Sign Up for Email News

  • CommuterPageBlog and CommuterPage.com
    Sign up to receive new blog posts by email. You can also choose to receive current stories from local news sources about biking, walking, Metro, local bus systems, and more.
Powered by TypePad

« New Smartphone App Makes it Easy to Find ART Stops, Bikeshare Stations | Main | You're Invited, "Policy to Practice: Planning for Bicycles" Seminar »

October 29, 2010

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d83451f42669e201348890ba7d970c

Listed below are links to weblogs that reference Capital Bikeshare Station Demand Map:

Comments

Bob

There should be multiple CaBi stations between Ballston and EFC, with a heavy concentration in Westover. It doesn't just solve the "last mile" issue for transit; it also provides backup to address Metro's reliability issues.

I see the theory that there is good bus transit on Washington Blvd between Ballston and EFC. However, Metrobus has more than enough problems with reliability. Having the CaBi stations will give us options besides walking to the Metro, or waiting for notoriously unreliable buses.

Gee

Thank you for responding Paul.

I did not spend a lot of time in Arlington in 2000. I'm unsure how the Census Bureau updates block level data with its demographic estimates. I suspect that they simply rake it to the county total which would not help internal estimates. Nevertheless, anecdotes from my neighbors suggests that it changed a lot.

Anyway, the model certainly corroborates our instincts about high demand/destinations/etc. How useful it is in distinguishing our "anecdotally-similar" areas would require more work, IMO. I'm guessing that some sort of sensitivity analysis combined with updated data would make it more convincing. But then again, I'm a curmudgeon who is probably going to ride his own bike. ;-)

bArlington

First off - GREAT! This map shows N Arlington such as the metro corridor getting stations. This is an area of high pedestrian traffic and I could see bike share really taking off in that area.

Then I agree with a lot of comments - which is destinations. Where do people bike to or from. Of course Ballston makes a lot of sense. So does Westover and Shirlington. Central Library. Places with a lot of traffic.

And consistent with some comments - where can bikes be used?? Places near or on bike paths have added value. I think that is why Shirlington is a the right place for this. BC it sits on the W&OD and you can imagine people using the bikes from here on those lanes.

Final thought: where are the universities?? There is GMU Law at Virginia Sq. There is Marymount at the golf course. There are probably a few other places. It seems like these students could be very receptive to this service.

For me personally, Rossyln - linking downtown to N Arlington and the N Arlington transportation system - is the most important station.

Paul DeMaio

Steve O., the Westover neighborhood is the center of the neighborhood and would certainly deserve a CaBi station. The map shows that lower density neighborhoods will require residents to walk further to a CaBi station, than more dense neighborhoods.

The East Falls Church Metro will certainly have a CaBi station in the future too. We haven't examined the VA Hospital Center yet, but it could get a station as well.

Paul DeMaio

Allen,

We did use population and employment density in developing the model. I would have liked to use data on retail activity, however, it was unavailable.

Paul DeMaio

Josh,

Shirlington will most definitely be getting a CaBi station. The model shows a need for 8 bikes there.

Bikesharing is for the first mile/last mile of transit commutes, so we will be installing stations in neighborhoods to better connect folks to the nearest Metro station and bus center, as in the case of Shirlington.

Paul DeMaio

Gee,

Thanks for your comments. Regarding the European model question, yes, it's a metric I'm using to calculate the number of bikes, rather than stations, within each block group. So instead of seeing zeros for the number of stations, you're seeing the number of bikes, which in many cases is fewer than needed for a station to be installed.


"Are these tabulation blocks and its population estimates from the 2000 or 2010 decennial census?"
The data is from the 2000 census with a 2009 estimated census dataset.

Regarding your neighborhood comment, I agree that census blocks don't show neighborhood boundaries. To cure this, we included neighborhoods, shown within the blue boundary lines.

"Is that the golf course with a 5?"
This was likely caused due to the number of buses that use 395. I don't see us placing a bikesharing station at the golf course.

"Do we really expect that much bike sharing on the military base?"
The Pentagon and Ft. Myer share the same block group. I could see a 21 bikes at the Pentagon, 13 bikes at the Arlington National Cemetery Metro, and 7 bikes at Ft. Myer, equally 41 bikes.

Paul DeMaio

Alex, thanks for your comments. I'll change the map name.

JRB

One thing to add, based on some prior comments: Bikeshare is successful in areas like downtown DC, because it's often faster than taking a bus or train.

In places like Westover, there are several bus routes; I don't imagine that people will stop using those in order to ride a bike.

For example: The benefit of riding a bike from Farragut Square to, say, the National Press Building, is fairly clear; it would definitely be faster than a bus at most times of day, and probably faster than taking Metro from F. West to Metro Center if you include the escalator/waiting time.

But getting from Westover to Ballston is already quite speedy by bus--maybe 7 minutes? And it's protected from the weather. As to EFC, it's so close to Westover on foot that I don't think many people capable of the 10-minute walk would bother with a bike.

Just my 2 cents.

JRB

I live near Westover, and I think I speak for many nearby residents when I say that this is not an optimal location for bikeshare facilities. First of all, Washington Boulevard (which is one lane in either direction) has a very high volume of automobile traffic, and it moves quite fast; bike riders would be in grave danger, and the presence of slower vehicles (like bikes) would cause many backups, probably all the way from Glebe to the I-66 ramp.

I also don't think there would be many users of bikeshare near this area. Most of the nearby residents walk to the library, restaurants, shops, etc.

Alex B.

Paul, I think this is a great map.

Some of the criticisms, I think, stem from a mis-wording of this as a 'demand' map. I think of it more as a location analysis for Bikeshare Suitability, rather than an attempt to gauge demand.

Demand would indicate what areas want bikesharing, when that really isn't the most important factor - what is the most important factor is where bikesharing will work, and where it will be a good investment. That's why the transit layer is crucial.

I also think including both the numbers and the colors is confusing - the comments here show that, as well. Steve O's point about Westover Village being underscored relative to other neighboring areas is a good one - but I'd point out that all of those areas scored really low, period.

As a model to guide investment, however, I feel this is spot on. The next phase of investment should be focused on the Rosslyn-Ballston corridor. Granted, that might be telling us what we already knew, but it's always good to have the data to back it up.

Josh

Looking at the data, it appears stations will be concentrated around areas with existing strong transit options. While I'm all for public transit, my expectation is that bike share is about non-linear transit and also serves the "last mile" problem that fixed route transit runs into. I'm hoping that when implemented, bike sharing will allow people who previously weren't served by transit to make that 10-20 minute bike ride to transit or to go about their errands (which consequently in Arlington are more than likely in an area with strong transit) without using a car. As a resident of Shirlington, I'm quite surprised that a station isn't anticipated in this location as it is a short hop from Crystal City and the existing bikeshare locations.

Allen Muchnick

I agree with the previous feedback that this model is seriously flawed.

Westover Village with its library, post office, retail, and restaurants--and with a significant concentration of affordable, medium density housing--should be--by far--most active bike-sharing node in all of 22205. For example, many Westover area apartment residents could use CaBi to commute to and from the East Falls Church and/or Ballston Metro stations, despite generally walking to Westover Village destinations.

Similarly, the model seems to poorly predict the bike-sharing potential along Columbia Pike between S George Mason Dr and Rte 27.

I think the primary criteria should be 1) population density, 2) employment density, and 3) retail and civic activity. Except for counting libraries and "community centers" (but not post offices, restaurants, shopping centers, churches, etc.), the model poorly considers the third category. Locating retail sales activity might be a useful criterion, especially if gasoline and car *sales* (but not car repair) can be excluded from this measure.

On the other hand, the public transit criteria seem redundant and perhaps also contraindicated, since bicycling is more useful where transit service is poor.

I also suspect that proximity to designated bikeways is largely irrelevant since virtually all Arlington neighborhoods and their neighborhood streets are eminently bikeable, with or without designated bikeways.

Steve O

Well, I'm not nearly as erudite on this subject as Gee, but when I think about bike sharing, I think of it not as a function of static variables (what is in a particular area), but rather as a connector between "nodes."

Westover Village, which I live near and am familiar with strangely scores lower on the map than the single-family residential neighborhoods nearby. That by itself tells me there's some weaknesses in the model.

Westover is a node of retail/restaurant activity. Two other nearby nodes that are in that a-little-far-to-walk-but-kind-of-short-to-drive distance are East Falls Church Metro (a transportation node) and Virginia Hospital Center (an employment node). I happen to know that many staff at VHC go to Westover for lunch. They drive.

A bikeshare station at Westover without a station at EFC (and to a lesser extent, VHC) is a lot less useful. And vice versa: having a CaBi station at EFC with no place to go would be worthless. Therefore assigning a value is dependent on where the other stations are, too. The model should be dynamic in that way.

Also, having a bike facility like a trail should be scored differently if it's in or near a "node" than if it just happens to be running through a single-family residential neighborhood.

So if it were up to me, I would have identified the "nodes" first and ranked them somehow. Then I would have looked for where the most valuable connections between them existed. It's a different way of thinking about the same issue. Perhaps its better; perhaps not, but it's the way I think about it.

Gee

Hi Paul,

From the page, the model is unclear and consequently it makes it difficult to comment. It seems to me that you're running some simulation -- as opposed to fitting a model with data on "bike demand" as the measured object of interest. My experience is that these simulation models are often highly dependent on structure and parametrization as well as the underlying data. In other words, it is hard to believe a black box model.

The "European model" aspect is just a metric so you can assign numbers to Census tabulation blocks. Correct? Given the verbiage, I expected to see some zeros where a bike station would serve multiple tabulation blocks.

Are these tabulation blocks and its population estimates from the 2000 or 2010 decennial census? What short form data do you use in your model?

Tabulation blocks, if my memory is correct, are largely about boundaries that make it "easy" to collect data -- which includes following up non-response with enumerators -- and not necessarily something that resembles a neighborhood in common parlance. One might question producing output that respects those boundaries since they are artificial for Arlington residents. So some type of smoothing over space might make sense -- perhaps you can use geographic or population centers for the tabulation blocks.

Is that the golf course with a 5?

Do we really expect that much bike sharing on the military base?

I'm not sure how to think about a place like Westover Village. Certainly lots of neighborhood folks go there and parking is limited. But I'd expect those same neighborhood folks to have bikes of their own. This raises the question of how bike demand is being measured. That is, if there are lots of places very "close" to where a person lives, my first thought is that bike demand would be low since people would walk. But if places are "far" from where a person lives, I suspect that many would drive. Moreover, the choice of whether people go to a particular location is not only dependent on what is located there but whether there are similar alternatives elsewhere.

It would be interesting to see how robust the model is to its parameter assumptions. How well does the model fit areas with existing bike share facilities (Paris)?

Otherwise, the model essentially fits our a priori guesses about high demand locations. Wilson Blvd corridor, Pentagon and Crystal City, and perhaps Shirlington.

The comments to this entry are closed.

CommuterPage.com Commuter News