Ask A Cartographer

Collapsing Edge of Pavement polylines into street centerlines.

October 24 2007 | 0 comments
Categories: ArcGIS Methods, Data Modeling

We have edge of pavement polylines generated with photogrammetric mapping. We need to collapse
these into a centerline polyline. It seems that ArcInfo has a tool to do this. What are the
parameters for this and cost of the additional software to enable us to use this tool.

Mapping Center Answer:

The tool that does this operation is called Collapse Dual Lines to Centerline, and requires an ArcInfo license level for ArcGIS; if you don’t have an ArcInfo license, you’ll need to get in touch with the ESRI account manager or sales rep who works with Avioimage in order to find out the answer regarding cost.

I’ve got some data that is of the character you described; I tried the Collapse Dual lines to Center line tool and found that it works, sort of, mainly for easy cases. Things that didn’t work:

1. Cul-De-Sacs, both the traditional variety with the turnaround at the end, and the variety where the sac is immediately adjacent to a larger street. These end up with a squiggle at the end, sometimes running around the back edge of the sac. It didn’t matter what I used for my maximum width pattern, these will need to be trimmed by hand.
2. Limited access highways and ramps; Here it wouldn’t have mattered whether you could specify that you want just one line or dual lines (one way lines), the tool just got confused and did poorly. It’s likely easier to use “free” highway data for this.
3. Medians; it depends on what you want here; If you want a fork in the road, that’s what you’ll get; once you identify them you could run the tool again on just those (that worked for me); selecting them for the subsequent run is the tedious bit.

Overall I’d say about 80% of the street centerlines I wanted were produced automatically on the first pass. But these were the easy ones. Frontage roads along highways will be problematic.

I ran a couple of scenarios, first using 80 feet as the max (to handled major streets and interstates), and a min of 8 feet, just knowing that some trails were captured that were less than 10 feet wide. I don’t think the minimum matters too much. I ran another scenario with 150 feet as the max to see if that cleaned up either the cul-de-sacs problem or the highways/ramps problem. No luck, just created different spaghetti. The problem is that medians and ramps that come close to other streets cannot be distinguished.

I, when asked about best practices, say it’s easier to capture center lines at the same time your capturing the edge of pavement. That method has the added benefit of producing a much higher quality dataset with respect to cartographic requirements because important characteristics such as asymmetrical lane counts and turn lanes can be accounted for in a meaningful way. Cartographic quality is bit of a loaded term there; whatever quality you’re aspiring to attain needs to be clearly spelled out. Items like how to handle medians, one-way streets, lane counts, parking, Z levels, etc. need to be clearly communicated to the folks doing the data capture. My guess is the Collapse Dual Lines to Centerline will save about 20% of the workload in a hybridized workflow, which would involve capturing, for easy areas the edge of pavement and then running the tool to get the centerlines, but then for the hard areas like interstates and areas of complex interchanges, the person doing the photogrammetric capture will have a far better perspective and the tools to just capture the centerlines properly as part of their work as usual. In fact, I would suspect you’d save a fair bit of time capturing the complex centerlines with the edge of pavement data just for not having to sort out the spaghetti that results from the Collapse Dual lines tool.

If you would like to post a comment, please login.

Contact Us | Legal | Privacy |