To split or not to split, that is the question..

[Will / NeoPhoenixTE]

I find myself as a crossroads, a fork in the road if you will, on whether or not I should put a fork in the road.

Most of us know the general rule of splitting a road around a physical barrier/median in it. But, I'm wrestling with the concept of splitting a road that is much like the example in the help files...

The road I'm looking at is much like this one, but without the hatch marks in the middle. It's one of those super-wide roads that's split by the pair of double-solid yellow lines in the middle. Sometimes the inner lines go dotted to turn it into that "chicken lane" for left turns.

I've not seen much discussion into splitting non-physically divided roads. I mean, it would make it easier to draw turn segments, and would give a user a better idea of just how thick this road is. Lane number doesn't seem to affect line width, so I can't think of any better way to define these roads.

Guess I just wanna get some conversation going on this and see what other people think.

[djboge]

I'm interested here too. I think if the road goes on for quite awhile and has at least a lane's width of buffer zone between both traffic directions it could be split but I also can see the other side. Simple is best, so if there's not a physical divider I don't see a hugely compelling reason to split the road - it adds a ton of extra segments and such.

This has always been my favorite example of an unnecessary road split: http://goo.gl/pXxWD

[Saikrishna Arcot]

I wouldn't split the road since if you were entering the road from a parking lot or similar, you can take left turns as there is no physical divider.

[Will / NeoPhoenixTE]

I would've split that one a bit further to the east to be perfectly honest.

Then again, keeping it as a unified road would've worked too.

I almost want to tinker with your example. Be careful what you link around me..

[djboge]

Haha you can do whatever. :) I can't edit any roads anymore without them going into moderation so I don't even bother trying to fix stuff like that.

[David Nesting]

The deciding factor for me is usually how the division would affect driving directions.

Are turns across the median permitted pretty much anywhere on the segment?  If so, don't split it, and don't mark it divided.

Are turns across the median permitted only at mostly existing intersections, or intersections that you're prepared to draw?  If so, you could just set the 'divider' attribute and ensure that all of the turns across the median occur at intersections.  Or you could split the road if it's aesthetically better.

If you choose to split the roads, make an effort to identify all of the places where turns across the median are permitted, and make sure connector roads exist to facilitate those turns.  My main goal when deciding to split is to avoid driving directions that instruct drivers to pass their destination (sometimes for a significant distance), and then make a U turn at the next intersection, when in reality they could have simply made a left turn at their destination.  Splitting the road should not cause a regression in driving directions like that unless it's actually improper for the driver to turn across the median there.

David

[Willy Hrachovina]

I have seen it done both ways, but in my opinion, I do not like the idea of splitting the roadway unless there is a physical barrier (median) that exists along a significant portion of the roadway. I believe that is how it is spelled out in the GMM user guides, and to me it makes the most sense. 

In the case of the "center turn lane" configuration I think that it is easier to use just the one segment marked as "two-way", otherwise the turn restrictions could make directions tricky.

If the yellow lines are wide like in the picture from the help guide, if they don't want you making turns across that, then it might make sense to have the divided segments. However, most times I see that either right before/after a median, or when the road expands for a left turn lane. When that happens, unless there is a median, I would not make split segments.

[I MAP]

I feel that if any restriction by law happens which prevents us from going from a Business on one side to another, it should be split

[Will / NeoPhoenixTE]

See, this right here is the impression I'm getting personally. A legal divider should be treated the same way as a physical one, because the two are one in the same if there's an officer nearby watching, heh.

[David Nesting]

On Sun, Aug 28, 2011 at 08:55, General Map Maker on behalf of Will / NeoPhoenixTE <google-...@googlegroups.com> wrote:

See, this right here is the impression I'm getting personally. A legal divider should be treated the same way as a physical one, because the two are one in the same if there's an officer nearby watching, heh.

I think setting the 'divider' attribute might also be sufficient here, right?  Won't that prevent turns across the median unless you're at an intersection?  If it's improper to turn across the median, we shouldn't give driving directions that ask users to do it, so either the 'divider' attribute or a split road is needed.  Which one to use is a judgment call based on length of the span, aesthetics on the map, etc.

 

David

[I MAP]

Divider attribute won't affect GPS in comparision to that of a split

[- deltafox -]

that's a problem with your GPS, you shouldn't alter Map just to get a workaround for your GPS problem !

In the help it says that even with a physical divider, roads don't have to be split (except if both sids do not match). Let use the 'divider' attribute instead !!! And turn restrictions as needed.

Ideally, GMM should improve rendering to show roads with 'physical divider' attribute as split roads, without a need to split the feature itself. And automatically set turn restrictions if there is a divider. Meanwhile, it's counterproductive to split these roads (and even more, roads with only a legal divider, or without any divider !) since it makes every change much more complex (in a case of image shift, which is likely to occur every couple of years). Think that a simple intersection between two roads become 4 intersections if both roads are split ! And there are many other disadvantages such as duplicating names, etc...

[John]

Hear, hear!  THIS is the answer.

"Ideally, GMM should improve rendering to show roads with 'physical divider' attribute as split roads, without a need to split the feature itself. And automatically set turn restrictions if there is a divider."

Great thought, deltafox.  If we could draw ONE line while having it auto-split in rendering, while still showing us when a connecting road goes through and when it does not, this would be a huge, huge improvement.  When a road then really does split off into two distant, non-parallel carriageways, we can split as appropriate - but splitting for a whole road that has a concrete divider makes no practical sense, even though it's unfortunately the best way to visually know which connecting roads allow access to both sides.

Let's leave the actual implementation to the design team, but the concept is a very elegant solution to this mess.

[Cory]

I've thought the same thing--a single pixel line down the middle of the road segment would probably be enough of a visual cue to differentiate a road with a divider from one without.

[John]

The tough part is, you still have to make it very obvious to the average Maps user which roads go through to the other side, and which ones are blocked by the divider.  This is one HUGE current benefit of road-splitting.  The user can see that easily.

[Will / NeoPhoenixTE]

One of the other reasons that I like to split some roads is an alignment issue. Sometimes the road is just so wide that showing it as a single line makes some of the intersecting turn segments over-extend, really. Meanwhile, if you have two roads that line up with the lanes, the over-extension isn't needed and the map is just all the more accurate.

[John]

Good point - all of this points to really legitimate reasons for splitting, and I can see the value in it.  Just wish Google would come up with easier ways to handle this, that wouldn't require us to make two edits on every piece of road that we want to change.

Maybe there'd be a way to allow a road segment to be "associated" with another - kinda like how in video editing, I can select two audio tracks and tell the editor that they're a stereo pair.  Sorry, best analogy I could think of at the moment.  Sometimes, you have more than two split roads, so that could be a complication - roads that have service roads, for example, and ramps connecting them.

Anyway, the implementation would be up to the software engineering team, but hopefully with some input from mappers.  I definitely think -deltafox-'s solution is a great starting point.  Single roads that auto-split into two (or more?) associated segments when you set a divider attribute!  Makes sense visually, and isn't as complicated for mappers!

[David Nesting]

On Sunday, September 4, 2011 3:45:27 PM UTC-7, John wrote:

Anyway, the implementation would be up to the software engineering team, but hopefully with some input from mappers.  I definitely think -deltafox-'s solution is a great starting point.  Single roads that auto-split into two (or more?) associated segments when you set a divider attribute!  Makes sense visually, and isn't as complicated for mappers!

This sounds elegant if the segments are close together, but if not:

1. How do you get the distance between segments right?  What if the distance changes?

2. How do you get intersections at either end of the span looking and behaving properly?  (I.e., how would you connect a traditional one-way segment to one side of the split?)

3. How would you do things like a dedicated left/u-turn lane (which would be drawn in the median) versus a conventional connector road?

I think another approach would be a Photoshop-style drawing mode, where you can first draw an approximation of what you want as a smooth curve, set the distance between one-way segments (or radius of a circular roundabout), preview where the intersections are going to be created (and maybe highlight nearby segments that aren't intersected), let you fine tune, and then 'commit' the change onto the regular MM canvas where you can proceed with fine-tuning and setting attributes.  This would solve the original problem at creation time, but still keep everything as standard segments with traditional intersections and connector roads.

(An extension of this idea would be to allow creating squares and rectangles, allow rotation, etc., to get better polygons, and automatically creating points to best draw a smooth arc.)

David

[Caboosey]

I got bunch roads that should be a single two way road marked with 4 lanes at http://goo.gl/33tji
Like 9th Ave, Davis Hwy, parts of Airport Blvd, Brent Ln, and Bayou Blvd in Pensacola, FL. I have the time to edit the roads, but MapMaker only lets me delete one at a time per route and these roads will automatically trigger a review for any changes since they are marked as major artery. I would spend more time on going back and forward deleting one road segment at a time and moving one road intersection at time to align the road, which would cause issues for regular Google Map users and require more time. It looks like some Google reviewer in June 2011 went on a frenzy and divided up these roads that should have never been divided.