More GPS

Roger said:

I took my etrex Summit on another visit to The Lakes today and
again had problems with cumulative ascent and descent. I am working my
way through AWs Outlying Fells so had 4 very short walks today. The
totals for ascent and descent (I am no longer sure which is which in
most cases are:

799 554 1082

1056

Quoted message said:

805 454 1202

1130

Quoted message said:

Discrepancy 6 100 120 74

GPS was set at home after making sure there was a 3D lock and was on
continuously until the final return to the car.

Do you think these odd figures could be another bug of the Summit Roger?
I haven't noticed anything as bad with mine, at least not yet.

I did notice after a short cycle ride last night that my ascent and
descent had a 5m discrepancy, but I put that down to not allowing any
time for the autocalibration to settle down, I just switched it on, got
a lock, calibrated it and rode off. When I got back I switched it
straight off.

The odd thing was that when I switched it back on indoors, with no lock,
the ascent and descent totals then matched perfectly! (although the
altitude had dropped by 5m or so)

Leaving it switched on for some time (with no GPS lock) while I checked
other data and played around with it, I noticed the ascent and descent
steadily increasing, as I've often noticed before. Even when perfectly
stationary (on the table), I've watched the ascent increase over time,
in this instance it increased from 66m to 138m over perhaps half an hour
to an hour (I didn't time it). This is no doubt due to adding up the
small fluctuations of the barometric altimeter - it typically "wobbles"
over a height range of 1-2 metres. All this adds up.

This explains the overly high ascents you've talked about, even when
stationary it increments the ascent. So while it may be true that it's
very sensitive and adds up all the little ups and downs, it's also
adding wobble. This no doubt happens when moving as well as when
stationary. When not moving, the ascent and descent increment
identically, but when moving perhaps the wobble causes one to be
incremented more than the other depending on whether you're moving
uphill or downhill?

This may explain some of your discrepancies. Have you noticed any
correlation between which one increments the most and how much time you
spent walking uphill compared to downhill?

Paul
--
http://www.wilderness-wales.co.uk
http://www.wildwales.fsnet.co.uk

The message <[email hidden]>
from "Paul Saunders" <[email hidden]> contains these words:

snip

Quoted message said:

Do you think these odd figures could be another bug of the Summit Roger?
I haven't noticed anything as bad with mine, at least not yet.

I am not sure. I have had a theory for years that the underlying problem
is temperature related but that could easily be wrong.

Quoted message said:

I did notice after a short cycle ride last night that my ascent and
descent had a 5m discrepancy, but I put that down to not allowing any
time for the autocalibration to settle down, I just switched it on, got
a lock, calibrated it and rode off. When I got back I switched it
straight off.

Quoted message said:

The odd thing was that when I switched it back on indoors, with no lock,
the ascent and descent totals then matched perfectly! (although the
altitude had dropped by 5m or so)

That sounds as though the underlying problem could be time lag. I
usually check the times as soon as I stop starting with trip time which
of course continues to accumulate. Dropping 16 feet would have been
painful had it really happened. :-)

Quoted message said:

Leaving it switched on for some time (with no GPS lock) while I checked
other data and played around with it, I noticed the ascent and descent
steadily increasing, as I've often noticed before. Even when perfectly
stationary (on the table), I've watched the ascent increase over time,
in this instance it increased from 66m to 138m over perhaps half an hour
to an hour (I didn't time it). This is no doubt due to adding up the
small fluctuations of the barometric altimeter - it typically "wobbles"
over a height range of 1-2 metres. All this adds up.

If you don't have a lock that would be equivalent to turning auto-calc
off wouldn't it?

Quoted message said:

This explains the overly high ascents you've talked about, even when
stationary it increments the ascent. So while it may be true that it's
very sensitive and adds up all the little ups and downs, it's also
adding wobble. This no doubt happens when moving as well as when
stationary. When not moving, the ascent and descent increment
identically, but when moving perhaps the wobble causes one to be
incremented more than the other depending on whether you're moving
uphill or downhill?

Except that yesterday the one set of figures I know for sure which way
round they are (still in the machine although both slightly increased,
1056 - 1060, 1130 - 1132) have greater descent than ascent.

Quoted message said:

This may explain some of your discrepancies. Have you noticed any
correlation between which one increments the most and how much time you
spent walking uphill compared to downhill?

Not something I have noticed but even I (slow as I am downhill) spend
more time in ascent than descent.

Looking at the end of yesterdays track I think there may be something in
the lag idea. The height recorded when I left the car was 772 feet,
immediately I returned 751 feet but by the time I turned the GPS off 1.5
minutes later the discrepancy had dropped from 21 feet to 6 feet. This
is curious because either of these discrepancies are anything like the
cumulative register discrepancy of 74 feet. Even more curious is that
the heights from the track list give the cumulative ascent to Hesk Fell
as 872 feet, not 750. And that is not the end of the curiosities. On The
Pike there seemed to be a point the other side of the summit wall that
might be at least as high so I marked both points. The waypoints give
the 2nd point as 5 feet above the first while the track has the first
higher than the second on both outward and return but the return pass 2
foot lower than the outward. No doubt I am reading too much significance
into these figures but I does seem odd that there should be such changes
over such a small time scale and between points only 23 feet apart.

--
Roger Chapman so far this year 69 summits
New - 47 (Marilyns 13, Sweats 5, Outlying Fells 35)
Repeats - 22( Marilyns 7, Sweats 16, Wainwrights 16)

Roger said:

The message <[email hidden]>
from "Paul Saunders" <[email hidden]> contains these words:

Quoted message said:

Quoted message said:

Do you think these odd figures could be another bug of the Summit Roger?
I haven't noticed anything as bad with mine, at least not yet.

Nor mine. In fact in the one test I've done of this, after reading one
of Paul's posts about it, I got very good agreement between figures of
ascent and descent. The figures were larger than I expected, but when
I checked the tracklogs against the actual route (a 5 mile cycling
commute, out in the morning, back in the evening), they seemed
reasonable, i.e., matched the very hilly terrain.

Quoted message said:

Quoted message said:

I did notice after a short cycle ride last night that my ascent and
descent had a 5m discrepancy, but I put that down to not allowing any
time for the autocalibration to settle down, I just switched it on, got
a lock, calibrated it and rode off. When I got back I switched it
straight off.

Quoted message said:

Quoted message said:

The odd thing was that when I switched it back on indoors, with no lock,
the ascent and descent totals then matched perfectly! (although the
altitude had dropped by 5m or so)

Quoted message said:

That sounds as though the underlying problem could be time lag. I
usually check the times as soon as I stop starting with trip time which
of course continues to accumulate. Dropping 16 feet would have been
painful had it really happened. :-)

There could be a time lag involved. I note that the Summit reports
altitude in the active log to a decimeter, but small movements up and
down are not reported in the up/down arrows which show ascent and
descent rate. It would be very silly of them not to employ some kind
of noise reduction in accumulating ascent and descent, because at the
resolution of a decimeter there is a lot of noise. And noise reduction
filters involve a time lag.

In fact auto-calibration incorporates a GPS altitude noise reduction
with a time period of about half an hour. There's also noise reduction
in speed reporting which seems to involve a time period of a few
seconds. I'd be very surprised if there wasn't noise reduction
filtering on the ascent and descent summations.

Quoted message said:

Quoted message said:

Leaving it switched on for some time (with no GPS lock) while I checked
other data and played around with it, I noticed the ascent and descent
steadily increasing, as I've often noticed before. Even when perfectly
stationary (on the table), I've watched the ascent increase over time,
in this instance it increased from 66m to 138m over perhaps half an hour
to an hour (I didn't time it). This is no doubt due to adding up the
small fluctuations of the barometric altimeter - it typically "wobbles"
over a height range of 1-2 metres. All this adds up.

Quoted message said:

If you don't have a lock that would be equivalent to turning auto-calc
off wouldn't it?

Yes.

I wonder if the altimeters in some Summits are noisier than others?
That would explain why some folk get so much worse over-estimations of
ascent and descent than others.

In the sci.geo.sat-nav group, one user, who has played with a lot of
eTrexes, has said that some have much poorer signal-to-noise on their
GPS reception than others. Maybe there are quality control
problems. Or maybe, like cheap binoculars, some are accidentally
pretty good by a fluke of component variations matching very well.

It also occurs to me that some locations may be noisier than others
with respect to barometric variation. For example, I would expect
someone moving around in the wind shadow of a mountain range during
high winds to have a lot more immediate variations in pressure due to
turbulence, gusting, etc., than someone moving around in the middle of
a wide plain on a calm day.

Quoted message said:

Except that yesterday the one set of figures I know for sure which way
round they are (still in the machine although both slightly increased,
1056 - 1060, 1130 - 1132) have greater descent than ascent.

How much of this is explained by differences in the reported start and
finish alitude?

Quoted message said:

Looking at the end of yesterdays track I think there may be something in
the lag idea. The height recorded when I left the car was 772 feet,
immediately I returned 751 feet but by the time I turned the GPS off 1.5
minutes later the discrepancy had dropped from 21 feet to 6 feet.

That's what would have happened if the Summit had been operating in 2D
mode due to poor sat view, and then when you stopped, you did
something which improved reception (took it out of a pocket and put it
down flat facing up?) which permitted it to recover 3D mode, which
would have caused autocalibrate to go into the fast recovery mode.

Quoted message said:

This
is curious because either of these discrepancies are anything like the
cumulative register discrepancy of 74 feet. Even more curious is that
the heights from the track list give the cumulative ascent to Hesk Fell
as 872 feet, not 750. And that is not the end of the curiosities. On The
Pike there seemed to be a point the other side of the summit wall that
might be at least as high so I marked both points. The waypoints give
the 2nd point as 5 feet above the first while the track has the first
higher than the second on both outward and return but the return pass 2
foot lower than the outward. No doubt I am reading too much significance
into these figures but I does seem odd that there should be such changes
over such a small time scale and between points only 23 feet apart.

That kind of disrepancy could easily arise if the unit was switching
between 2D mode and 3D mode, which brings in the fast recovery mode of
autocal.

I'm beginning to wonder if the reason I haven't found such
discrepancies is that one of the first things I taught myself to do
with my Summit was how to get the best tracks by giving it the best
sat view, and whenever I do these kinds of experiments I'm careful to
get as good tracks as possible, simply because in poor tracking
conditions so many extra confounding variables are introduced.

How do I get good tracks with good sat view? The unit is always
horizontal, and whenever I'm near a cliff or slope I make sure that
any head or body shadow from me is on the same side, rather than
subtracting from the already clear view, e.g., when I walk up the unit
is behind me, when I walk down it's behind, when I walk beside a cliff
the unit is on the outside.

Of course I can't always be bothered taking such precautions, and
often simply leave it in a pocket hoping to get a mostly reasonable
track. Under such circumstances I've even had it report sudden
excursions at impossibly high speeds some hundreds of miles in the
direction of Norway. Another time it nearly got to Greenland.

--
Chris Malcolm [email hidden] +44 (0)131 651 3445 DoD #205
IPAB, Informatics, JCMB, King's Buildings, Edinburgh, EH9 3JZ, UK
[http://www.dai.ed.ac.uk/homes/cam/]

The message <[email hidden]>
from [email hidden] (Chris Malcolm) contains these words:

Quoted message said:

Quoted message said:

Quoted message said:

Do you think these odd figures could be another bug of the Summit Roger?
I haven't noticed anything as bad with mine, at least not yet.

Quoted message said:

Nor mine. In fact in the one test I've done of this, after reading one
of Paul's posts about it, I got very good agreement between figures of
ascent and descent. The figures were larger than I expected, but when
I checked the tracklogs against the actual route (a 5 mile cycling
commute, out in the morning, back in the evening), they seemed
reasonable, i.e., matched the very hilly terrain.

I have had very good agreement on some trips in the past but on more
occasions the difference is disturbingly large. Until this last trip I
have taken it as read that the cumulative heights given must be
exaggerations to some degree but that was certainly not the case on the
Hesk Fell work where even the larger figure was no greater than that
culled from the map.

snip

Quoted message said:

Quoted message said:

That sounds as though the underlying problem could be time lag. I
usually check the times as soon as I stop starting with trip time which
of course continues to accumulate. Dropping 16 feet would have been
painful had it really happened. :-)

Quoted message said:

There could be a time lag involved. I note that the Summit reports
altitude in the active log to a decimeter, but small movements up and
down are not reported in the up/down arrows which show ascent and
descent rate. It would be very silly of them not to employ some kind
of noise reduction in accumulating ascent and descent, because at the
resolution of a decimeter there is a lot of noise. And noise reduction
filters involve a time lag.

I use Imperial measurements and if Ozi can be believed altitude is
quoted in feet with no finer discrimination although distance is given
to the nearest 0.1feet between track points.

Quoted message said:

In fact auto-calibration incorporates a GPS altitude noise reduction
with a time period of about half an hour. There's also noise reduction
in speed reporting which seems to involve a time period of a few
seconds. I'd be very surprised if there wasn't noise reduction
filtering on the ascent and descent summations.

Quoted message said:

Quoted message said:

Quoted message said:

Leaving it switched on for some time (with no GPS lock) while I checked
other data and played around with it, I noticed the ascent and descent
steadily increasing, as I've often noticed before. Even when perfectly
stationary (on the table), I've watched the ascent increase over time,
in this instance it increased from 66m to 138m over perhaps half an hour
to an hour (I didn't time it). This is no doubt due to adding up the
small fluctuations of the barometric altimeter - it typically "wobbles"
over a height range of 1-2 metres. All this adds up.

Quoted message said:

Quoted message said:

If you don't have a lock that would be equivalent to turning auto-calc
off wouldn't it?

Quoted message said:

Yes.

Quoted message said:

I wonder if the altimeters in some Summits are noisier than others?
That would explain why some folk get so much worse over-estimations of
ascent and descent than others.

I am no longer convinced that it does overestimate on a regular basis.

Quoted message said:

In the sci.geo.sat-nav group, one user, who has played with a lot of
eTrexes, has said that some have much poorer signal-to-noise on their
GPS reception than others. Maybe there are quality control
problems. Or maybe, like cheap binoculars, some are accidentally
pretty good by a fluke of component variations matching very well.

Quoted message said:

It also occurs to me that some locations may be noisier than others
with respect to barometric variation. For example, I would expect
someone moving around in the wind shadow of a mountain range during
high winds to have a lot more immediate variations in pressure due to
turbulence, gusting, etc., than someone moving around in the middle of
a wide plain on a calm day.

Friday was very close to being totally calm all day.

Quoted message said:

Quoted message said:

Except that yesterday the one set of figures I know for sure which way
round they are (still in the machine although both slightly increased,
1056 - 1060, 1130 - 1132) have greater descent than ascent.

Quoted message said:

How much of this is explained by differences in the reported start and
finish alitude?

Already covered, see below.

Quoted message said:

Quoted message said:

Looking at the end of yesterdays track I think there may be something in
the lag idea. The height recorded when I left the car was 772 feet,
immediately I returned 751 feet but by the time I turned the GPS off 1.5
minutes later the discrepancy had dropped from 21 feet to 6 feet.

Quoted message said:

That's what would have happened if the Summit had been operating in 2D
mode due to poor sat view, and then when you stopped, you did
something which improved reception (took it out of a pocket and put it
down flat facing up?) which permitted it to recover 3D mode, which
would have caused autocalibrate to go into the fast recovery mode.

I learnt my lesson long ago and now my summit always sits on my right
shoulder when in use. Either mounted on my rucksack strap or, on the
rare occasions I am not carrying a rucksack, on my braces.

Quoted message said:

Quoted message said:

This
is curious because either of these discrepancies are anything like the
cumulative register discrepancy of 74 feet. Even more curious is that
the heights from the track list give the cumulative ascent to Hesk Fell
as 872 feet, not 750. And that is not the end of the curiosities. On The
Pike there seemed to be a point the other side of the summit wall that
might be at least as high so I marked both points. The waypoints give
the 2nd point as 5 feet above the first while the track has the first
higher than the second on both outward and return but the return pass 2
foot lower than the outward. No doubt I am reading too much significance
into these figures but I does seem odd that there should be such changes
over such a small time scale and between points only 23 feet apart.

Quoted message said:

That kind of disrepancy could easily arise if the unit was switching
between 2D mode and 3D mode, which brings in the fast recovery mode of
autocal.

It could be getting a poor view of the available satellites when mounted
on the dashboard. It certainly shouldn't when out walking except when in
tree cover or on a steep hillside and the route on Hesk Fell is devoid
of trees and while I struggled to keep going on the upper slopes of Hesk
Fell it couldn't really be described as steep and was a broad ridge so
the lateral view was good, at least to one side. Anyway that doesn't
explain why the cumulative register lost about 100 feet on the ascent
compared with the lowest/highest point range on the track. Looking at
the map I would have said that the track height was 15- 20 feet too high
for the whole of that first ascent and even at the end of the walk was
still registering too high although the dip shortly before the end could
have been too low if the contours were accurate (big if).

Quoted message said:

I'm beginning to wonder if the reason I haven't found such
discrepancies is that one of the first things I taught myself to do
with my Summit was how to get the best tracks by giving it the best
sat view, and whenever I do these kinds of experiments I'm careful to
get as good tracks as possible, simply because in poor tracking
conditions so many extra confounding variables are introduced.

Quoted message said:

How do I get good tracks with good sat view? The unit is always
horizontal, and whenever I'm near a cliff or slope I make sure that
any head or body shadow from me is on the same side, rather than
subtracting from the already clear view, e.g., when I walk up the unit
is behind me, when I walk down it's behind, when I walk beside a cliff
the unit is on the outside.

I don't go quite to such lengths but OTOH when in use for walking it is
always on my shoulder or in my hand. I had it in my hand coming off Hesk
Fell towards The Pike to see whether the slope was steep enough to slow
me down. (It was, most of the time I was between 3 and 3.5 mph and on
occasion down to 2.7 mph. 3.5 mph was my flat surface rate shortly
afterwards. The steep part of the slope averaged 1 in 4 according to the
first series 1:25000)

--
Roger Chapman so far this year 69 summits
New - 47 (Marilyns 13, Sweats 5, Outlying Fells 35)
Repeats - 22( Marilyns 7, Sweats 16, Wainwrights 16)

Roger said:

Quoted message said:

The odd thing was that when I switched it back on indoors, with no
lock, the ascent and descent totals then matched perfectly!
(although the altitude had dropped by 5m or so)

I've just checked the track profile carefully in Ozi and calculating the
total ascent myself from the high and low points (ignoring three
obviously spurious points), to a tenth of a metre, the result is 71.9m.
This is 5.9m more than the 66m total ascent reported by my GPS!

Clearly the ascent is calculated independently within the GPS and is not
derived from the track, as is the case with distance. I've always
thought that the track is a better source of statistics if you take the
trouble to check it later - you can also correct any spurious points to
get a more accurate result.

So could it simply be that the GPS itself isn't very good at calculating
ascent? I was surprised that my track calculation resulted in higher
figure, since that shouldn't include the "wobble" factor.

Quoted message said:

That sounds as though the underlying problem could be time lag.

I mentioned a 5m discrepancy between ascent and descent immediately at
the end of my ride, but the track doesn't support this. The starting
ascent recorded in my track is 70.8m, while the last track point
(recorded in exactly the same spot) lists 71.3m, a discrepancy of only
0.5m. So there definitely seems to be a lag. As I say, they matched
correctly when I switched it back on later.

One slight curiosity. I started from my driveway, which is slightly
higher than the road, perhaps a foot. The GPS didn't detect a drop to
the road at the start of my ride, but at the end it recorded three
heights of 71.7m on the road, then it jumped up to 73.2m as I entered my
driveway, then dropped back to 71.3m 16 seconds later. That was the
last point in the track.

I treated the 73.2m height as a momentary error and ignored it in my
calculation of ascent, but it's interesting. I've read that the
altimeter is exremely sensitive, so could the slight rise to the
driveway have initiated that sudden jump of 1.5m? (or perhaps a pressure
change as I entered the drive?)

Curious also that the height settled at 0.4m lower than the three road
heights, even though it was slightly higher, but that's probably just
the wobble effect, or maybe a lag? Given that it typically tends to
wobble over a range of 1-2m, I doubt I should try to read anything
significant into such a small change.

Quoted message said:

I
usually check the times as soon as I stop starting with trip time
which of course continues to accumulate.

I never had a total trip time with the GPS12 so I just timed my walks
with a stopwatch instead (which I'd been doing for years anyway before
even getting a GPS). I'm still doing that, I don't like the way the
trip time continues to accumulate on my new GPS (and interestingly, it
only does this if it has a lock). My GPS12's elapsed time was useful
for measuring battery life, and that incremented constantly whenever the
unit was on, irrespective of lock. Pity my new GPS doesn't have that
feature (the total trip time is reset with the rest of the trip data,
unlike the 12's elapsed time).

Anyway, you don't really have to time your walks at all if you record a
track, you can easily determine the length of your walk by simply
comparing the times of the first and last track points, if of course
you've got that data, which you may often not.

As for the cumulative ascent, I'd check that first if I were you, since
that tends to increment quite quickly due to wobble.

Quoted message said:

Quoted message said:

Leaving it switched on for some time (with no GPS lock) while I
checked other data and played around with it, I noticed the ascent
and descent steadily increasing, as I've often noticed before.

If you don't have a lock that would be equivalent to turning auto-calc
off wouldn't it?

Yes, but I'm pretty sure I've seen the same thing happen when sitting in
the garden when I've had a lock. I'll have to check that.

Quoted message said:

And that is not the end of
the curiosities. On The Pike there seemed to be a point the other
side of the summit wall that might be at least as high so I marked
both points.

Did you do any averaging when marking? I don't know if yours works the
same as mine, but if so, a normally marked waypoint records the
altimeter elevation, while an averaged waypoint records an average of
the GPS elevation. These can of course be quite different.

Quoted message said:

The waypoints give the 2nd point as 5 feet above the
first while the track has the first higher than the second on both
outward and return but the return pass 2 foot lower than the outward.

If you've recorded altimeter height in the waypoints then much will
depend on the wobble factor. I've tried marking three bumps on a summit
to try to determine which was higher, using both normal and averaged
waypoint methods, and also compared with the track later. In each
instance I observed the wobble and tried to mark the waypoint at the
middle point of the wobble (so if it wobbled from 590 to 592 I tried to
mark it when it was at 591). Checking the track later though, I found
that many trackpoints were dropped in each spot, recording the different
heights as it wobbled. Taking the highest point of each of these gave
me the best indication of the relative height of these three points as I
imagined them to be.

Perhaps it would suffice to simply walk over the high points and check
the track later? Marking a waypoint seems to be a hit or miss way to
record an accurate altimeter height, it's a pity that it won't average
the altimeter wobble instead of the GPS altitude. It would be nice to
have a choice.

Quoted message said:

No doubt I am reading too much significance into these figures but I
does seem odd that there should be such changes over such a small
time scale and between points only 23 feet apart.

Your 5ft difference is only 1.5m. I see my GPS wobbling between 70m and
72m just sitting here at my desk. Either it's extremely sensitive and
there are tiny pressure fluctuations going on all the time in what seems
to be still air, or it just isn't accurate enough to measure altitude to
that degree of accuracy. I suspect the latter.

Paul
--
http://www.wilderness-wales.co.uk
http://www.wildwales.fsnet.co.uk

Chris Malcolm said:

There could be a time lag involved. I note that the Summit reports
altitude in the active log to a decimeter, but small movements up and
down are not reported in the up/down arrows which show ascent and
descent rate.

They do on my 60CS.

Just a quick experiment, leaving it sitting on the table in front of me
for five minutes, after calibrating it to 71m.

Approx 20 wobbles between 71 and 72, 73, 72, lots more 71-72, 74, 73,
72, many more 71-72. So a range of 71-74 over 5 minutes, mostly 71-72.
Perhaps 50-60 fluctuations over the 5 minute period.

During this time I saw many up/down arrows indicating ascent/descent
rate. I tried to write all of these down as they flashed up. I got
most of them, but missed a few due to three or four sometimes happening
in quick succession. I recorded the following 20 fluctuations during
the five minute period, and probably missed at least another 5, if not
10.

u 7.7, d 6.7, d 5.6, u 6.6, u 4.8, u 4.7, u 6.4, d 4.7, u 5.6, d 5.0,
u 7.3, u 4.6, d 4.8, u 4.9, u 4.8, u 4.9, u 7.0, d 5.1, u 5.3, u 1.6

Many of these accompanied the height changes reported above, but many
didn't. Most of the 71-72 fluctuations had no up or down rate
accompanying them, but occasionally they did. The 73 and 74 readings
had lots of up/down rates associated with them. Occasionally ups and
downs were reported with no elevation change.

I wanted to record these changes in cumulative ascent/descent totals,
but my GPS refused to do that with track recording switched off. It
wouldn't let me switch track recording on without getting a lock first.
I couldn't be bothered to go outside and stand in the cold wind to get a
lock.

If track recording had been switched on though, I'm sure the
ascent/descent totals would have incremented, because that's exactly
what they did last night, after forgetting to switch track recording off
after my ride. It seems that altimeter stats are linked to track
recording, and will continue to increment if track recording is switched
on - even if indoors with no lock! Having no lock prevents turning
track recording on (if it's been switched off), but if it's still on it
will continue to record trackpoints and update altimeter data, even
without a lock.

Quoted message said:

It would be very silly of them not to employ some kind
of noise reduction in accumulating ascent and descent, because at the
resolution of a decimeter there is a lot of noise.

So I've noticed. But it seems there isn't any noise reduction in my
60CS, judging by the above test, unless you care to interpret it
differently.

Quoted message said:

In fact auto-calibration incorporates a GPS altitude noise reduction
with a time period of about half an hour. There's also noise reduction
in speed reporting which seems to involve a time period of a few
seconds.

I've seen direction lag when turning sharp corners too.

Quoted message said:

I'd be very surprised if there wasn't noise reduction
filtering on the ascent and descent summations.

Well there doesn't seem to be any on mine.

Quoted message said:

I wonder if the altimeters in some Summits are noisier than others?
That would explain why some folk get so much worse over-estimations of
ascent and descent than others.

Well my 60CS seems to be very noisy, or very sensitive, yet as I said,
my manually measured ascent turned out to be greater than the GPS
estimate yesterday.

Quoted message said:

In the sci.geo.sat-nav group, one user, who has played with a lot of
eTrexes, has said that some have much poorer signal-to-noise on their
GPS reception than others.

Mine has a different type of aerial and gets very good reception, at
least as good as the GPS12.

Quoted message said:

It also occurs to me that some locations may be noisier than others
with respect to barometric variation. For example, I would expect
someone moving around in the wind shadow of a mountain range during
high winds to have a lot more immediate variations in pressure due to
turbulence, gusting, etc., than someone moving around in the middle of
a wide plain on a calm day.

I've noticed some sudden jumps when turning around in strong wind.

Quoted message said:

How do I get good tracks with good sat view? The unit is always
horizontal,

Mine needs to be vertical for best sat reception, which is great for use
in the car, and for hanging on my rucksack strap. Not so good for
cycling though, it's almost horizontal on the handlebar mount (but I
suppose that's good if I ever need to turn the compass on when in a
forest). I still get pretty good reception on a bicycle though, since
it's a bit further away from my body.

Quoted message said:

Of course I can't always be bothered taking such precautions, and
often simply leave it in a pocket hoping to get a mostly reasonable
track. Under such circumstances I've even had it report sudden
excursions at impossibly high speeds some hundreds of miles in the
direction of Norway. Another time it nearly got to Greenland.

I've never had anything remotely that bad with my GPS12. A jump of a
few hundred metres is about the worst I've seen.

Paul
--
http://www.wilderness-wales.co.uk
http://www.wildwales.fsnet.co.uk

The message <[email hidden]>
from "Paul Saunders" <[email hidden]> contains these words:

snip

Quoted message said:

Clearly the ascent is calculated independently within the GPS and is not
derived from the track, as is the case with distance. I've always
thought that the track is a better source of statistics if you take the
trouble to check it later - you can also correct any spurious points to
get a more accurate result.

Quoted message said:

So could it simply be that the GPS itself isn't very good at calculating
ascent? I was surprised that my track calculation resulted in higher
figure, since that shouldn't include the "wobble" factor.

It is beginning to look that way.

Quoted message said:

Quoted message said:

That sounds as though the underlying problem could be time lag.

Quoted message said:

I mentioned a 5m discrepancy between ascent and descent immediately at
the end of my ride, but the track doesn't support this. The starting
ascent recorded in my track is 70.8m, while the last track point
(recorded in exactly the same spot) lists 71.3m, a discrepancy of only
0.5m. So there definitely seems to be a lag. As I say, they matched
correctly when I switched it back on later.

Quoted message said:

One slight curiosity. I started from my driveway, which is slightly
higher than the road, perhaps a foot. The GPS didn't detect a drop to
the road at the start of my ride, but at the end it recorded three
heights of 71.7m on the road, then it jumped up to 73.2m as I entered my
driveway, then dropped back to 71.3m 16 seconds later. That was the
last point in the track.

I just don't trust height difference with adjacent heights. Too much
risk of error.

Quoted message said:

I treated the 73.2m height as a momentary error and ignored it in my
calculation of ascent, but it's interesting. I've read that the
altimeter is exremely sensitive, so could the slight rise to the
driveway have initiated that sudden jump of 1.5m? (or perhaps a pressure
change as I entered the drive?)

Quoted message said:

Curious also that the height settled at 0.4m lower than the three road
heights, even though it was slightly higher, but that's probably just
the wobble effect, or maybe a lag? Given that it typically tends to
wobble over a range of 1-2m, I doubt I should try to read anything
significant into such a small change.

Quoted message said:

Quoted message said:

I
usually check the times as soon as I stop starting with trip time
which of course continues to accumulate.

Quoted message said:

I never had a total trip time with the GPS12 so I just timed my walks
with a stopwatch instead (which I'd been doing for years anyway before
even getting a GPS). I'm still doing that, I don't like the way the
trip time continues to accumulate on my new GPS (and interestingly, it
only does this if it has a lock). My GPS12's elapsed time was useful
for measuring battery life, and that incremented constantly whenever the
unit was on, irrespective of lock. Pity my new GPS doesn't have that
feature (the total trip time is reset with the rest of the trip data,
unlike the 12's elapsed time).

Quoted message said:

Anyway, you don't really have to time your walks at all if you record a
track, you can easily determine the length of your walk by simply
comparing the times of the first and last track points, if of course
you've got that data, which you may often not.

Grr!!! I don't think I have ever been out for a days walking when I
didn't have to save the track at least once and saved tracks lose their
time stamps. Batteries OTOH usually last the whole day and I don't think
I have had to change them in the field since I switched to 21 mah
capacity.

Quoted message said:

As for the cumulative ascent, I'd check that first if I were you, since
that tends to increment quite quickly due to wobble.

I am beginning to think I should wait a bit longer so that any lag is
overcome. The odd 2 or 3 feet of stationary wobble is unlikely to
feature in the final tally as I tend to round totals down anyway. Or at
least I have done so far in the expectation that the registers were over
recording. I must go back to the record some time and see if I can
establish whether that 5800 foot walk I did last year when I was fitter
than usual was, as I thought, somewhat exaggerated.

snip.

Quoted message said:

Quoted message said:

And that is not the end of
the curiosities. On The Pike there seemed to be a point the other
side of the summit wall that might be at least as high so I marked
both points.

Quoted message said:

Did you do any averaging when marking? I don't know if yours works the
same as mine, but if so, a normally marked waypoint records the
altimeter elevation, while an averaged waypoint records an average of
the GPS elevation. These can of course be quite different.

Averaging is something I haven't even investigated. As you know I don't
like hanging about wasting time. :-)

snip

Quoted message said:

Perhaps it would suffice to simply walk over the high points and check
the track later? Marking a waypoint seems to be a hit or miss way to
record an accurate altimeter height, it's a pity that it won't average
the altimeter wobble instead of the GPS altitude. It would be nice to
have a choice.

I noted, looking at the track over the fairly flat summit of Hesk Fell
that the highest point on the recorded track was at the point where I
realised that I had already passed the summit which was marked by little
more than a single stone a few yards from the path. The height
difference might not have been much but I was sure at the time that it
was higher rather than 2 feet lower but seeing as how I stood on the
summit of Bleatarn Hill (951 ft) and thought it looked higher than the
adjacent summit of Boat How (1105 ft) a mile and a half away perhaps my
judgement in these matters cannot be relied upon.

Quoted message said:

Quoted message said:

No doubt I am reading too much significance into these figures but I
does seem odd that there should be such changes over such a small
time scale and between points only 23 feet apart.

Quoted message said:

Your 5ft difference is only 1.5m. I see my GPS wobbling between 70m and
72m just sitting here at my desk. Either it's extremely sensitive and
there are tiny pressure fluctuations going on all the time in what seems
to be still air, or it just isn't accurate enough to measure altitude to
that degree of accuracy. I suspect the latter.

I have just tried the same test with mine. Calibrated to 535 feet. In
the last couple of minutes it had been as low as 536 and as high as
541so I recalibrated it again (540>535) and the range has again
gradually drifted back, this time to 536-540. I haven't been watching
the arrow for much of the time but currently it is firmly stuck at down.

--
Roger Chapman so far this year 69 summits
New - 47 (Marilyns 13, Sweats 5, Outlying Fells 35)
Repeats - 22( Marilyns 7, Sweats 16, Wainwrights 16)

Roger said:

I have just tried the same test with mine. Calibrated to 535 feet. In
the last couple of minutes it had been as low as 536 and as high as
541so I recalibrated it again (540>535) and the range has again
gradually drifted back, this time to 536-540.

If you've got auto-cal on then that may just be due to the GPS altitude
being slightly over when you did the test. GPS altitude has cycles of
wobbling, both short jumpy wobbles and longer cycles, that's probably
why you need at least half an hour for auto-cal to settle down.

Paul
--
http://www.wilderness-wales.co.uk
http://www.wildwales.fsnet.co.uk