Ride With GPS crazy speed chart- Mtbr.com
Results 1 to 16 of 16
  1. #1
    Professional Crastinator
    Reputation: Fleas's Avatar
    Join Date
    Jan 2006
    Posts
    5,885

    Ride With GPS crazy speed chart

    How is average speed calculated?

    Total distance / moving time? OR using a moving average?

    So I am a tad technology-averse.

    This morning I decided I'd time trial my entire loop and record it. I never do this, mostly because the results always seem suspect.
    In this case, it looks like the tracker recorded every last corner and switchback with full accuracy.

    But the speed looks like it freaked out. 40mph? 65mph?

    With Total Distance / Moving Time these speed anomalies wouldn't matter. But if it compiles a moving average, then the final avg. speed could be skewed.

    Conditions were dry and fast on an easy trail, but I felt slower, and actually stopped several times to clear the trail and pick up trash and air down tires (meaning my real speed was even faster). Auto pause is enabled, but seldom actually pauses when I'm off the bike because the GPS is on my person.

    I do not have any history on this exact route. Just avg. speeds usually about 1 mph slower.

    If my avg. speed is for real, I should feel pretty good about myself.

    Ride With GPS crazy speed chart-speedcrazy.jpg

    -F
    It's never easier - you just go faster.

  2. #2
    mtbr member
    Reputation:
    Join Date
    Dec 2018
    Posts
    371
    "Auto pause is enabled, but seldom actually pauses when I'm off the bike because the GPS is on my person."

    What recording device ? Sounds like a smartphone with RWGPS's tracking app ?

    The Garmin and Wahoo units as well as the Endomondo app for smartphone, had the ability to configure Autopause to be when the device had stopped moving. The Garmins can also be set for something like "speed less than 6mph" or some such. In general though, and if autopause for stopped is enabled, the avg. is going to be distance for the time spent moving.

    If the recording device is your phone and you get off the bike and move around, then obviously the device thinks you are still moving and will not pause. Thus the avg. is going to be incorrect.

    Also note that if you are riding in a wooded area, the GPS may have tracking errors due to weak signal. That in turn can screw the indicated speed.

  3. #3
    since 4/10/2009
    Reputation: Harold's Avatar
    Join Date
    Dec 2003
    Posts
    29,345
    Those unrealistic for the conditions speed spikes tell me a couple of things.

    First off, it tells me that you are not using a wheel sensor. If you're not using a wheel sensor, it's just senseless to try to drill down into speed or distance metrics with much accuracy because of the inaccuracy inherent with using GPS as your primary measuring system for such on a bicycle.

    The second thing those spikes tell me is that there are some inaccurate GPS points that suggest larger position jumps than actually occurred. This is part of the error inherent with using GPS as your means of calculating speed and distance.

    Without a speed sensor, your averages are going to be suspect. The tools most likely just divide total distance by time. The GPS is great at measuring time, so the real issue is with the distance measurement. Because you had jumps that recorded speed spikes, the GPS track itself is showing some inaccuracy, which will show up with the distance. An accurate GPS is ALWAYS shorter than actual. Maybe your inaccuracies with those positional jumps added enough back to make the total distance close to actual. Maybe not, though. An INaccurate GPS has a strong potential to measure distances TOO long because of those positional jumps. You simply won't know unless you get a really accurate wheel rollout distance of the trail.

    The use of autopause is going to complicate things. Averages will mostly be calculating using only moving time, as described. You're really screwing that part up when you keep the computer on your body, and then walk around cleaning up trash and trail debris. Moving time won't be accurate. It'll also be adding distance that you didn't actually ride.

    The issue in wooded areas isn't really with weak signals. It's with multipath errors (scattering of the signals by the trees). Trees aren't substantial enough to completely block GPS signals most of the time. But they ARE substantial enough to make it bounce around. This introduces inaccuracies because the GPS receiver has a hard time correcting or filtering those errors (a better antenna and a better GPS processing chip help, but phones often don't list the specifications of these components, so you can get wildly different results from one phone to the next and have no idea how the components differ from one device to the next and cannot predict the results), not because the signal from the satellite is weak. It can be strong, but also have strong scattered or and reflected signals that add errors.

  4. #4
    Professional Crastinator
    Reputation: Fleas's Avatar
    Join Date
    Jan 2006
    Posts
    5,885
    Quote Originally Posted by Catmandoo View Post
    ...

    If the recording device is your phone and you get off the bike and move around, then obviously the device thinks you are still moving and will not pause. Thus the avg. is going to be incorrect.

    ...
    Incorrect, yes. But it would be slower. I felt like my speed was surprisingly high (by a "whopping" 1 mph - a lot if you are maxed out AND riding in the dark).

    The signal, in the woods, appeared to be very steady as all of the tightest turns are recorded.



    Thanks,
    -F
    It's never easier - you just go faster.

  5. #5
    since 4/10/2009
    Reputation: Harold's Avatar
    Join Date
    Dec 2003
    Posts
    29,345
    Quote Originally Posted by Fleas View Post
    The signal, in the woods, appeared to be very steady as all of the tightest turns are recorded.
    The evidence you've shown illustrates that there were still problems with reception, at least on a spotty basis.

    If I could play around with the data on an interactive map, I could even find the bad points that contributed to the speed spikes and I bet you would see a problem with the position of those recorded points on the map.

    And without a wheel sensor, the kind of analysis you want to do just isn't reliable.

  6. #6
    Professional Crastinator
    Reputation: Fleas's Avatar
    Join Date
    Jan 2006
    Posts
    5,885
    Quote Originally Posted by Fleas View Post
    ... I never do this, mostly because the results always seem suspect.
    In this case, it looks like the tracker recorded every last corner and switchback with full accuracy.

    ...
    -F
    The track doesn't look like it jumped around... there are no straight segments even zoomed in.
    (And I know pretty much every zig and zag by heart)

    Ride With GPS crazy speed chart-speedtrack.jpg

    I realize that we're zoomed out a bit, but even in places where I stopped and walked 20 ft, those movements show up when zoomed in.

    Let's just say I set a PR this morning. ...in the dark, with multiple stops. ...and leave it at that.


    -F
    It's never easier - you just go faster.

  7. #7
    since 4/10/2009
    Reputation: Harold's Avatar
    Join Date
    Dec 2003
    Posts
    29,345
    Quote Originally Posted by Fleas View Post
    The track doesn't look like it jumped around... there are no straight segments even zoomed in.
    (And I know pretty much every zig and zag by heart)



    I realize that we're zoomed out a bit, but even in places where I stopped and walked 20 ft, those movements show up when zoomed in.

    Let's just say I set a PR this morning. ...in the dark, with multiple stops. ...and leave it at that.


    -F
    I see a couple suspect spots even at this zoom level. But since I can't manipulate anything, I can't ID the exact location of the speed spikes.

    Let's just say I know what I'm looking for.

    Here's the speed chart from my last ride. I use a wheel sensor. pay special attention to how much smoother the graph of my speed is over the course of the ride.



    The simple nature of calculating speed and distance from GPS data is going to result in lots of noise both above and below the actual speed. GPS positions have error inherent in them. There is ALWAYS error inherent in each position, whether you think you see it or not. Overlay multiple rides onto each other and it starts to become more obvious the kind of error your GPS has. That error is going to cause the noise in any measurement based upon it. This is why if you want the most accurate elevation, you should have a barometric altimeter taking entirely separate measurements. It's why if you want accurate speed and distance readings, you should be using a separate sensor. It's why if you want accurate power measurements, you should have a separate power meter. Look at the noisy estimated power calculations that Strava put in that chart. They're almost worse than worthless.
    Attached Thumbnails Attached Thumbnails Ride With GPS crazy speed chart-speedchart.png  


  8. #8
    mtbr member
    Reputation:
    Join Date
    Dec 2018
    Posts
    371
    Generally all GPS signals are considered weak when they are are finally received by the device. The OpenStreetMap WiKi well describes the problem of signal strength as well as the results. 2 Quotes:

    "By the time a GNSS signal has covered the distance from a satellite in space to Earth's surface, the signal is typically as weak as -125dBm to -130dBm, even in clear open sky. In built up urban environments or under tree cover the signal can drop to as low as -150dBm (the larger the negative value, the weaker the signal). At this level some GNSS devices will struggle to make an initial signal acquisition/fix (but may be able to continue tracking if a signal was first acquired in the open air"

    As well as; My italics

    "GPS requires a direct line of sight between the receiver and the satellite. When an object lies within the direct path, accuracy suffers due to reflections and weakening of signals. This is particularly problematic in urban environments, within valleys and on mountain slopes. In all three situations, the objects (buildings and the Earth itself) are substantial enough to completely block the GPS signals. When weak signals are received, they may have been reflected off buildings and the surrounding landscape. Reflections generate multi-path signals arriving with a small time delay at the receiver. This results in inaccurately calculated position.

    Even when the object is less substantial (tree cover, car roof, your body), reflection and weakening of signals may still occur. This can sometimes be observed when viewing your recorded GPS track logs on top of aerial imagery. In the image on the left, the true position of the footpath follows the shadowy area in the forest. However, as the GPS receiver enters the forest (walking from east to west), it can be observed that reflections cause the recorded track to incorrectly shift slightly to the south.'

    https://wiki.openstreetmap.org/wiki/...cy_of_GPS_data

  9. #9
    since 4/10/2009
    Reputation: Harold's Avatar
    Join Date
    Dec 2003
    Posts
    29,345
    Quote Originally Posted by Catmandoo View Post
    Generally all GPS signals are considered weak when they are are finally received by the device.
    It should generally be expected that my use of the words "strong" and "weak" above are relative. Yeah, GPS signals in general are pretty weak compared to other terrestrial wireless transmissions we use.

    Quote Originally Posted by Catmandoo View Post
    "By the time a GNSS signal has covered the distance from a satellite in space to Earth's surface, the signal is typically as weak as -125dBm to -130dBm, even in clear open sky. In built up urban environments or under tree cover the signal can drop to as low as -150dBm (the larger the negative value, the weaker the signal). At this level some GNSS devices will struggle to make an initial signal acquisition/fix (but may be able to continue tracking if a signal was first acquired in the open air"

    As well as; My italics

    "GPS requires a direct line of sight between the receiver and the satellite. When an object lies within the direct path, accuracy suffers due to reflections and weakening of signals. This is particularly problematic in urban environments, within valleys and on mountain slopes. In all three situations, the objects (buildings and the Earth itself) are substantial enough to completely block the GPS signals. When weak signals are received, they may have been reflected off buildings and the surrounding landscape. Reflections generate multi-path signals arriving with a small time delay at the receiver. This results in inaccurately calculated position.

    Even when the object is less substantial (tree cover, car roof, your body), reflection and weakening of signals may still occur. This can sometimes be observed when viewing your recorded GPS track logs on top of aerial imagery. In the image on the left, the true position of the footpath follows the shadowy area in the forest. However, as the GPS receiver enters the forest (walking from east to west), it can be observed that reflections cause the recorded track to incorrectly shift slightly to the south.'

    https://wiki.openstreetmap.org/wiki/...cy_of_GPS_data
    None of that negates what I said. It just provides more detail and context. I was trying to keep my explanations fairly simplistic. Quality modern receivers with good GPS antennas and high quality processing have minimal problems with multipath errors and the usual signal weakening from forest cover and "urban jungles" with lots of tall buildings. Yes, loss of accuracy occurs, but it's less than a lot of people make it out to be. Trees without major terrain impediments? Psh, doesn't even rate for me as problematic. On early GPS gear? Yeah, it was a major problem. I remember climbing trees in the early 2000s to get enough signal quality to record a SINGLE POINT. I remember trying to record a track of a trail and getting only a handful of gps points recorded because of a combination of flat terrain, 2nd growth midwestern forest cover, and cloudy skies, but returning on a sunny day in the wintertime with the leaves down resulted in something I could actually use. Modern receivers don't have that problem anymore.

    The real problems start to arise when the terrain completely blocks signals from large swaths of the sky, limiting the number of satellites the receiver can get any sort of fix on.

    You also cannot trust that a satellite image shows the true location of something with complete accuracy. Those images have to be georeferenced. To georeference them, control points must be identified. In areas with lots of forest, it can be really hard to obtain enough control points for high quality georeferencing across the entire image. Positional accuracy of features across that image will vary as the image gets 'warped' to the terrain, depending on the quality and distribution of control points used for georeferencing.

    Unless you're comparing your GPS tracks with information from a professional survey conducted recently enough that the trail hasn't migrated or been rerouted, you have to figure that the basemap you're overlaying your ride onto has some baked-in inaccuracy, too.

    My point is, when GPS is involved, everything is a little fuzzy. Yeah, there are factors involved that can make it more or less fuzzy, but it's still going to be a bit fuzzy.

    For Fleas, most of these details are completely irrelevant, anyway. All he needs to know is that there's enough inaccuracy in GPS that the speed spikes are going to happen and that the inaccuracies will affect ALL of the data. Only way to get rid of the speed spikes is to use a wheel sensor. There are a couple of other aspects of his use (that can be addressed) that affect his measurement of speed and distance, too, and that's it.

  10. #10
    mtbr member
    Reputation:
    Join Date
    Dec 2018
    Posts
    371
    Quote Originally Posted by Harold View Post
    It should generally be expected that my use of the words "strong" and "weak" above are relative. Yeah, GPS signals in general are pretty weak compared to other terrestrial wireless transmissions we use.



    None of that negates what I said. It just provides more detail and context. I was trying to keep my explanations fairly simplistic.
    What you posted was as usual, terrific. My post about “weak signal” was even more simplistic. When you mentioned that weak isn’t really it, I recalled reading that a while back and needed to refresh my memory as to the nature of GPS (and others). The OSM WiKi is a good read.

  11. #11
    Professional Crastinator
    Reputation: Fleas's Avatar
    Join Date
    Jan 2006
    Posts
    5,885
    There are actually some long straights on this trail in the form of boardwalks. There are also some hard right angles due to last minute trail layout changes/adjustments.

    This makes me wonder how my Strava would look (if I participated).

    And speaking of inaccuracies...the graphic doesn't match the numbers. I think I'm sorta done with this.

    -F

    Ride With GPS crazy speed chart-screenshot_20190925-211537_ride-gps.jpg

    -F
    It's never easier - you just go faster.

  12. #12
    Professional Crastinator
    Reputation: Fleas's Avatar
    Join Date
    Jan 2006
    Posts
    5,885
    I finally sat down and read this. Thanks, guys, for chiming in. It is as I had thought. But in spite of all the "fuzz" in the GPS track, I have obtained unexpectedly good accuracy on occasion. Always taken with a grain of salt, though.

    -F
    It's never easier - you just go faster.

  13. #13
    since 4/10/2009
    Reputation: Harold's Avatar
    Join Date
    Dec 2003
    Posts
    29,345
    Quote Originally Posted by Fleas View Post
    I finally sat down and read this. Thanks, guys, for chiming in. It is as I had thought. But in spite of all the "fuzz" in the GPS track, I have obtained unexpectedly good accuracy on occasion. Always taken with a grain of salt, though.

    -F
    The fuzz will often average out when looking at those metrics, because along with unusually high numbers, you'll get unusually low ones, too.

    I've seen from the Strava app (who has written a great deal about the kinds of processing they do to GPS data), in particular, that a ride might have a reasonable overall distance for the ride, even though the positional accuracy of the track is kinda bad. Some of it is related to processing and smoothing that strava does, and some of it is related to the apparent contradiction of the "an accurate GPS will always measure distances short, but an INaccurate one will often measure them more accurately" statement. It's because the "wander" of an inaccurate GPS will add distance back. There's a threshold for that, though, beyond which the measured distance is just as bad, if not worse, but it's long instead of short.

    Pretty much every other decent app is going to have some corrections. One thing not too many people have written much about is the fact that pretty much all raw GPS data needs some processing to fix it. Garmins do it, too. Filtering out outliers, processing multipath errors, and so on.

    About the only way to see the TRUE raw data is to use a survey grade GPS that allows you to see the "point cloud" when you record a single point. First, there is some processing done on-device (mostly averaging, but most survey grade devices can also filter points by signal quality and only record the ones that meet certain standards). But you can also have some post-processing corrections applied when you download your data to the office computer. With extra equipment, you can even apply corrections in real time in the field (these are higher level corrections than Garmin or Strava does, or are even capable of doing).

  14. #14
    Professional Crastinator
    Reputation: Fleas's Avatar
    Join Date
    Jan 2006
    Posts
    5,885
    Quote Originally Posted by Harold View Post
    The fuzz will often average out when looking at those metrics, because along with unusually high numbers, you'll get unusually low ones, too.

    I've seen from the Strava app (who has written a great deal about the kinds of processing they do to GPS data), in particular, that a ride might have a reasonable overall distance for the ride, even though the positional accuracy of the track is kinda bad. Some of it is related to processing and smoothing that strava does, and some of it is related to the apparent contradiction of the "an accurate GPS will always measure distances short, but an INaccurate one will often measure them more accurately" statement. It's because the "wander" of an inaccurate GPS will add distance back. There's a threshold for that, though, beyond which the measured distance is just as bad, if not worse, but it's long instead of short.

    Pretty much every other decent app is going to have some corrections. One thing not too many people have written much about is the fact that pretty much all raw GPS data needs some processing to fix it. Garmins do it, too. Filtering out outliers, processing multipath errors, and so on.

    About the only way to see the TRUE raw data is to use a survey grade GPS that allows you to see the "point cloud" when you record a single point. First, there is some processing done on-device (mostly averaging, but most survey grade devices can also filter points by signal quality and only record the ones that meet certain standards). But you can also have some post-processing corrections applied when you download your data to the office computer. With extra equipment, you can even apply corrections in real time in the field (these are higher level corrections than Garmin or Strava does, or are even capable of doing).
    Don't some trail apps contain their own route data, so that if you are riding a known/established route, it automatically places your position on the route, unless you make a huge excursion?

    -F
    It's never easier - you just go faster.

  15. #15
    since 4/10/2009
    Reputation: Harold's Avatar
    Join Date
    Dec 2003
    Posts
    29,345
    Quote Originally Posted by Fleas View Post
    Don't some trail apps contain their own route data, so that if you are riding a known/established route, it automatically places your position on the route, unless you make a huge excursion?

    -F
    Not exactly. You might be thinking about Strava's "Slide" tool in which you can take a recorded track and it will process your data using Strava's heat map to "snap" it to the "highest heat" portion of the track. But Strava doesn't just do that by default. You have to specifically process your ride with the tool (which is buried under the Strava Labs sub-domain).

    Car Garmins (and auto nav apps) have a setting where you can tell it to automatically snap your position to the nearest road. It works so long as you're following a road that's on the map. I've had issues with this feature going bonkers in situations where I'm driving on a gravel road that doesn't appear on the map, or when I'm driving totally off-road on someone's ranch. This would be a particular problem for mt bike trails since they aren't nearly as well mapped as streets. And even worse, trails are not mapped as accurately as streets. Try following a course on a computer (I have with my Garmin) that's based on aforementioned relatively poorly mapped mtb trails sometime. Your position as determined by your computer will deviate from the planned one, even though you're still on the trail (the actuality of the case is that the problem is a result of the inaccuracy of your planned route in addition to the inaccuracy of your calculated position in real time), and you'll get lots of warnings that you've deviated from your planned route. It doesn't really help when mt bike trails are less permanent than roads. Reroutes (both formal and informal) around problem areas, slower migration of the main line over time, construction and closure of whole trails, etc, all add to the difficulty of making this sort of functionality work for trails the way it does for roads.

    Even then, I've had the occasions where my auto nav absolutely flips out because the GPS position is wandering too much and the app keeps moving me from one road to the next and changing the directions every 10 seconds.

    It's really not a simple thing, and that level of processing in real time is a bit much to ask our devices even now. They just aren't THAT smart.

  16. #16
    Professional Crastinator
    Reputation: Fleas's Avatar
    Join Date
    Jan 2006
    Posts
    5,885
    Quote Originally Posted by Harold View Post
    Not exactly. You might be thinking about Strava's "Slide" tool in which you can take a recorded track and it will process your data using Strava's heat map to "snap" it to the "highest heat" portion of the track. But Strava doesn't just do that by default. You have to specifically process your ride with the tool (which is buried under the Strava Labs sub-domain).

    Car Garmins (and auto nav apps) have a setting where you can tell it to automatically snap your position to the nearest road. It works so long as you're following a road that's on the map. I've had issues with this feature going bonkers in situations where I'm driving on a gravel road that doesn't appear on the map, or when I'm driving totally off-road on someone's ranch. This would be a particular problem for mt bike trails since they aren't nearly as well mapped as streets. And even worse, trails are not mapped as accurately as streets. Try following a course on a computer (I have with my Garmin) that's based on aforementioned relatively poorly mapped mtb trails sometime. Your position as determined by your computer will deviate from the planned one, even though you're still on the trail (the actuality of the case is that the problem is a result of the inaccuracy of your planned route in addition to the inaccuracy of your calculated position in real time), and you'll get lots of warnings that you've deviated from your planned route. It doesn't really help when mt bike trails are less permanent than roads. Reroutes (both formal and informal) around problem areas, slower migration of the main line over time, construction and closure of whole trails, etc, all add to the difficulty of making this sort of functionality work for trails the way it does for roads.

    Even then, I've had the occasions where my auto nav absolutely flips out because the GPS position is wandering too much and the app keeps moving me from one road to the next and changing the directions every 10 seconds.

    It's really not a simple thing, and that level of processing in real time is a bit much to ask our devices even now. They just aren't THAT smart.
    Thanks!

    -F
    It's never easier - you just go faster.

Similar Threads

  1. gps, gps w/ speed sensor, or gps w/ bike computer.....
    By sk8z27 in forum GPS, HRM and Bike Computer
    Replies: 1
    Last Post: 06-13-2015, 05:44 AM
  2. Two speed speed chart at 100RPM
    By richwolf in forum Singlespeed
    Replies: 1
    Last Post: 08-22-2008, 07:46 PM
  3. How do you use your GPS to track to ride? Tips on GPS wanted!
    By Ilikejeeps2 in forum GPS, HRM and Bike Computer
    Replies: 5
    Last Post: 04-02-2008, 02:26 PM

Members who have read this thread: 0

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  

THE SITE

ABOUT MTBR

VISIT US AT

© Copyright 2020 VerticalScope Inc. All rights reserved.