>> In other words, some GPS receivers (or firmware) are junk.
>That certainly has been my experience each time I bought a new phone (use
>sat nav app) or less frequently GPS computer seems to be more accurate, I
>certainly remember older units getting very confused as to which road/trail
>I was on thinking I was on a parallel road or so on.
Non-smartphones, dumb phones or feature phones handle GPS very
differently than Smartphones. In a smartphone, the GPS chipset
produces the location, speed, altitude, etc directly to the CPU, which
then displays it on a map or a screen full of numbers. In a dumb
phone, the satellite delays are sent to an internet service provider,
which calculates the location, speed, altitude, etc and sends the
results back to the dumb phone. That saves battery power and
eliminates the cost of a processor, but only works if you want a fixed
location, not a track displayed on a map.
Meanwhile, there were big improvements in GPS receiver design, which
improved sensitivity, accuracy, battery life, cold start, etc. The
big improvement was high-sensitivity GPS in about 2009.
<
https://www.furuno.com/en/gnss/technical/tec_high>
Prior to this, most GPS receivers had problems in highly reflective
environments (urban canyons) and in weak signal areas (forests). With
high-sensitivity GPS, it was possible to obtain usable signals in
areas where GPS was previously useless.
There are a bunch of other improvements, such as additional
constellations (GLONASS, Galileo, DBS, etc), better augmentation (WAAS
via Inmarsat), additional signals (L5, L1C) and better post processing
in the phones. I can explain these and others but not right now.
>Clearly some could be software ie locking on to the road/trail etc when
>navigating though the GPS traces seem tighter.
That was a feature, not a bug. The problem was that the FCC demanded
that for E911 emergency service, the accuracy of a GPS location had to
be far better than what was available at the time. Basically, they
wanted to know which land of the street the caller was located. If in
a building, they wanted to know the floor number. Most cellular
service providers failed miserably and had to resort to marginal
tricks. One of these was to deliver a position to the PSAP (public
safety answering point) that was rounded off to the location of the
nearest road. The logic was that since about 80% of the 911 emergency
calls were from vehicles, which presumably were on a road, assuming
that the call was coming from a location on a road was tolerable.
Never mind people calling from wilderness location and areas away from
roads.
After a few screwups, the PSAP's quietly asked the cellular vendors
involved (mostly AT&T) to disarm this feature. The other vendors were
more into improving the accuracy of the GPS system and ran into a
different problem. The road maps were horribly inaccurate. They were
good enough for driving from point A to point B, but not good enough
for GPS location. Besides the ongoing confusion between different
datums, different agencies used different mapping methods, resulting
in different map errors. For example, 911 service used the telephone
company property maps, which were not very well maintained. I was
marginally involved in trying to reconcile map errors in Santa Cruz
county. I could overlay various maps on top of each other and find
wide variations between maps. Of course, none of the map "owners"
wanted to fix their maps and insisted that their maps were perfect.
At this time, most of the major errors have been fixed, but the
millions of minor discrepancies remain. LIDAR mapping and satellite
imaging has done wonders for fixing the maps. However, it will
probably be a few more years before you're able to look through the
viewfinder of a phone, eyeglasses, binoculars or computer and obtain
an accurate map overlay using augmented reality.
<
https://www.google.com/search?q=augmented+reality&tbm=isch>
Programs, such as Strava, improve things so that they appear better on
a map. Instead of a series of jagged "connect the dots" lines, these
programs smooth the track so that it looks like a very accurate and
smooth line. I don't believe that they snap the track to the nearest
road because there are so many runners, hikers, swimmers and cyclists
that do not follow the roads. Unless you were using a smartphone GPS
for generating your riding track, you shouldn't have seen it snap to
the nearest road.
Accurate elevation is a different horror story. In general, the sports
GPS computers use the map elevation (via Garmin Connect) or calibrated
barometric pressure. GPS elevation data is horribly inaccurate.
"How Are Elevation Readings Calculated for My Activity in Garmin
Connect?"
<
https://support.garmin.com/en-US/?faq=dRY70Lc6yv2oY3eam1ZWxA>
"Barometric Altimeter Accuracy of Outdoor Products"
<
https://support.garmin.com/en-US/?faq=WlvNrOungC28xGtwB7hLY5>
However, Garmin will record GPS elevation data if the GPS receiver
provides it.