[Fwd: [TowerTalk] GPS receivers]

[Jim Lux]

At 10:38 AM 9/20/2004 -0400, Pete Smith wrote:
>At 08:52 AM 9/20/2004, Jim Brown wrote:
>>My friends who claim to be educated in matters of GPS say that
>>because of the manner in which the system is modulated for security
>
>I think this is a bit confusing.  Before 2003, there was something in 
>place called Selective Availability, which limited civilian GPS systems to 
>tens of meters accuracy, and appears to be what your friends are telling 
>you about.  That practice was discontinued then, and civilian systems, 
>unaugmented, should have about the same accuracy as anyone else's, with a 
>given set of satellite data.
>
>The following, from the FAA's web site, may be of general interest:

<snip an excellent summary from the FAA... by the way, WAAS isn't available 
everywhere, particularly outside the US)


>To get back on the original topic, I don't think anyone associated with 
>HFTA would argue that positioning errors on the order of tens of meters 
>would have a significant effect on the accuracy of HFTA predictions, 
>except in some fairly special cases where significant near-field obstacles 
>are quite close to the antenna system.

Tens of meters is potentially many wavelengths.

Position precision on that order makes a difference if you're near the top 
of a hill, of course.

The documentation supplied with HFTA doesn't provide any guidance on the 
accuracy of the model used internally, as well.

I think that your best bet is to compare the terrain model extracted by 
MicroDEM (or whatever other tool you use) with your actual situation and 
place the antenna location in an appropriate place on that model, 
independent of what the actual lat/lon happens to be.  "National Map 
Accuracy Standards" only require positional accuracy of 1/50th inch for 
features and elevations, etc.  On a 1:24000 map (the usual 7.5 minute 
topo), that's on the order of 10-12 meters.

The SRTM data (currently the most accurate topography data available over 
wide areas) is distributed with +/-20m error (90% circular) in the absolute 
sense and 15m in a relative sense. If you're matching up to GPS, it's the 
absolute error that you'd worry about.  The relative vertical accuracy 
(which is what you'd use for HFTA) is spec'd at +/-6m (90% vertical error).

The older USGS DEM files are distributed in a variety of horizontal grids 
and accuracies.  I have a set of CD-ROMS with the older 1 degree DTED data, 
sampled at 3 arc second posts, but the underlying data is basically drawn 
from the 1:250,000 topo map data, and has comparable accuracy.  Comparing 
feature positions from 1:24000 maps against topography from that DEM data 
shows occasional errors of as much as 100 meters.

Lets also not forget whether your terrain database derives from a "visual 
top of the trees", "bald earth", or "trained estimate by USGS 
cartographer".  This is particularly relevant in areas like the Pacific 
Northwest where 100 ft tall forests cover much of the surface, and can 
obscure variations of 10-20m.

There are more accurate databases available, for some areas. For instance, 
Thousand Oaks CA has a survey done by flying a lidar over the city with a 
precision GPS, and fairly decent ground truth/validation survey.


Upshot... Use anyold GPS or one of the web mapping programs to get close. 
Extract your local terrain database.  Look at it to see if it looks 
reasonable, and site your modeled antenna according to the terrain, not the 
numbers.  Run the model several times with variations in the profiles, to 
get a feel for the sensitivity of the results to smallish changes in the 
terrain.

Jim, W6RMK