Author Topic: GPS average speed  (Read 4105 times)

GPS average speed
« on: 05 June, 2017, 11:10:24 am »
This weekend saw an aborted Asparagus and Strawberries. I used two devices, a 500 which I use just for recording and an eTrex 35 for navigation. As the ride was a 400K event I recharged the 500 twice causing the timer to stop (I really need a reliable OTG cable).

When I got home I uploaded the activities on to Garmin Connect and I have three rides, much as I was expecting. The first stage shows as 66 miles at an average speed of 15.5 mph in 4 hours 17. The second one shows as 70 miles at 13.8 mph in 5 hours 5. The third shows as 61 miles at 11.2 mph in 5 hours 29. I know the reason why the speeds are getting less, but that is not important here. When I stitch the files together I get 198 miles at 9.5 mph. How can I possibly get that? Average moving speed is 13.3 which is something like I would have expected.

If I do a quick calculation 198 miles divided by 20.75 hours does come to around 9.5. So 4:17 + 5:05 + 5:29 = 14:51 is my moving time which the other files have used to calculate the average speed. So how come this changes when I join the files together?

Strangely, the eTrex has also recorded the ride at, you guessed it, 9.5 mph.

Re: GPS average speed
« Reply #1 on: 05 June, 2017, 11:41:01 am »
That behaviour sounds reasonable. When you stitch the files together you're adding in the two chunks of non-moving time to the combined file, so those two stops are included in the total time from end to end - since you weren't moving, average drops. The individual files don't have that data in them so effectively you've created more information, when concatenating the files, of the 'these three rides are are really all one ride' variety.

Re: GPS average speed
« Reply #2 on: 05 June, 2017, 12:08:21 pm »
I hear what you are saying and that certainly makes sense. The problem is though that there is non-moving time in all the files and that is not taken into account on the individual stats. Just seems that when the files were stitched it changed the way it calculated the speed. Or it could be that I had to manually upload it.

Re: GPS average speed
« Reply #3 on: 05 June, 2017, 01:44:16 pm »
More than likely each new file starts somewhere slightly different to where the previous file ended, so missing time between each stitched segment is considered to be moving instead of non-moving because the simple algorithm it uses doesn't know any better.
"Yes please" said Squirrel "biscuits are our favourite things."

Re: GPS average speed
« Reply #4 on: 05 June, 2017, 02:47:22 pm »
I don't buy the gap between the files at all. There is only about an hour and a half in gaps between the start and end of each file.

Time between files 1 & 2 - 13:23:20 - 12:43:12 = 00:40:08
Time between files 2 & 3 - 21:00:31 - 20:04:43 = 00:55:48

No. The problem is how the average speed is calculated. For some strange reason Garmin Connect has used the elapsed time to work out the average speed. This is completely different to the way it is calculated on the 3 individual files which use the moving time, and the sums do bear this out.

Karla

  • car(e) free
    • Lost Byway - around the world by bike
Re: GPS average speed
« Reply #5 on: 05 June, 2017, 02:58:52 pm »
Did you stop your GPS recording for these non-moving periods in each file, and how long was the longest period?  I wouldn't be at all surprised if Garmin Connect reacts differently to a) a period of constantly-recording trackpoints that just so happen to be centred on the same place, and b) a 40 minute blank with no recorded data at all.

Also, mind if I ask why you use the Edge to record?  You know you can get the track out of the Etrex, right?

Re: GPS average speed
« Reply #6 on: 05 June, 2017, 03:41:01 pm »
You may well be right about Connect reacting differently when it sees a blank section, ie the space between files. What may be the issue is that it has seen these blank sections and decided to use elapsed time rather than moving time to calculate the average speed.

I use the 500 as a data recording device and the eTrex as the navigation device. I have never had a problem with keeping it like this, and it usually works well. Maybe when the 500 packs up I will only use the eTrex.

Karla

  • car(e) free
    • Lost Byway - around the world by bike
Re: GPS average speed
« Reply #7 on: 05 June, 2017, 04:06:55 pm »
Perhaps try downloading the gpx file from the etrex, opening both files in a text editor and spotting the differences.

Kim

  • Timelord
    • Fediverse
Re: GPS average speed
« Reply #8 on: 05 June, 2017, 05:13:26 pm »
I use the 500 as a data recording device and the eTrex as the navigation device. I have never had a problem with keeping it like this, and it usually works well. Maybe when the 500 packs up I will only use the eTrex.

I'm sure it works fine (all of these devices are capable of both), but it does seem to me to be about 50% backwards.  Logging is something that eTrexes do well.  They just work.  While Edges require all sort of user intervention re starting and stopping, even if they don't suffer the long-track-related bugs.

How you (or the analysis tool of your choice) interprets the data is really a separate issue, although Garmin devices do their own on the fly interpretation to give the "Moving average" and "Overall average" fields in the trip computer, which may or may not use the same algorithm.

Re: GPS average speed
« Reply #9 on: 05 June, 2017, 06:38:30 pm »
It may seem a little backwards but I bought the 500 about 6 years ago and I simply didn't want to get rid of it. Never had a problem with data recording so I didn't see the need to use anything else. When I bought the eTrex I did so because of its mapping capability. I use both at the moment because I can fit both on my handlebars.

If I could use an eTrex as both a mapping tool and a data recorder then I would like to do so, especially as the battery life in the 35 is very poor. I went through 4 sets of batteries on the Asparagus and Strawberries. I want to look at a 20x or a 30x to see if I can get all that I want and better battery life.

Anyway I have uploaded the file from the eTrex and it too agrees with the concatenated file. Therefore it uses the overall time not the moving time. Is there anybody that can confirm that the 20 or 30 does that or if it uses moving time to get average speed?

Kim

  • Timelord
    • Fediverse
Re: GPS average speed
« Reply #10 on: 05 June, 2017, 06:43:00 pm »
If I could use an eTrex as both a mapping tool and a data recorder then I would like to do so

Sounds like you already have done.  An eTrex is always logging, unless you switch it off or into demo mode.  It's that simple.


Quote
Anyway I have uploaded the file from the eTrex and it too agrees with the concatenated file. Therefore it uses the overall time not the moving time. Is there anybody that can confirm that the 20 or 30 does that or if it uses moving time to get average speed?

As I said above, the eTrex calculates and is able to display both.  There's no 'average speed' of any kind logged in the file, it's just a list of positions and times.  How you interpret that is up to the software you use to analyse it.

Phil W

Re: GPS average speed
« Reply #11 on: 05 June, 2017, 07:08:24 pm »
The average speed on GC is the calculated distance divided by calculated elapsed time.  if you go into an activity and scroll down below the map you should see an entry average moving speed. Distance divided by calculated moving time. *

* The edge units (since the 500) record to the fit file format. That does include summary records for distance etc. separate from the raw time / location data you find in a GPX file. Which bits of the summary data is used in GC is not documented that I'm aware. But I do know it is used from when I've had my old edge 500 crash and lose the location data but still show the summary data in GC

Re: GPS average speed
« Reply #12 on: 05 June, 2017, 09:06:36 pm »
The average speed on GC is the calculated distance divided by calculated elapsed time.

That is the way it seemed to work after the weekend but GC has always used the moving time to calculate the average. That is the bit that has changed. Consider the links below. The top one is from Yesterday and the other from last month. Calculated in different ways. Maybe Garmin changed something.

https://connect.garmin.com/modern/activity/1721326876
https://connect.garmin.com/modern/activity/1778392460

Re: GPS average speed
« Reply #13 on: 05 June, 2017, 10:43:51 pm »
The average speed on GC is the calculated distance divided by calculated elapsed time.

That is the way it seemed to work after the weekend but GC has always used the moving time to calculate the average. That is the bit that has changed. Consider the links below. The top one is from Yesterday and the other from last month. Calculated in different ways. Maybe Garmin changed something.

https://connect.garmin.com/modern/activity/1721326876
https://connect.garmin.com/modern/activity/1778392460

Have I missed something here...the first activity is "cycling" and the second is "other". Is the average possibly calculated differently ? What happens if you change the second to "cycling" ?

Phil W

Re: GPS average speed
« Reply #14 on: 05 June, 2017, 11:57:10 pm »
The calc was the same back in December looking at tracklogs loaded from my Etrex 20. So I suspect the difference is the Fit file format of the edge 500. I suspect that the average speed comes from the summary records in the fit file of the Edge 500. The fit file average moving speed being used. But for straight GPX , GC calcs the average based on elapsed (which it also calls) and doesn't try to eliminate stopped time from the calc.

So load an activity straight from the Edge 500 and I suspect you'll see the moving average you are expecting. With the other suspicion being that your merge eliminate the fit file summary data or produced a GPX to load which has no such equivalent record.

A consequence of the edge 500 firmware no longer being updated but GC still being actively developed and updated is that their chosen calcs (if they have any equivalent ones) can get out of sync.

Re: GPS average speed
« Reply #15 on: 06 June, 2017, 11:36:18 am »
The whole issue is down to the way GC handles file imports. Import a fit file and it uses moving time to calculate average speed. Export that activity as a gpx and then delete the uploaded activity. Import the file back into GC and it will use total time in its calculation. Don't know why there is a difference, I haven't studied the format of a fit file. Also for some reason, when you use a fit file you get a calorie count.

Re: GPS average speed
« Reply #16 on: 06 June, 2017, 11:56:00 am »
GPX files only contain a fraction of the data that is present in a .fit file. There's simply no way to bung all of the data in a .fit file into the rather limited XML schema of a GPX file.

(I have studied the .fit file format in order to write my own analysis tools.)
"Yes please" said Squirrel "biscuits are our favourite things."

Re: GPS average speed
« Reply #17 on: 06 June, 2017, 12:51:32 pm »
GPX files only contain a fraction of the data that is present in a .fit file. There's simply no way to bung all of the data in a .fit file into the rather limited XML schema of a GPX file.

(I have studied the .fit file format in order to write my own analysis tools.)

I believe GPX files can have custom extensions to store any data.  And you can export lots of it from Strava...

https://support.strava.com/hc/en-us/articles/216918437-Exporting-your-Data-and-Bulk-Export?mobile_site=true

Re: GPS average speed
« Reply #18 on: 06 June, 2017, 01:43:49 pm »
GPX files only contain a fraction of the data that is present in a .fit file. There's simply no way to bung all of the data in a .fit file into the rather limited XML schema of a GPX file.

(I have studied the .fit file format in order to write my own analysis tools.)

I believe GPX files can have custom extensions to store any data.  And you can export lots of it from Strava...

Sure, but the problem with custom extensions is that most other pieces of software don't understand them, so they are effectively useless.

From that link you'll see that GPX doesn't support storing power data, laps and a few other bits in GPX files (only in TCX files).
"Yes please" said Squirrel "biscuits are our favourite things."

Kim

  • Timelord
    • Fediverse
Re: GPS average speed
« Reply #19 on: 06 June, 2017, 01:47:09 pm »
Sure, but the problem with custom extensions is that most other pieces of software don't understand them, so they are effectively useless.

You could say the same thing about FIT files.  The advantage of custom XML extensions is that the regular data can at least still be read.

frankly frankie

  • I kid you not
    • Fuchsiaphile
Re: GPS average speed
« Reply #20 on: 06 June, 2017, 01:48:39 pm »
GPX doesn't even support 'speed' - let alone average speed and moving average speed.  All these things can be derived from the raw GPX data of course, but that is down to the software in use, whether it bothers to do it or not.

Phil's explanation upthread is surely the complete answer:
* The edge units (since the 500) record to the fit file format. That does include summary records for distance etc. separate from the raw time / location data you find in a GPX file. Which bits of the summary data is used in GC is not documented that I'm aware. But I do know it is used from when I've had my old edge 500 crash and lose the location data but still show the summary data in GC
when you're dead you're done, so let the good times roll

Re: GPS average speed
« Reply #21 on: 06 June, 2017, 01:51:50 pm »
Sure, but the problem with custom extensions is that most other pieces of software don't understand them, so they are effectively useless.

You could say the same thing about FIT files.  The advantage of custom XML extensions is that the regular data can at least still be read.

One of the main drivers for FIT was that many of the things that were custom (which each manufacturer implemented with slightly different XML schemas) were standardised in the FIT format. If you can read and process power data for a bike from one manufacturer you can do it for all of them.

The FIT format is self describing, so it's perfectly possible for anything to read it and skip over the bits it doesn't understand. Indeed there's plenty of data present in the file that aren't documented although people have begun to work out what they might be.
"Yes please" said Squirrel "biscuits are our favourite things."