Bakeoff plus Tecumseh hike

Today Doug Paul, Paradox and I met to do the GPSr bakeoff, hiking up the Greeley Pond Trail to the lower pond and back with a variety of equipment. I suspect that Doug Paul will post a description of the experiments.

It was a gorgeous fall day, and we were finished (including downloading tracks to a laptop) a little after 1 PM. We decided to hike up Tecumseh.

Rather than drive up to the ski area I suggested that we take the Pipeline XC ski trail. It is not well known, but allows hikers to do an interesting Osceola-Tecumseh loop with one car: Greeley Ponds to Osceola Trail, over both peaks to Tripoli Road, down road to Tecumseh trailhead, and over Tecumseh to ski area. From there the Pipeline Trail affords a short (0.5 miles) way back to Depot Camp. (Note: We only went up Mt. Tecumseh. I mention the longer loop as it is an interesting loop that uses the Pipeline Trail.)

There is not much to say about the Tecumseh Trail, we followed the trail all the way to the summit, and left the summit by the Sossman Trail (i.e. we looped over the summit). I suggested that we continue on the Sossman Trail and go down the ski slopes, a trip I find very enjoyable in foliage season. The Sossman Trail has two fine outlooks, one to the west and the other, with a bench, to the east. On the eastern outlook we met a couple who were planning to go down the ski trails; this is the first time I have met other hikers on the Sossman Trail.

Going down the ski slopes we had great views of the surrounding mountains with their foliage. The views changed with changes in the light and with our loss of elevation.

This trip report is mainly a plug for the full Sossman Trail in all seasons. Steve Smith and Mike Dickerman, in their book The 4000-Footers of the White Mountains write:

Arguably, the mountain's most stunning vistas are those obtained from the viewpoints established several years ago along the ridgeline Sossman Trail.

Click to expand...

If you go to its end and return to the main trail it adds 1.2 miles (0.6 each way) to a very short hike, giving you views far superior to the minimal ones available on the summit. The ski slopes are a very attractive option in the fall; in other seasons I am satisfied with the views from the two outlooks on the Sossman trail.

After the hike we went to my place where Doug completed the downloading of the tracks, and perhaps we drank a beer or two.

A very enjoyable day; good company and great views. What more can one ask for?

It was a great experience meeting both DougPaul and Mohamed Elozy. The photos of mountains in the distance in Vermont and the Carters were taken from approximately NAD 83 19 T 294786 4870877. Well worth the trip.
pictures

Nice TR, Mohamed _ and nice pix, Paradox. It looks like the foliage is close to peak and beautiful this year.

I was carrying 3-4 lbs of electronics--3 GPSes, spare batteries, external antennas, etc. (What Paradox's first picture doesn't show is 2 external antennas under my white hat. It was fitting a bit loose with all that extra junk up top...). Traveled light and left most of the extraneous toys in the car for Tecumseh.

We have tracks for 7 GPS/antenna combinations. (Might be a problem with 1 track.) I'll have to do a bit of clean up and then I'll try to do an analysis. A quickie preliminary conclusion is that an external antenna on top of one's head gives a better track than the same GPS carried on one's pack shoulder strap. But that is hardly news.

GPS toys or not, it was a pleasant hike with good companions on a nice day. The colors weren't bad either...

Doug

Preliminary analysis

OK, I've done some preliminary work on the data and looked at it plotted on Garmin Mapsource in varying magnifications. All my comments are just my impressions, I haven't done anything quantitative.

My comments so far:
GPS-antenna ... Track Seg* .... Comments
60csx-ext ..... 5=1N+4S ....... coarse sampling, cannot tell
60csx-int ..... 3=1N+2S ....... break at end, best track?, good+

60cs-ext ...... 8=5N+3S ....... breaks at ends, best track?, good+
60cs-int ...... 7=4N+3S ....... .4mi break, otherwise good

vista-ext ..... 33=28N+5S ..... lots of breaks N, S track good
vista-int ..... 85=34N+51S .... bad GPS?, lots of breaks N+S, good-
eTrex-int ..... 33=22N+11S .... carried badly**, lots of breaks N+S, good-

* 2 Track Segments would have been perfect --we power cycled the GPSes at the turn around point to break the N and S into 2 separate tracks. The damage caused by the extra breaks depends on where they are, how big they are, and how long it takes the location to settle down after reacquiring the satellites.
** carried dangling upside down on users back. (We ran out of places to carry the GPSes...)

* int antennas: the GPSes were carried on pack straps at the front of the shoulder
* ext antennas: on top of head, on or in hat

* We hiked from Depot Camp (Livermore Rd parking lot, a wide open area) to the lower Greely Pond (higher and in a tight notch), and returned. We kept going pretty steadily, but there were a few short stops (insulation adjustments etc) and we stood around for a bit at the pond.
* S generally better than N, better satellite constellation?, more open sky for GPSes carried on pack shoulder straps?

* Something appears to have gone wrong with 60csx-ext. For some reason, it gave us a very coarse sampling, so we cannot meaningfully compare it to the other tracks. (All GPSes were set to give us trackpoints at 20 ft intervals.)
* Otherwise, the 60csx-int and 60cs-ext tracks are the best. Can't say at this point which looks better.
* The 60cs-int, except for a .4mi break, also did fairly well.
* The (eTrex) vista-ext had a lot of breaks going northward, but produced a good track going southward.
* The (eTrex) vista-int (possible poor GPS) and the (basic yellow) eTrex-int both produced a lot of breaks

* Note: I believe the eTrex Vista and the basic yellow eTrex GPSes use the same GPS chipset, and, except for manufacturing variation, should yield equivalent quality tracks.

I have attached the tracks in .gpx format. If you highlight individual track segments, you can distinguish the N from the S portions of the hike.

(edit: use the attachment from my next post.)

Doug

Quantitative analysis

I have now done a quantitative analysis of the GPS tracks by writing a program to analyze the tracks. The program removes all gaps from the tracks (by connecting the segments with straight lines) and computes the following:
1) The round trip track length. Just adds up the distance between successive points.
2) The 3 biggest distances between successive track points.
3) The average distance beween track points (=length/no_points).
4) I then computed the area of the zone outlined by the round-trip track and divided it by the one-way track length (= round_trip/2) to estimate the average width of the zone outlined by the track. This average width is hopefully proportional to the average cross-track error. The area of a track is not well defined (and difficult to estimate on messy tracks), so this measure could be questionable. But it is the best that I have...
5) The number of track points.
6) The number of track segments. (Since we power-cycled the GPSes at the turn-around point, the ideal number is two.) In a number of cases, there are very short (frequently only 1 point) tracks recorded at turn-on or turn-off time. The gaps in the tracks between the track segments were filled by drawing a straight line (not very accurate if the gap is big).

All distances are in meters.
** coarse track log, not comparable to the others
*+ carried dangling upside down on users back. (We ran out of places to carry the GPSes...)

Notes:
* The data were recorded on Sat, Sept 30, 2006. The leaves were mostly still on the trees. The hiking route was from Depot Camp (Livermore Rd parking lot, a wide open area) via Livermore Rd and Greely Pond Tr to the lower Greely Pond (higher and in a tight notch), and return. Location, Waterville Valley, White Mountains, NH, USA.
- start point: http://www.topozone.com/map.asp?lat=43.96575&lon=-71.51434&size=l&u=4&datum=nad27&layer=DRG
- turn-around point: http://www.topozone.com/map.asp?lat=44.00576&lon=-71.50463&size=l&u=4&datum=nad27&layer=DRG
* All GPSes were set to give fairly frequent track points (20 ft). (We may have set this improperly, leaving the GPSes set to "auto". There appear to be two places where one can set this, but there is no clarification in the manual.) For some unknown reason, the 60csx-ext recorded a much coarser tracklog, so these data do not give a meaningful measure of its performance in this configuration. I would expect it to be as accurate or more accurate that the 60csx-int.
* int (internal) antennas: the GPSes were carried on pack straps at the front of the shoulder
* ext (external) amplified antennas: on top of head, on or in hat
* The hiking path was chosen because the SW portion was relatively open and the N portion was in a fairly tight notch. Both sections were treed.
* There was a spurious point (not from this hike) in one of the tracks made available in the Preliminary analysis--it was removed. The corrected tracks are attached.
* We also tried to include a Magellan GPS, but it shut down at some unknown point during the hike, so we dropped it from the study.
* vista = eTrex Vista
* etrex = basic (yellow) eTrex

Conclusions:
1) An external antenna on top of one's head gives better tracks than the internal antenna when the GPS is carried attached to one's pack shoulder strap. Presumably, the internal antenna would perform as well as the external antenna if the GPS were carried on top of one's head (and oriented properly).
2) On internal antennas: the 60csx-int > 60cs-int > vista-int ~ etrex-int. (">" = better than, "~" = similar)
3) On external antennas: 60cs-ext > vista-est. (We cannot draw any conclusions about the 60csx-ext due to the coarse track.)
4) The vista-int GPS was previously believed to be a sub-standard unit--the data comparing it to the etrex-int is somewhat conficted--slightly better track width, but more track breaks. (I believe the vista-int and the etrex-int contain the same GPS chipsets.)
5) I would expect the 60csx-ext to work better than the 60csx-int and likely better than the 60cs-ext, but these data do not support this supposition. (They also do not disprove it.)

Doug

Attachment: a map of the route and the corrected tracks in GPX format.

Nicely done Paul, thank you . We can do it again with an improved protocol, when the leaves are back on the trees. All things considered, it looks like an incredible piece of technology, these GPS's. BTW that was a pretty slick camera you had that Saturday, do you have the pictures posted anywhere?

Paradox said:

Nicely done Paul, thank you . We can do it again with an improved protocol, when the leaves are back on the trees. All things considered, it looks like an incredible piece of technology, these GPS's. BTW that was a pretty slick camera you had that Saturday, do you have the pictures posted anywhere?

Click to expand...

I think the protocol was basically ok, but a few details could be improved. And out of 8 GPSes, we had only 1 failure and 1 didn't-quite-do-what-we-wanted. Not too bad for a first try with no dry run.

At least I now have an analysis program written--it took a good bit longer to work out the details of the analysis than it did to collect the data...

The camera (a DSLR) is reasonable, but not all that slick--others have the same or better. Unfortunately I don't have a web site and haven't investigated the picture album sites, so I don't have any place to post them. That reminds me--I should email you a copy of the pic of you and the dogs.

Doug

Yes, email the photos to me and I will post them to my Webshots.

Paradox said:

Yes, email the photos to me and I will post them to my Webshots.

Click to expand...

Thanks--I got off my tail and activated my website privilages...

You can look at them here:
http://mysite.verizon.net/dbpwebjunk/bake-off/index.html

Doug