead of biking and b) it doesn't reflect how you would bike around Bellevue.

So I set some margins (in the case of Bellevue this is mostly across the street of Bellevue). The cercle below is an approximation.
For HB Zurich, I choose 2 entries (Europaallee entry (bikeable), Sihlquai entry while being on the street) as well as the head of HB Zurich after leaving the tunnel at the stop light.

Bellevue

HB

[rogerfischer]

PS: Hier noch wie die Strecken mit Posmo Segments aussehen: https://twitter.com/posmo_coop/status/1330564450110365697?s=20

[rogerfischer]

PS: I updated the article. Let me know if you have any comments. https://medium.com/@posmo_coop/bikemapping-your-city-e3fe44e276a4