What's the most accurate elevation tool on the net (freebie)

Unless you are dealing with a surveyed peak, I would assume all the elevation data is derived from USGS NEDs (national elevation dataset). For the most part, they are only 1/3 arc second. I'm speculating that some of these programs are reporting data from the NEDs as if they are section of flat areas on the 1/3 arc second grid, while others are interpreting the elevation using a weighting scheme.

Have you considered running SPLAT! ? It uses 1/3 arc second data. It can predict line of sight. Generally I find I do better than the SPLAT! prediction, so if SPLAT! says no, the answer is maybe, but if SPLAT! says yes, you have line of sight. The program has hooks for predicting signal strength, models knife edge diffraction, etc.

Similar software is

For pure line of sight, there is GRASS. However GRASS has a very steep learning curve.

What takes maybe 5 minutes on SPAT can take 2 days on GRASS. I don't know if the results are that much better.

For the bay area, I suspect SPLAT! is fine. I think it's shortcomings are in areas where there is really rugged terrain that probably isn't modeled well.

Youre asking wrong questions. First define a surface with height ZERO as a reference. There are about 100 different definitions alone for that. Distance from Earth center, median sea level at Novosibirsk, or a San Francisco? Water isn't level, it follows gravitation. And so on.

w.

I'd use the info from another licensee's application, (on the same tower). With the FCC you also must specify which survey you are using. All towers over 200'(AIR) have their own entry in a tower database.

On Sat, 03 Mar 2012 00:10:32 -0800, miso wrote:

I just installed the Splat (Surface Path Length And Terrain) RF analysis application (version 1.3.0-1) from the default Ubuntu Software Center.

-

$ sudo apt-get install splat (dpkg -s splat)

It's apparently a command-line tool which needs me to download the database so it may take a while to figure out.

Googling for a "splat!" tutorial, I find it's also a photoshop hack so it's actually hard to figure out how to use it on the fly.

Q: Do you have a working example or two for how to use it from the command line?

[quote] $ splat --help

*** ERROR: No transmitter site(s) specified!

$ splat

--==[ SPLAT! v1.3.0 Available Options... ]==--

-t txsite(s).qth (max of 4 with -c, max of 30 with -L) -r rxsite.qth -c plot coverage of TX(s) with an RX antenna at X feet/meters AGL -L plot path loss map of TX based on an RX at X feet/meters AGL -s filename(s) of city/site file(s) to import (5 max) -b filename(s) of cartographic boundary file(s) to import (5 max) -p filename of terrain profile graph to plot -e filename of terrain elevation graph to plot -h filename of terrain height graph to plot -H filename of normalized terrain height graph to plot -l filename of path loss graph to plot -o filename of topographic map to generate (.ppm) -u filename of user-defined terrain file to import -d sdf file directory path (overrides path in ~/.splat_path file) -m earth radius multiplier -n do not plot LOS paths in .ppm maps -N do not produce unnecessary site or obstruction reports -f frequency for Fresnel zone calculation (MHz) -R modify default range for -c or -L (miles/kilometers) -db threshold beyond which contours will not be displayed -nf do not plot Fresnel zones in height plots -fz Fresnel zone clearance percentage (default = 60) -gc ground clutter height (feet/meters) -ngs display greyscale topography as white in .ppm files -erp override ERP in .lrp file (Watts) -ano name of alphanumeric output file -ani name of alphanumeric input file -udt name of user defined terrain input file -kml generate Google Earth (.kml) compatible output -geo generate an Xastir .geo georeference file (with .ppm output) -dbm plot signal power level contours rather than field strength -gpsav preserve gnuplot temporary working files after SPLAT! execution -metric employ metric rather than imperial units for all user I/O

If that flew by too fast, consider piping the output through 'less':

splat | less

Type 'man splat', or see the documentation for more details.

This compilation of SPLAT! supports analysis over a region of

8 square degrees of terrain. [/quote]

GPS < 3 meter error

SPLAT! isn't a photoshop hack, but you can use gimp (or I suppose photoshop) to hack with the png files it creates. I take the png files and chop them up so they are acceptable to google earth, then make an overlay out of them. Splat can do it, but unless things have changed, it makes a kml file that can be too large for GE to accept. GE wants tiled imagery. I haven't run it in maybe 18 months, so I can't really say much without setting it up again.

I saw a GPS mentioned. I never found GPS elevation to be too accurate. I have a barometer in my GPS. I never used it, so I can't vouch for it's accuracy. But you can enter in the pressure from the airport, or calibrate it to a known reference (known altitude for a position).

One idea would be to check the USGS monuments and find the closest reference.

Note that sometimes these markers are on private property. Other times they are in the middle of the street! it pays to look at the reference on google earth before trying to use it. My Garmin gps60cs was good to

4ft, basically one lsb.

You could get a nearby reference, call the barometer, then quickly move to your spot before the pressure changes. Or you might get lucky and find there is a marker there already.

Topo maps have lines of constant altitude, usually on 20ft contours. You could interpolate from the map.

If you really need accurate data, just pay for a survey. I would guess something that simple is under a grand. I've paid for land surveys and they are a few grand, but the altitude at one point is pretty simple. Potentially the civil engineering firm might have topo data on file that is not available to the general public. For instance, I paid for a satellite topo to be done on some property. The civil engineering firm has it on file and I'm sure it gets peeked at by other people.

If in the US, I would assume that the LAT/LONG uses either WGS84, NAD27, or NAD83 datums. That reduces the number of available options.

Unfortunately, the USGS is still hanging onto NAD27, while most mapping programs and displays are on WGS84.

In the People's Republic of Santa Cruz, the error is about 20 meters east-west, and about 1 meter north-south. I forgot which way. The problem becomes really bad when trying to locate a mountain top.

20-200 meters of horizontal error can easily move a position from the peak, to somewhere on the slope, resulting in large altitude errors.

SRTM and SRTM2 are another oddity. They were created from the space shuttle, using a radar altimeter. Depending on the whether it's looking at buildings or trees, there's no really good way to determine of the indicated altitude is the top of a 100ft redwood tree, the top of a 10 meter high building, or at ground level.

So, my list of rhetorical questions are:

1. What is the OP trying to accomplish? If for an FCC license HAAT calculation, almost any reasonable guess will suffice.

If for doing Radio-Mobile coverage contours, you'll need to use the built in mapping tool to find the peak or exact location on the mountain top. The choice of application depends on what one is attempting to accomplish.

1. What level of accuracy is really required? The original position of 37.337408N -121.644073W is specified to 1 millionth of a degree, or about 0.09 meters.

It would be interesting to know where this highly accurate number came from. Better GPS receivers, without DGPS, can do 3 meters accuracy.

1. What datum are you using? I suggest WGS84. If the lookup tool offers a choice of datums, pick one and stay with it.
2. Are you interested in ground level, building rooftop level, or tree top level?

The neighbors and are discussing setting up a neighborhood mesh network and we need to ascertain, beforehand, where to place masts and which direction to point them in our mountainous neighborhood (Skyline & Summit area).

For us, probably any reasonable answer would suffice - but why not pick the most accurate for starters is what we're thinking.

A few feet would probably work just fine for the neighborhood. We each have acres of land, but the terrain is so rough that only a few spots for antennas would be useful. That's why we want to choose them ahead of time.

We didn't want to put our actual location on the net, so, we picked an arbitrary set of numbers from one of the elevation calculators just as an example. But we're in the roughly 37,-122 range.

WGS84.

We have some numbers in NAD83 from the various WISP providers but they drive us crazy since we have to imperfectly convert them to WGS84 to keep our numbers consistent.

All three because we want to site a dozen or more antennas which need to have clear line of sight over rooftops and trees by at least the first Fresnel zone.

Interesting quotes from that helpful reference (with my comments in parenthesis).

"In the Continental United States the difference between WGS 84 and NAD

27 can be as much as 200 meters." (I wonder how they handle the constant creep which occurs out here near the San Andreas fault line).

"Every map that shows a geographic coordinate system such as UTM or Latitude and Longitude with any precision will also list the datum used on the map." (I'd change "will" to 'should' based on my experience the past two weeks on the web)

"The Global Positioning System uses an earth centered datum called the World Geodetic System 1984 or WGS 84." (That's what I prefer.)

"For all practical purposes there is no difference between WGS 84 and NAD

83." (Good to know.)

"On a USGS topographic map ... The datum will always be NAD 27... A dashed cross in the SW and NE corners of the map gives a visual indication of the difference between the two datums." (This is good to know.)

"If you are engaged in a mission that requires more [than several hundred meters] precision, then your datums should match." (Since we're siting antennas on private hilly land, we probably want two or three meters accuracy in position and a half-meter to a meter in elevation accuracy so our datums must match.)

What I mean by that is that it's a right-slip fault, and it moves by centimeters to inches each year (sometimes in feet to yards, both in elevation and in position) ... but ... how do they know if the west side moved north or if the east side moved south?

I wonder what they use for their frame of reference since it depends on which side of the fault you're on if you want to say the west moved north or that the east moved south.

Ok. Go thee unto:

Follow the destructions at:

For maps, download the SRTM3 maps for your area from:

Do not bother with DEM, SRTM1, or other maps. Do NOT unzip the maps. My directory shows about 600MBytes of SRTM3 data for everything for the SF Bay and Monterey Bay areas. You can set Radio-Mobile to automatically download a map if needed, but it's easier to just download the maps ahead of time.

Follow a simple example such as:

to get started. There are also numerous tutorials on the web. Note that the program uses the concept of "networks" which will be key to modeling a mesh. Locate your nodes, use realistic values, and build a model. This part is a PITA and requires considerable time and effort. Draw the (optical) coverage areas for each node, and the line of sight:

You're going to have a big problem in the Santa Cruz Mountains call trees. These are cellulose and water obstructions that may or may not appear at the correct altitude on the elevation profiles. 2.4GHz will NOT penetrate foliage, especially when wet. You'll need to manually adjust your path profiles for the tree line. If the trees are inside the Fresnel zone, you'll have losses.

If you have problems, ask here, or preferably the Yahoo Radio-Mobile group at:

I have a really bad attitude about mesh networks. Bug me if you want to hear the full rant. For a sample, see the dismal performance of an early mesh network (MIT Roofnet -> Meraki).

"Surprisingly, the performance over a two hop route is less than 1/2 that of one hop routes, implying routes tend to interfere with themselves." Also:

covers the main problems. Do you really want a phone call at 2AM from a neighbor asking if the network is down?

This is an interesting antenna HAAT (Height Above Average Terrain) program ... but I'm not quite sure what use it is because it gives a 360 degree average height ... but most 2.4Ghz antennas I'm dealing with are directional.

You enter the latitude, longitude, & height of the antenna, and then it tells you, for example, for 360 degrees, the average antenna height above ground for 2 to 10 miles along each radial, the result of which can go negative.

The output is a text file. It's interesting, but, without graphics, I'm not sure how to use the results properly when just going point to point.

I wrote that before you disclosed what you were trying to accomplish. Had your intent to obtain accurate altitude readings been for the purpose of applying for an FCC station license, you would have needed the HAAT calculations to estimate coverage area. For building you mesh network, you don't need HAAT calculations.

The first paragraph should explain what HAAT means.

Hint: You can always take a table of number and create a graph or graphic. Going the other direction is not so easy. Most propagation and antenna design software generates an output table (text file), from which a graphic is later generated.

I've used Radio Mobile and SPLAT!. I never got a warm and fuzzy with Radio Mobile. Of course, it is a bit more complicated to use SPLAT!.

One obvious advantage to SPLAT! is it can analyze very large areas. Not all that useful in the case of this wifi setup, but very useful in sigint.

Oops. I meant the SRTM1 maps.

I've used both. Radio-Mobile has a very steep learning curve. Important functions are buried deep into obscure menus, useless trivia is scattered all over the menus, there's no logical sequence of operation, and many of the terms require expertise in cartography. Debugging errors is tricky as important items, such as the performance characteristics of the radios, are scattered over a half dozen menu pages. I find myself constantly referring to my cheat sheet in order to get anything done. However, I haven't found anything else that even comes close to what it does.

Splat is somewhat easier to use, but as you note, is designed to display repeater coverage. It's less useful for close in coverage, or showing coverage details, as in mountainous or urban jungle terrain.

Both programs put considerable effort into implementing complex terrain models. For 2.4 and 5.7Ghz, optical line of sight is close enough.

The first problem I'm having is locating a Linux (Ubuntu) Radio Mobile download ...

Is that suppose to be some kind of thanks for doing your research? In the future, if you need assistance, get it from someone else.

RM mostly runs under Wine:

The problems listed are not fatal as you can download the SRTM maps manually, and can simply export the result as a Google Earth overlay to obtain street map detail.

My recollection of Radio Mobile is you need to crank down the minimum angle that it sweeps to get any accuracy. Like I said, I prefer SPLAT! for the accuracy. Even so, it is only as good as the NED. However, if SPLAT! says you can see it, then you can see it. I thought Radio Mobile was simple to run, at least for one transmitter at a time. Far easier than SPLAT, which requires compilation parameters to set the array size. Radio Mobile, at least when I read it, was stuck at 3600x3600. If you exceed that array, and note it uses a 1/3 arc second grid, the program interpolates.

The grid is 10 meters on a size for 1/3 arc second. That means you can't "see" more than 36km. Plenty for wifi, not so good for repeaters or even photography.

I generally do two runs with SPLAT. First I check the altitude when the radio is to be located. If it doesn't match the topo map, I add the difference in altitude to the transmitter height. Then run it again.

I have a 90 mile path to analyze, so I guess I'll see what these programs can do lately. But if Radio Mobile is stuck at 3600 pixels, that is a show stopper.