[Qgis-user] wishing for accurate lattitude/longitude from a cell phone

[Garth Fletcher]

I got involved in a project to locate the actual boundary monuments 
which delimit our town.  The USGS 7.5' topographic maps appeared to be 
in error at some locations by several hundred feet.  About 30 monument 
locations were involved.

We are in rural NH which is mostly wooded, which results in a lot of 
satellite signal attenuation, i.e., fewer receivable satellites.

I did a first effort using a Garmin eTrex 20 (~ $200) which received 
both the US GPS satellites and the Russian GLONASS satellites. Receiving 
both is important in our high attenuation environment because it 
significantly increases the number of receivable satellites.

Those measurements supported my suspicions about topo errors, but had 
error bounds in the tens of meters - not accurate enough for my purpose.

Next I used a Bad Elf Surveyor (~ $600 + Mac iPad) to record 30 minutes 
of data into RINEX files which were then sent to CSRS-PPP for 
post-processing.  This somewhat reduced the error bounds, but they were 
still ~10 meters wide, even for some 60 minute recordings.


Finally I used an iGage iGS3 receiver (~ $2400) to record US GPS and 
GLONASS satellites for at least 30 minutes each (up to 1 hour under 
heavy foliage) into RINEX files sent for post-processing to CSRS-PPP. 
This approach finally realized the ±1 meter with 95% probability I needed.

A graph of the error ellipses for the 30 monuments as predicted by 
CSRS-PPP post-processing, shown on a 1 foot grid, can be seen here:
<http://www.Mason-NH.org/Specials/Mon_Ellipses_Final_1_ft_grid.jpg>
Note that almost all are within ± 2 feet.


GPS satellites broadcast on two frequencies, L1 and L5. A key difference 
is that the hand-held Garmin units and the Bad Elf Surveyor only use the 
L1 frequency whereas the iGS3 is a dual frequency receiver (L1 and L5).

One large source of errors is the variable signal propagation delays in 
the ionosphere, which have predictable differences between the L1 and L5 
frequencies. Recording both signals allows a better estimation of, and 
correction for, the ionospheric delays.

A word about post-processing.

I use the Canadian Geodetic Survey's CSRS-PPP processing because they 
accept data from both US GPS and Russian GLONASS satellites whereas the 
US Geodetic Survey's OPUS only accepts US GPS satellite data.  In our 
heavily wooded environment the ability to use both constellations of 
satellites provides a crucial boost in performance.

Post-processing services continuously record L1/L5 signals from hundreds 
of fixed sites.  This allows them to accurately model the time-changing 
errors in GPS signals, primarily ionospheric delays but also errors in 
the satellite orbits and their clocks.

When RINEX data is submitted, the service can look at its 
contemporaneous data from fixed receivers to model the errors at the 
time and location of the RINEX recording and thereby apply corrections.

CSRS-PPP offers 3 levels of correction: ultra-rapid (90 minutes), rapid 
(24 hour) and final (2 weeks).  Which is applied depends on how long you 
wait between recording the RINEX file and submitting it for post-processing.

The ultra-rapid will use the 90 minutes of data preceding the recording; 
the rapid will use 24 hours of data, and the final will use 2 weeks of 
data in calculating the corrections.  More data gives better models and 
corrections. In my experience the differences between 24 hour and 2 week 
corrections are minor compared to my ±1 m target accuracy.


A fixed + rover approach is the gold standard for accuracy.  This 
technique uses 2 receivers - one fixed for the duration and the other 
"roving" around to the various locations to be measured.  As long as the 
two stations are fairly close they will be affected identically by the 
same GPS errors which can then be cancelled out.  The cancelling can be 
done in real-time if the fixed station broadcasts its data to the rover, 
or it can be done by post-processing. Of course using two receivers 
doubles (or more) the equipment cost.--
Garth Fletcher