JonnyW wrote:Hi All

I do own the Tracks.org.nz website and I am keen to see if there is any real interest in me opening it up to cover more (or all) of New Zealand?

Hey Jonny

I vote for expanding tracks.org.nz to cover regions outside of Wellington. The site already has support for multiple regions (states) and there seems to be a bunch of people from other areas who have data to contribute, seems like a no brainer to me

Don

iodi wrote:As for the GPS, I have a Vista HCx. The H stands for high sensitivity antenna, which makes quite a difference. Even so, reception is a bit flakey in places like the Karapoti Gorge, under dense tree cover and surrounded by steep hills. The only time it has ever lost signal was in the Rimutaka tunnel - I suppose a hundred metres of rock over head might have something to do with that.

The HCx has an altimeter so I do collect elevation data. I believe that elevation charts are on the Tracks list of things to do.

Hi Iodi

Do you modify the tracks you have added to tracks.org.nz at all or is what we see on the site effectively unmodified track logs? It looks like the data I have been collecting on my new Holux m241 will need a bit of "massaging" for accuracy.

donovan wrote:Hi Iodi

Do you modify the tracks you have added to tracks.org.nz at all or is what we see on the site effectively unmodified track logs? It looks like the data I have been collecting on my new Holux m241 will need a bit of "massaging" for accuracy.

Raw GPS logs are a bit rough to use directly. In particular:
- GPS logging is inherently imprecise, having a degree of inaccuracy that depends on the quality of the satellite reception conditions; and
- the Google Maps satellite/aerial photo images don't always align well with reality - eg. in the Wainuiomata area the Google maps are misaligned by 40-50m, which is way too much to ignore.

So, to create the kml files I do the following:
1. Save the GPS log as a gpx file.
2. Open the gpx file in Google Earth (gpx is a Garmin file format, though any other format that Google recognises would do).
3. Manually draw the track as a Path using the GPS log as a guide, while looking for clues on the aerial photo as to where the track actually is. I generally zoom in quite close, such that the Google Earth scale is around 40m long.
4. Save the drawn Path as a kml file.
5. Copy the coordinates from the saved kml file into one of the existing kml files so as to pick up the correct file format (including changing the track name in two places). I use a spreadsheet to help with this, which does two things: truncate the coordinates to 6 decimal points (which is a little tidier, but not necessary), and calculate the length of the drawn path.
6. Upload the kml to the Tracks site.

This sounds a bit complicated, but it is actually straightforward and easy to do once you get the hang of it.

iodi wrote:Raw GPS logs are a bit rough to use directly. In particular:
- GPS logging is inherently imprecise, having a degree of inaccuracy that depends on the quality of the satellite reception conditions; and
- the Google Maps satellite/aerial photo images don't always align well with reality - eg. in the Wainuiomata area the Google maps are misaligned by 40-50m, which is way too much to ignore.

So, to create the kml files I do the following:
<snip>instructions</snip>

This sounds a bit complicated, but it is actually straightforward and easy to do once you get the hang of it.

Thanks for the detailed explanation. I collected some tracks on Mt Vic on the way home and went through a similar process except I used gpsbabel to grab the data as kml and loaded that in google earth. I then edited the downloaded track by selecting it and choosing properties on the right click menu. I managed to get a pretty good looking track this way.

Some observations:
1. Google earth is not the best track editor as its not really designed with editing in mind I guess, a feature that would be nice would be the ability to drag the whole track which would be useful where there is a consistent offset.
2. When the google data and the overlayed tracks are misaligned with respect to reality this may cause problems later, if a, google change their dataset/provider (which they do periodically) or b, the data is imported into an application with a more accurate/different dataset. This is just an observation I don't see howto avoid it.I am interested to see if its feasible to import the tracks.org.nz track data into openstreetmap (now possible due to the licence change) and whether this is an issue as most of their data (afaik) for nz comes from either gps or tracing on yahoo maps.
3. Its easier to visually improve a track when its not under trees and tracks under trees are generally less accurate, therefore I guess they will always be less accurate than exposed tracks.

donovan wrote:Google earth is not the best track editor as its not really designed with editing in mind I guess

True. Though it uses the same aerial photos as Google Maps, so tracks created in Google Earth work well when shown on a Google Map as on the Tracks site.

donovan wrote:a feature that would be nice would be the ability to drag the whole track which would be useful where there is a consistent offset.

That would be good. I use Excel to adjust the coordinates by a constant amount, which achieves the same effect albeit with a bit more effort.

donovan wrote:When the google data and the overlayed tracks are misaligned with respect to reality this may cause problems later, if a, google change their dataset/provider (which they do periodically) or b, the data is imported into an application with a more accurate/different dataset.

Yes, that is true. Fortunately the aerial photos around most of the Wellington region are aligned very well. Wainuiomata is an exception.

donovan wrote:Its easier to visually improve a track when its not under trees and tracks under trees are generally less accurate, therefore I guess they will always be less accurate than exposed tracks.

That's right. One solution to this problem is to record a track several times to smooth out the noise.

iodi wrote:

donovan wrote:Google earth is not the best track editor as its not really designed with editing in mind I guess

True. Though it uses the same aerial photos as Google Maps, so tracks created in Google Earth work well when shown on a Google Map as on the Tracks site.

donovan wrote:a feature that would be nice would be the ability to drag the whole track which would be useful where there is a consistent offset.

That would be good. I use Excel to adjust the coordinates by a constant amount, which achieves the same effect albeit with a bit more effort.

donovan wrote:When the google data and the overlayed tracks are misaligned with respect to reality this may cause problems later, if a, google change their dataset/provider (which they do periodically) or b, the data is imported into an application with a more accurate/different dataset.

Yes, that is true. Fortunately the aerial photos around most of the Wellington region are aligned very well. Wainuiomata is an exception.

donovan wrote:Its easier to visually improve a track when its not under trees and tracks under trees are generally less accurate, therefore I guess they will always be less accurate than exposed tracks.

That's right. One solution to this problem is to record a track several times to smooth out the noise.

For the auto routing maps I ran most of the tracks several times rather slowly with my GPS set to record every second and picked the one most to the center. I also found that different times of the day would also bring the track slightly south by about 2 or 3 meters depending on the time difference between the recorded tracks. Personally I wouldn't alter them in any way unless there is something obviously wrong. Chances are that Google will move or update their maps and then you will need to redo it.

I am buying a garmin 405 to accompany me on future rides and I might leave my 60csX on the bike aswell. This would give me 2 tracks from 1 run. In theory this should be enough as the two should be very close to each other at all times.

iodi wrote:That would be good. I use Excel to adjust the coordinates by a constant amount, which achieves the same effect albeit with a bit more effort.

Yea I was considering storing my stuff in a database so i could manipulate it with code more easily to do things like this.

iodi wrote:That's right. One solution to this problem is to record a track several times to smooth out the noise.

Yea I was considering doing that although its quite a lot of extra work, ok for tracks that are close by but harder to do for more remote ones.

Some years ago when working in outback Australia we used a technique called Differential GPS where we had a second gps at a know location recording points at the same time as the second was recording a track, we would then calculate the offset from the known point and use that offset to correct the track waypoints for each second. It seemed to work reasonably well. The units themselves supported this tho, setting this up with by hand would probably be a bit of a mission and may require close synchronization of the units.

Some other techniques: http://freegeographytools.com/2008/impr ... ers-part-i

ArjanH wrote:For the auto routing maps I ran most of the tracks several times rather slowly with my GPS set to record every second and picked the one most to the center. I also found that different times of the day would also bring the track slightly south by about 2 or 3 meters depending on the time difference between the recorded tracks. Personally I wouldn't alter them in any way unless there is something obviously wrong. Chances are that Google will move or update their maps and then you will need to redo it.

GPS accuracy varies by time according to the position of the satellites, which is probably why you're seeing different results at different times of the day. This phenomenom is called Dilution Of Precision, which can be predicted and hence it is possible to determine the best times to record positions. However, that's all getting a bit sophisticated - I just ride when I can and if the GPS log isn't good then I'll just have to ride the track again

The Google maps are spot on for most of Wellington, so there shouldn't be any significant issues if/when Google update the images. Wainuiomata is the main exception, where the 40-50m error is too big to ignore. If Google update the images around Wainuiomata, then the tracks will need to be updated too but that shouldn't be a big job.

donovan wrote:Some years ago when working in outback Australia we used a technique called Differential GPS where we had a second gps at a know location recording points at the same time as the second was recording a track, we would then calculate the offset from the known point and use that offset to correct the track waypoints for each second. It seemed to work reasonably well. The units themselves supported this tho, setting this up with by hand would probably be a bit of a mission and may require close synchronization of the units.

I used a similar technique for adjusting the tracks around Wainuiomata. I would record a waypoint at a point that would be clearly identifiable on Google Earth's images. I then record the track. Then when I load the GPS log into Google Earth I can see the difference between the recorded position of my first waypoint and where that point is on the image. This tells me how much to adjust the log by, which I do in a spreadsheet and then load the adjusted log into Google Earth as the basis for the track. Fortunately, this process has only been necessary in the Wainuiomata area.

donovan wrote:Some other techniques: http://freegeographytools.com/2008/impr ... ers-part-i

That's a great site, with all sorts of useful info and software.

iodi wrote:

groklem wrote:I assume you are also logging elevation data? any chance of putting that into the kml? would be great to see some kennett bros style distance vs elevation graphs on tracks.org.nz

I agree that the Tracks site is an awesome resource, which is why I contribute to it. You're welcome to contribute too.

snip

The HCx has an altimeter so I do collect elevation data. I believe that elevation charts are on the Tracks list of things to do.

I knocked together a quick script to do elevation charts and put an example on the site here:
http://tracks.org.nz/track/show/177

At the moment the script does kml only, but I will make it do gpx too because all your logs that contain height data are in gpx right?

The plan is to make a web form that allows upload of gpx/kml and displays a png or a gif that can then be embedded in the site.

Any thoughts let me know...

groklem wrote:I knocked together a quick script to do elevation charts and put an example on the site here:
http://tracks.org.nz/track/show/177

At the moment the script does kml only, but I will make it do gpx too because all your logs that contain height data are in gpx right?

The plan is to make a web form that allows upload of gpx/kml and displays a png or a gif that can then be embedded in the site.

Any thoughts let me know...

That's very cool.

I don't use the raw gpx log, for reasons described above, though the altitude profile would be easier to arrange if I did use the gpx directly. However, I could add altitude data to the kml by doing a bit of translation from the gpx coordinates to the manually drawn kml coordinates, so I'm sure something could be arranged.

Meanwhile, it would be best to have a chat with Jonny about adding altitude charts to the site, in terms of data format, data storage, presentation of the charts, managing the script, etc.

ArjanH wrote:I am buying a garmin 405 to accompany me on future rides and I might leave my 60csX on the bike aswell. This would give me 2 tracks from 1 run. In theory this should be enough as the two should be very close to each other at all times.

I would be interested to see the results. My gut feeling is that it will smooth out the results to a degree by geting rid of some of the noise but I would think that tracks recorded at different times of day/different days would be more valuable for creating an aggregate track. I'm riding Mt Vic 4+ times a week at the moment and logging everything so I should build up a bit of a corpus to test out some of these theories

ArjanH are you nefarious on tracks?

I'm off to ride/record Hylton park and Matipo park in Wanganui this weekend ...

groklem wrote:I knocked together a quick script to do elevation charts and put an example on the site here:
http://tracks.org.nz/track/show/177

At the moment the script does kml only, but I will make it do gpx too because all your logs that contain height data are in gpx right?

The plan is to make a web form that allows upload of gpx/kml and displays a png or a gif that can then be embedded in the site.

Any thoughts let me know...

Nice work groklem.

It would be nice to see a section for all the tracks. Personally I think the height should be less and the width more, from memory its also common to have sections with a multiplier on the height with respect to the width, so tracks that have less overall change have a higher multiplier. Some sort of standard would be cool, like having the width the same number of pixels as the gmaps window or something.

Speaking of standards it would be nice to have a consistent style and colour on the TrackLineStyle for the kml, billy T looks a little yellow and thick.

Speaking of Billy T, grade=Beginner? WTF, Im not taking my girlfriend down there ...

donovan wrote:

ArjanH wrote:ArjanH are you nefarious on tracks?

I'm off to ride/record Hylton park and Matipo park in Wanganui this weekend ...

Ah yes that's me. Sorry about the name change, I normally use Nefarious but for Some reason Vorb wouldn't let me have it.

I haven't been on a bike in over a month. Broke two ribs just before Xmas and been really crook with some sort of bug for the last 10 days.

I am hoping to dump a whole pile of tracks on tracks tonight ...