Mile Marker Accuracy

[twowheeler]

I am wondering how accurate the intervals are, and if it varies between states or within states.

I've made a doohickey that looks at the rotation ratio between front and back wheels of a bicycle. I can readily detect a 20 psi deficit in one tire. Pressing a button at two successive mile markers allows an independent pressure reading for each tire, but the mile markers need to be consistent within 5 feet for good results.

I've heard that a marker will shifted if the "correct" spot is inconvenient, as when a big rock sits there. I think I've noticed an odd interval on a ride. I'd like to know how common those offsets are in various regions.

The code will soon be posted for the world to see. If mile markers vary too much for that function to be useful, I'd rather pretend I was smart enough to do the error analysis and research before writing and testing the function.

[MNHighwayMan]

The MUTCD (government publication that covers specifications for signs, traffic signals, etc) specifies that mile markers can be posted up to ±50 feet of the actual point where it belongs. However, realignments which may cause shortening or lengthening of the route may shift mile markers even more from their proper spot, and it varies from state to state whether the markers are adjusted or short/long miles get marked as such.

[oscar]

Alaska and Hawaii both are conservative about recalibrating mileposts, especially ones used for street/mailing addresses or to give directions. At least in Alaska, the private Milepost travel guide alerts travelers to instances where the distance between consecutive mileposts is significantly more or less than one mile. That is a fairly common entry in its route logs.

California's postmiles will sometimes have codes indicating mileage adjustments where a highway has been straightened or realigned. But the codes are not widely publicized, since the postmiles really aren't intended for use by the general public. They aren't always a mile apart, either.

Like milemarkers on non-freeway routes in many other states, the postmiles usually reset at county lines, though it can get interesting when a route flits back and forth across a county line.

[wxfree]

TxDOT's Statewide Planning Map shows the reference markers and gives the location of each one, in terms of the distance from the origin of each highway.  It's surveyed to a thousandth of a mile.  I assume the map is pretty accurate, because it's just a matter of entering the actual location, not putting the sign in the right place.  I've noticed over time that they aren't all that accurate.  They're often hundreds of feet off from one to the next (they're placed every two miles).  Interstate markers, which are placed every mile, are generally more accurate, since placement isn't challenged by things like intersecting roads, but it's still common for them to be off by one or two hundredths of a mile, from one to the next.

[twowheeler]

Tougher question: What would a histogram of mile marker separation look like?  If a majority are within a millimile, I could take a few measurements. My function automatically discards impossible values. If the cumulative interval over several miles is more accurate than individual miles, I could adapt to that easily enough.

[jeffandnicole]

Overall, they're going to be pretty accurate.  However, any two random ones may be a few feet off.  You'll normally think of it as the starting one being accurate and the one you're looking for may be a few feet off, but it could be the opposite - your starting market could've been the one not in its exact location.

[kalvado]

There should also be some uncertaintly due to curves, as outer lane in a curve is longer than inner one.  I had an impression that centerline is actually measured, and that can easily result in 10-20' deviation. And I really wonder what is measured when two directions of interstate are separated by a very wide median..
And I-88 in NY has (had?) a specific location with "measured mile begins/ends" signs. I thought that one is for some sort of calibration..

[MNHighwayMan]

There should also be some uncertaintly due to curves, as outer lane in a curve is longer than inner one.  I had an impression that centerline is actually measured, and that can easily result in 10-20' deviation. And I really wonder what is measured when two directions of interstate are separated by a very wide median..
And I-88 in NY has (had?) a specific location with "measured mile begins/ends" signs. I thought that one is for some sort of calibration..

The MUTCD actually specifies how it's to be done for divided highways:

Section 2H.05 Reference Location Signs

...

Standard:

12 For divided highways, the distance measurement shall be made on the northbound and eastbound
roadways. The reference location signs for southbound or westbound roadways shall be set at locations
directly opposite the reference location signs for the northbound or eastbound roadways.

[kalvado]

There should also be some uncertaintly due to curves, as outer lane in a curve is longer than inner one.  I had an impression that centerline is actually measured, and that can easily result in 10-20' deviation. And I really wonder what is measured when two directions of interstate are separated by a very wide median..
And I-88 in NY has (had?) a specific location with "measured mile begins/ends" signs. I thought that one is for some sort of calibration..

The MUTCD actually specifies how it's to be done for divided highways:

Section 2H.05 Reference Location Signs

...

Standard:

12 For divided highways, the distance measurement shall be made on the northbound and eastbound
roadways. The reference location signs for southbound or westbound roadways shall be set at locations
directly opposite the reference location signs for the northbound or eastbound roadways.

Which makes sense - although in spots like this one things may end up being funny anyway:
https://www.google.com/maps/@42.6028645,-74.7207642,14z

And still, in a tighter turn, length of the road as measured along right shoulder is not the same as measured along left shoulder. Same map, a little bit to the west has a turn of almost 90 degrees in 1 mile. That is pi()/2*(2x lane width)~45 feet difference. still within 50 feet MUTCD allowance, and should cancel out in longer trips due to right and left curves on longer stretches. 

[cjk374]

The MUTCD actually specifies how it's to be done for divided highways:

Section 2H.05 Reference Location Signs

...

Standard:

12 For divided highways, the distance measurement shall be made on the northbound and eastbound
roadways. The reference location signs for southbound or westbound roadways shall be set at locations
directly opposite the reference location signs for the northbound or eastbound roadways.

LaDOTD really screwed this up a few years ago. They hired a contractor to pour the concrete bases for the new enhanced milepost signs on I-20. Not only did the contractor put the new mileposts in the wrong places (compared to where the old milepost was), the new ones on the opposite side did not line up at all. This occurred between Bossier City & the Bienville/Lincoln Parish line, and from Monroe to I think Tallulah. But DOTD signed off on their work so they are at fault as well.

[DrSmith]

When I was young, there was about a half mile error for Route 322 Eastbound mile 23 marker (this is in NJ), as it was placed near Jobs Lane, while the westbound one was placed at Tuckahoe Road. At some point, the eastbound one was moved in line with the westbound one.

More recently, it is interesting to notice the difference in the mile markers on the Mass Pike. As a part of the sign updates are also the inclusion of enhanced mile markers. The new ones have been placed on the westbound side so far. For the eastbound side they are only there where there is a narrow median and both directions are placed on the same post (the mile markers are being placed in the median along the Pike). Where there is a wide median, the older mile markers are still there in the eastbound direction  (they haven't finished the project yet). In Western Mass, you can see a quarter mile difference in where the old markers were placed compared to where the new westbound ones are being placed in the locations with the wider median (at least in the range of Exits 2-8 or so where I have noticed).

[Beltway]

In Virginia the milepost locations were set by VDOT survey parties.
That would mean accurate to within a fraction of an inch.

[twowheeler]

On a day when both work and snow are absent, I'll test on my nearest road, but most of your reports are encouraging. I've almost never ridden on a divided highway. I hadn't considered the effect of curves. If I'm riding 15 feet from the centerline, and I want the error < .001, the radius of a constant curve must be >= 15000 feet. On winding roads, the amount of left curving could match the amount of right curving, but instead I'll just say the radius for this test should be infinite or nearly so. Since this is a secondary feature, that's not much of a limitation. I think I should tell it to figure out whether the test was over 1, 2, 3 , or 4 miles.

I intended this function as a boredom reducer on all-day trips. It's probably more useful on a regular commute. As long as there's a straight mile with no stop, I could find two objects a mile apart and always use them for the test.

I set up the regular ratio test to be completely automatic, giving very few false positives. Testing over a mile is a different thing, and I probably won't cry much if it doesn't always work.

Thanks to all.

[roadman65]

How does roads like I-84 in OR or I-24 in TN where both carriageways shift miles apart?  One side has got to be longer than the other, so how does both Oregon and Tennessee compensate?

[MNHighwayMan]

If their respective DOTs comply with the MUTCD, then the miles on the westbound or southbound roadways are distorted to be more or less than a mile. In fact, it's going to be true in every state, to varying degrees, as in no state is an Interstate perfectly straight.

It is worth pointing out that the MUTCD does not call them "mile markers," but rather reference location markers. The fact that the distance between them on one carriageway is not exactly one mile misses the point.

[wanderer2575]

There is also the issue of carriageways having different lengths through an interchange.  For example, the I-96/I-275/I-696/M-5 interchange in Novi, MI -- the westbound I-96 carriageway has to be a good half-mile longer than eastbound.

https://goo.gl/maps/qzSsBLbwCfn

[Super Mateo]

What happens in a situation like this?  Eastbound I-80 takes a loop ramp to continue its route.  Is the mileage measured on the ramp?

[MNHighwayMan]

I would think it'd be measured from where the centerlines of the two carriageways would intersect if it were an at-grade intersection.

[roadman65]

The NJ Turnpike and Garden State Parkway both erect them at every tenth of a mile.  I assume these two roadways are pretty more accurate than the regular mile interval signing roads.

Also to point out that GA has them on one pole where both sides of the road's mile sign are erected.  You will see on one side of a two lane roadway to have the marker on the left instead of the typical right side.  Plus GDOT uses the southern or western terminus of the road in each individual county for the zero point is easier to keep track of as well.

[20160805]

Well, I can't speak for a general rule across the board, but the ones along US 10 through Wisconsin are hopelessly far off.  The entire length of the highway across the state is 291 miles IIRC, ending in Manitowoc; exit 291 is about an hour west of that, however, located in metro Appleton.

[ftballfan]

I-75 in Ohio has some exit numbers that are one off their mile markers.

[cbeach40]

Ontario's preferred tolerance is +/- 10 m from the posted location. On the kinds of routes where we post them (freeways, expressways, remote highways) the chances of not meeting that are very low.

[kalvado]

Ontario's preferred tolerance is +/- 10 m from the posted location. On the kinds of routes where we post them (freeways, expressways, remote highways) the chances of not meeting that are very low.

And how that is getting measured? Thinkin about it, task may not be trivial, especially in curves..

[cbeach40]

Ontario's preferred tolerance is +/- 10 m from the posted location. On the kinds of routes where we post them (freeways, expressways, remote highways) the chances of not meeting that are very low.

And how that is getting measured? Thinkin about it, task may not be trivial, especially in curves..

Everything is surveyed, and from that you get your centreline chainage - which will be reflected in the construction drawings and title records. During construction it's certainly easier when everything is staked out.

[Beltway]

Ontario's preferred tolerance is +/- 10 m from the posted location. On the kinds of routes where we post them (freeways, expressways, remote highways) the chances of not meeting that are very low.

And how that is getting measured? Thinkin about it, task may not be trivial, especially in curves..

Everything is surveyed, and from that you get your centreline chainage - which will be reflected in the construction drawings and title records. During construction it's certainly easier when everything is staked out.

On a section with a graded median they could use the centerline of the original design plans in the center of the median.

With a section with independantly graded roadways, the design plans would have a separate centerline for each roadway.  That would be more complex, but for the sake of mileposting the survey party could establish a temporary centerline halfway between the two roadways and then base the mileposting from that.