Mar 8, 2013 at 9:46 am #1300189
Loki CuthbertBPL Member
@lokbotLocale: Portland, OR
I've been playing around with the DeLorme North America 9.0 software and really enjoy it's auto routing function that works on known trails so I don't have to click on every little bend to get accurate distances on the trail. I am noticing mileage errors of aproximately 13%.
I have mapped out the Eagle Creek/Tanner Butte loop which I know is 24.2 miles, but Delorme North America says it's 21.1 miles. The Rogue River trail is 40 miles and DeLorme says it's 35 miles. The Wonderland Trail is 93 miles, but DeLorme is saying it's 78.6 miles.
Is this kinda mileage problem expected with routing on mapping software?
Are there other programs that have similar routing options where it will stay on the route the trails for you? Or do you think this feature is part of the problemMar 8, 2013 at 11:23 am #1963167
Ryan BresslerBPL Member
That does seem like a lot of variation…can you view their trail data to see if it is oversimplified and missing switchbacks and things? Some data out there for things like trail routes is also way out of date or just plain wrong ;)
They may also sample a set number of points along the route to do their distance calculation which works well for short distances but can mis switchbacks etc.
In my experience working on Hillmap distance calculations are hard to get accurate. Even not considering where they source the trail data calculating distances involves a lot of approximation of spherical geometry and rarely considers changes in elevation etc.
For example I've found that, if I supply an entire user built path vs supplying it piece by piece to the same google maps api functions for calculating distances I get different results. (I do this to generate elevation profiles for elevation at points along a path). I think this has to do with the approximations getting done slightly differently depending on how long the path is and how many points it has I've been trying out a couple of different ways to deal with it but there isn't a great solution and you end up having to write code to handle a fair amount of "fuzzyness" in the data.Mar 8, 2013 at 11:52 am #1963174
Greg MihalikBPL Member
"Is this kinda mileage problem expected with routing on mapping software? "
It depends a lot on the route you are plotting, but yes, 10%, although high, is not surprising.
A roller-wheel odo will get every jog and turn which is just not possible with a data points on 10 meter intervals.
Find a "straight" route – like a dirt road – and see what differences you get.Mar 9, 2013 at 11:24 am #1963502
There are a lot of small errors in maps. one of the biggest is in acccurately showing switchbacks. A mile of switchback may ocupy such a small area on the map that the mape only shows one. In other cases no switchbacks are shown. The amount of errors is partially dependant on the scale of a mape. A mape showing a large area will have more errors that another mape showing a smaller area on the same size sheet of paper will have fewer errors. In the past these small errors would not have been a big deal. However now we have GPS and computers that are being used with maps that were not made with GPS.
Your software is probably following every wiggle on the line in digitized mape to determine its length and then compare that to the scale listed on the map and calculate the distance. But if the map doesn't show every small wiggle in the trail the computer will give you a distance that is too short.
Another thing to watch out for are the trail signs. Don't trust the distances they show. You don't know if that distance is someones guess, a physical measurment, or the distance shown on a mape made 30 years ago. Some show accurate distances. Others don't. The only way to know the distance accurately is toe give a hiker a GPS to to automatically record every minute of the hike and then use that data to make a new mape. On many new maps the distance between trail junctions is now listed but often the fine detail of the trail including switchbacks is not.
You must be logged in to reply to this topic.