View Single Post
Old September 26th 16, 05:47 PM posted to sci.geo.satellite-nav
Terry Pinnell[_3_]
external usenet poster
Posts: 9
Default Puzzling change in trackpoint rate

Alan Browne wrote:

On 2016-09-25 03:16, Terry Pinnell wrote:
Alan Browne wrote:

Perhaps it updates less often when there is poor quality in the position
solution. Where do you have the phone when you're recording such
tracks? (pocket?). When the points were more numerous did you have it
in a better position? Was it hilly or urban canyons? (The later should
have good cell aiding, however).

Even sat geometry could have an impact at the time that you're recording
on the position quality.

Just a guess.

Thanks Alan, and apologies for not acknowledging sooner.

There has been no change of those kinds at my end. Still carrying the
iPhone either in shirt pocket or (thin) trouser pocket. Same types of
terrain (mainly English countryside).

How about the satellite end? Could there have been some change there?

Not enough to matter. IAC the position solution on your phone is
composed of:

GPS (and SBAS P-R and corrections if in view)
GLONASS (and SBAS corrections if in view (not sure about P-R))
Cell Tower ranging
WiFi reception (if any - probably not out in the fields...)

So the various changes to the satellites (in/out of service, etc.) won't
show in user land unless you're into plate tectonics - and even then...)


More important (as the very high rate seems here to stay) am I right
that this significantly increases estimates of gross ascent and
descent? Elevation profiles chart now look very much noisier. Should I
just increase the smoothing factor until they 'sort of look right'?!

Depends on how your ascent estimator is designed. If it's simply based
on delta-alt from the GPS and/or the altimeter in the iPhone it should
be pretty good. Smoothing 'til it looks right should be okay - 10
second time constant, I'd guess.

There is an app you can download called SensorLog which will record all
(most?) of the sensors in the iPhone to a file. (or o/p to an IP
address via WiFi - not practical for you...). Makes a CSV file that can
be easily processed or loaded into a spreadsheet for plotting and

On my tracks I don't care much about elevation (though maybe I should)
and when plotting a trail I usually have 6 or more recordings of it,
walking in each direction. In the woods, in winter, the variance is
easily 5 - 10 m. (No SBAS received either as at 45 N or so, in the
woods, the WAAS sats are too low; the 1 EGNOS sat I get is only when on
an open southeast facing hill)

Opinions vary widely on this topic, including the 'fractal' view that
you can regard just about *any* estimate as 'accurate'. (Distance as

"Fractal" - "Estimate" - "Accurate".

If you're looking for 5 m accuracy and the estimates have 10 metre
uncertainty then the fractal nature of the data (if so) doesn't matter a
whit, does it?

More useful to get a feel for the error by making repeated recordings on
a given trail or route. You won't get the truth, but you'll get a range.

With many recordings you can plot the middle (eyeball it or write some
code - the later is much harder to do than say) and get a feel for the
errors in different situations. Open field, good antenna position,
receiving SBAS ... expect a couple m most of the time; in dense woods,
north side of the mountain (northern hemi), expect 20 - 50 m error.

well as cumulative ascent/descent.) But I haven't given up on trying
to come up with reasonably realistic estimates, so that helpful
comparisons can be made, both about walks done and planned.

Any position estimate has errors. When I plot trails, I do it in the
winter when the leaves are down and I put the receiver under my tuque so
that there is minimal blocking of the satellites. This tightens up the
error considerably - esp. if SBAS is received.

If you have your iPhone in your shirt or trouser pocket, you are denying
the receivers a huge part of the sky as the signal will not go through
your body - the errors will be larger than if you can get that iPhone up
above your head where it can see as much sky as possible, all of the
time. I'd bet that would calm down your elevation plot a bit too.

Note that smartphone sat receivers are excessively sensitive. When you
put the receiver in a pocket, it still gets the signal at a very
suppressed level. Noisy - therefore less accurate. Further, it would
be very prone to multipath signals bouncing off rocks and perhaps trees.
That of course makes for more error.

So, if accuracy is important, get a hat, put the iPhone in a ziplock bag
and stuff it in the hat and wear it.

Many thanks Alan, much appreciate that comprehensive explanation.

I'm fairly satisfied with my distance measurements (after editing out
any serious errors by visual inspection). It's those gross ascent
descent estimates that vex me!

Walks (including holidays with my wife) are sometimes led, sometimes
self-led. The former often include a briefing beforehand from the
leader along the lines of "I'll be leading the higher of the three
walks tomorrow, which will be 10 miles, 1500 ft of gross ascent and
1750 ft of gross descent." Leaving aside the obvious question as to
the source of his/her estimates, they do at least give us a rough idea
of expected difficulty based on past experience. And for our own
planned self-led walks I like to prepare my own estimates.

BTW, my walks never get repeated exactly. When they're close I do try
to make some comparisons.

Do you think that estimates based on digital elevation models, like
the one that Google Earth uses, or the UK's LIDAR model with 2 m
are more likely to give a more realistic GA/GD than recorded GPS data?
(BTW, no altimeter on the iPhone.)

I like the hat idea, although I suspect I'd quickly lose my iPhone!
Used to have my ancient Garmin strapped to my rucksack, so maybe I'll
try that again.

Terry, East Grinstead, UK