Welcome to MountainViews
If you want to use the website often please enrol (quick and free) at top right.

Site Feedback Site announcements, requests for features, bug reports about the software of the service.
Sort by >

More controls

<< Prev page 1 .. 19 20 21 22 23 24 .. 39 Next page >>
Post details Post   (Contract pics)
simon3
2012-04-19 13:44:54
Re Automatic Calculation of Ascent
Context of comment:At least two people have queried the ascent / descent figures created by the track system. I have also noted a number of anomalies in results with test gps tracks from various sources including ones I made myself.
Message:Anomalies include a) usually an apparently high estimate for height gain b) a different figure for ascent and descent. Captain V's apparently lower estimate is a further exception.

The current algorithm used to determine the height gain or loss is extremely simple: it looks at the height recorded for each trackpoint (of which there can be typically between 500 and 5000 for a track). It then adds a positive increase in height between heights reported to the total ascent and adds a decrease to the total descent.

There are several ways of calculating height gain and each have pros and cons. The "spreadsheet" way that I used years ago was to use a spreadsheet accumulating height gain/loss between waypoints. This has the advantage of being self consistent, but the disadvantage of not recording minor ascents/descents between waypoints.

Then there is the way it is done by the GPS units. However two GPS units going over the same route do not always seem to produce the same figures and are often different by 10% as well as being often 10 to 15% higher than the "spreadsheet" method.

The most consistent method I have seen is that employed by Trailmaster which uses an accurate DEM (digital elevation model) of the land, however this DEM is not at present public and the public versions that are available would not necessarily improve the situation enough to be worthwhile using.

The reason for choosing the current method is that it is transparent. No smoothing or calculation other than addition or subtraction is done, however I would consider a different approach if necessary.

(Note: for speed estimation as used for estimating the users speed and colouring the track an elaborate smoothing algorithm IS used which seems to work well.)
Additional points:So I would be happy to make changes however based a. on evidence and b. on a clear argument as to why another algorithm would be better.

Call for voluntary help. I could use help from someone with statistical ability and some appreciation of what can and cannot be done using numerical methods to analyse sample tracks and devise a proposed, better method of calculating height calculation.

In the meanwhile, here are some known things you can do to improve accuracy.
a. turn on gps in advance of walking, for example in your house but in view of the sky, to allow the unit to figure out its position ("finding the ephemeris"). Turn it off after not more than 12 minutes. When you then turn it on at or immediately before the walk starts let it have a good view of the sky (reduces high initial errors in readings)
b. when carrying the gps position it so it always has a good view of the sky. This reduces trackpoint height measurement errors, which can accumulate and give spuriously high figures.
c. in general leave the gps on and in good view of the sky for the whole walk. (reduces start errors)

Remember also that no matter what solution is arrived at, there is no perfect answer as the problem is similar to that notorious one of measuring the length of a coastline, which depends on size of the ruler. However that said, there could be a better algorithm with an explicable rationale.
Context of comment:   (Show all posts)
Context of comment:   (Show all posts)
Context of comment:   (Show all posts)
Context of comment:   (Show all posts)
Context of comment:   (Show all posts)
Context of comment:   (Show all posts)
Context of comment:   (Show all posts)
Context of comment:   (Show all posts)
Context of comment:   (Show all posts)


RECENT CONTRIBUTIONS 1 2 3 .. 19 Next page >>
Forum: General
Muckish in monochrome
mcrtchly 7 hours ago.
First snow of 2017 on Muckish, Co. Donegal.

  
Forum: General
A hike in the Alps Maritimes
mallowman 4 days ago.
I went to the Southern Alps for a trek at the end of October for hopefully some sun and a dose of vitamin D to get me through the winter. Here is a link to my blog of the trip. I hope you enjoy, S...

  
Track
Lake District: Extended Kentmere Round
Onzy 4 days ago.
Classic Kentmere Round extended to take in High Street. 9 Wainwrights in total. walk, Len: 22.2km, Climb: 1108m, Area: Yoke, Lake District - Eastern Fells (Bri...

Track
Trekking in the Caucasus Mountains: The Chaukhi Massif, Georgia’s Dolomites
mcrtchly 3 days ago.
Mention Georgia and most people automatically think of the State of Georgia in the USA, but there?s another arguably m walk, Len: 34.0km, Climb: 1766m, Area: Georgia ()

  
Track
Lake District: Fairfield Horseshoe
Onzy 4 days ago.
Lakes Classic round taking in 8 Wainwrights... walk, Len: 19.2km, Climb: 997m, Area: Nab Scar, Lake District - Eastern Fells (...

  
Track
Croagh Patrick
Paul_Rooney 5 days ago.
walk, Len: 7.7km, Climb: 709m, Area: Croagh Patrick, Croagh Patrick (Ireland) C...

Summit Comment
Binn Mhairg: A well worthwhile but neglected scramble
IainT a week ago.
The quartzite slabs on the NE flank of Binn Mhairg make a good scramble, one of the best in the Maum Turks. They are reachable by traversing round from the Mam Ean track, pathless but largely gras...

  
Forum: General
"Deaded"
BleckCra a week ago.
Never would have thought it but it turns out mountains can can be named after insects. Despite all my decrying this site for its translation of Meelmore as Hill of the Ants, I find right on the do...

  
Summit Comment
Aughinish: Clare Island
sandman a week ago.
This island is located in County Clare but can only be accessed via County Galway. Originally connected to Clare but the connection was lost in 1755 as a result of an earthquake The British built ...

Summit Comment
Scanlan's Island: Clare Island
sandman a week ago.
Access to the Island is via the gravel road/track located at M2575111537 .During low tide you have no problem driving on to the island and following the track to the left over concrete to within m...

  
Track
Near Truskmore, Dartry Mountains (Ireland)
Paul_Rooney 5 days ago.
walk, Len: 8.5km, Climb: 424m, Area: Truskmore, Dartry Mountains (Ireland) Trus...

  
Summit Comment
Munnia Top: Coastal Hill
sandman a week ago.
Best to access this hill via the farm gate located on the the coastal side at M2929112163 in order to avoid the houses on the opposite side. Just a quick up and down over grazing land to begin and...


RECENT CONTRIBUTIONS 1 2 3 .. 19 Next page >>