VE3NCQ

22 Comments »

1. 

   Thanks. Just what I’m after. I’ve been using “Path Profile” (by M.J. Willis) to get me Path losses but is a bit buggy and irritating to use. I’ve just downloaded Splat and have run it up for the first time. Looks good.

   Are there GUI options available for generating kml files? I know Splat can do this. Also any options for using ASTER files (“Path Profile” claimed to be able to use these but I couldn’t get it to work)?

   Cheers
   Carlos

   Comment by Carlos — December 1, 2010 @ 05:38

2. 

   Had a further play with the software. Some comments:
   1. For the life of me I cannot generate graphs (whether png, ps or gif) for point to point graph analysis. I checked and I do have gnuplot (wgnuplot).
   2. Selecting Open then cancel produces an error message, and Tx and Rx file entries are wiped from the GUI. You’d want the default behaviour to be to do nothing?
   3. Fractions of seconds in Tx/Rx positions are ignored (no rounding done)
   4. GUI is resizeable. Perhaps shouldn’t be?
   5. I did a point-to-point terrain analysis and the ppm showed a broken green line although there is definitely LOS between my Tx and Rx
   6. What is gmon.out?

   BTW. I found the kml option, but haven’t tried it yet.

   Comment by Carlos — December 1, 2010 @ 07:11

3. 

   On the GUI, you can check the box to get a kml file if the analysis type supports it. This option disables the legend colour key, otherwise it would also show on the map. The kml file shows a transparent overlay when opened in Google Earth.

   Right now, SPLAT doesn’t handle the ASTER format, but this is a very interesting suggestion. The SRTM one arc-second data will already work in SPLAT, but is limited to the United States due to NIMA restrictions. Perhaps ASTER data can get around this restriction. The srtm2sdf converter utility is currently used to make a suitable SDF terrain file for SPLAT. I don’t know if John Magliacane has done any work on the ASTER format, but I will try getting some files to see what I can figure out, if anything. In theory, it should be possible to modify the converter to make an acceptable SDF file from ASTER. SPLAT won’t care where the data came from as long as the SDF files are valid.

   More information about ASTER is available from the USGS.

   Thanks for the comment Carlos, and it certainly would be worthwhile to work ASTER compatibility into the program.

   Comment by VE3NCQ — December 1, 2010 @ 07:49

4. 

   1. The graph problem can be traced to the system PATH environment variable. If you have wgnuplot installed, SPLAT needs to know where it is. It gets this from the PATH environment variable. Mine is installed in the root directory of c:\ in case SPLAT doesn’t like spaces in the path name. Once wgnuplot is on your PATH, it should work.

   At a command prompt, type echo %PATH% and see what’s there. Mine looks like this:

   C:\>echo %PATH%

   C:\WINDOWS\system32;C:\WINDOWS;C:\WINDOWS\System32\Wbem;C:\Program Files\Symantec\pcAnywhere\;C:\Program Files\Samsung\Samsung PC Studio 3\;C:\Program Files\Common Files\GTK\2.0\bin;C:\Program Files\QuickTime\QTSystem\;C:\PROGRA~1\HAMRAD~1\gnuplot\bin
   

   If gnuplot isn’t there, type: c:\> path = %PATH%;c:\PATH-TO-GNUPLOT

   2. The ability to open a saved file in the GUI is new, and I didn’t bugcheck every combination. I forgot to trap the cancel error and gracefully exit. I’ll do a bugfix release to correct that.

   3. SPLAT recognizes decimal coordinates, such as 74.6864. It also accepts DMS format, like 74 41 11.0. Are the fractions being dropped in the site report? I looked at the code, and the site report only uses the first 4 decimal places. They aren’t dropped during the calculations. I’ll see what happens if it’s increased to 6 places.

   4. You’re right, the GUI should not be resizable. I never tried that.

   5. This has been reported from time to time. There could be several causes. If you make a kml file, and overlay it on Google Earth, can you see anything it might be picking up as an obstruction? In one case I looked at, a tree line was the obstruction, and the attenuation would have been less than predicted (depending on the frequency). You can create a file with terrain overrides in case there are anomalies in the SDF files.

   6. As far as I know, gmon.out has something to do with debugging. I thought that debugging symbols had been turned off when I compiled the releases. It seems I left the profiler on, which Dev-C++ includes to help locate excessive processor usage. The file isn’t written by SPLAT itself, and can be safely ignored for now. Next time I compile the binaries I’ll make sure the profiler is off as intended.

   Thanks for the feedback, and I’ll put those corrections in the GUI and credit you with the discovery.

   Comment by VE3NCQ — December 1, 2010 @ 09:00

5. 

   I have SPLAT.
   I have WGNUPLOT.
   The WGNUPLOT directory is in the PATH statement.
   When I run any point-to-point graph the console output says that the plot has been written, but no file is actually written anywhere.
   Any further ideas?

   Comment by Bill Weeks — December 1, 2010 @ 15:00

6. 

   At a command prompt, type pgnuplot. Does the Gnuplot terminal come up? If not, there is a path issue. Make sure that the path ends in “gnuplot\bin” with no trailing slash.

   When SPLAT runs a graphing analysis, it saves the data to a file called “splat.gp” in the working directory. It then invokes gnuplot with the system command “pgnuplot splat.gp”. If gnuplot opens successfully, splat.gp is deleted, but it is retained if gnuplot fails. The thing to look for here is whether or not splat.gp is sticking around. If it’s there, cd into that directory with a command prompt, and type “pgnuplot splat.gp” which should bring up the graph in gnuplot. If not, it’s a PATH environment issue.

   By looking for what happens to splat.gp, you should be able to figure out where things are going wrong. On the GUI, there is a checkbox under the “Point-to-Point Options” that will prevent splat.gp from being deleted.

   Comment by VE3NCQ — December 1, 2010 @ 16:11

7. 

   Many thanks for that Austin. I can now get my graph. It would be nice if a plot can be displayed as well, either in the GUI or a separate window.

   Other comments:
   1. For the path-to-path graph option, it would be useful to display the calculated path loss somewhere on the GUI

   2. Put units against text (e.g. on the Longley-Rice analysis option, display km against “Radius to plot”, m against “Receive antenna height” etc.)

   3. The software doesn’t like spaces in the qth (and presumably the lrp) files. Probably a Windoze thing. Any chance of your software placing quotes around any inputted filename to get around this?

   4. A nice have: Batch process a number of Rx sites with one Tx site to get path losses (to put this in context I’m examining the impact of a windfarm on some Radars here in the UK).

   Other than that, a very useful tool!

   P.S. I notice that the weather over where you are in Chatham is similar to what we have in the UK, except snow here is major news!!!!!

   Comment by Carlos — December 2, 2010 @ 05:29

8. 

   That Gnuplot problem seems fairly consistent for a lot of people. Perhaps I could add some tests to my GUI that check for its presence on the machine and in the PATH. It’s GNU so maybe I could offer a bundled version that’s pre-configured.

   As for a plot with graph, I think SPLAT does a different kind of analysis with that function that doesn’t create a plot. You’d have to run Longley-Rice as well. On the Point-to-Point Terrain analysis, there is an option to force terrain maps to white. That is to allow you to superimpose the plotted line over a Longley-Rice Path Loss Analysis using layers in PhotoShop or GIMP. Just make the white transparent, or multiply the layers.

   1. Path loss is saved to a file. I could make it appear on the console output, but that would require changes to all the SPLAT executables. Maybe the GUI could have an option to open that path loss file in Notepad on completion.

   2. I will see if there’s a way to add units. The problem as I recall is that each console line is added as the command line is parsed. In most cases, the units are not known yet because the -metric switch hasn’t been encountered, and may not in fact exist at all where Imperial units are desired. When -metric is encountered, the console will say “-metric switch processed. Analysis done in metric units”. The best I can do is maybe have SPLAT parse for the -metric switch first, then it will know which units are in use when the other switches are encountered. This is also a matter for the SPLAT executables.

   3. I thought about putting quotes around file names, or escaping the spaces like we do in Linux. It will need some experimenting, because the GUI only passes – and SPLAT only accepts – some but not all of the file names. The rest are calculated. If I’m not mistaken, SPLAT just changes the extension, so adding quotes would be a logical step. I’ll put that in the GUI and see what happens.

   4. Yikes! Here’s what we, in theory, could do. My GUI saves the command line which is really what invokes SPLAT. Take that, and copy it to a batch file. Copy and paste as many sites as you need, changing your RX sites as needed, and the output file names. Save that file as a DOS .bat file, and run it from the command line. That would probably be the fastest way. You’d have to enter all that info one at a time in the GUI anyway, but using more mouse clicks.

   You probably have your parameters all figured out already, but my approach would be to create a user-defined terrain file that adds each of the turbine sites. This will plot the effects, keeping in mind that the blades are turning so some of the shadowed area will be disturbed as opposed to blocked, and energy will be scattered off in different directions which is then reflected back as artifacts. There is a way to produce a text output file in SPLAT that can be edited and fed back in with obstructions added. This is off by default in the GUI, because the files are huge – in the order of GigaBytes for some runs. Add the -ano switch to get one and -ani to read it back.

   I’m assuming that your Radars are for weather? The wind turbines absolutely will show up as ground clutter if they are close enough. I think 13 km from the Radar is the closest they can be. We have many wind farms here in Chatham-Kent. I have noticed that when propagation is enhanced, during inversions or tropo ducting events, our turbines show up as strong echoes on the NWS Radar 100 km away. When that happens, hams can be fairly certain of good openings on VHF and UHF. You could probably model this by making changes to the .lrp parameters, such as decreasing the N units. In this case, you not only have a shadow area, but also ground clutter returns and Radar energy scatter effects.

   The N bending constant used by SPLAT is 301, which coincides with the standard 4/3 earth radius model (k) of 1.333 and the standard atmosphere (1013.25 hPa at 15° C). SPLAT only performs a 2D analysis. To get an idea of the 3D effects on pattern, you might model the antenna in 4NEC2 with the turbines added as passive far-field radiators. You may have to model them proportionately taller to compensate for the 4/3 earth radius which NEC doesn’t consider.

   As for our weather, it’s been a milder autumn. This year was the latest we’ve ever had a killing frost, and we’ve only been down to -2° C. I still have a tree in the backyard with coloured leaves. Any similarities with our weather will soon end, as we will be into snow and cold season very shortly (antenna season).

   Comment by VE3NCQ — December 2, 2010 @ 08:03

9. 

   Hi Austin. Thanks for the comments.

   Currently I’m interested in ATC and AD radars (operating at about 3GHz) and the effect of windfarms on detection of targets in the vicinity of the farm. I have a separate SW for modelling radar systems effects, e.g. antenna patterns, pulse compression, signal processing, and separate SW for predicting radar cross-section of each component of a turbine. Second-order effects that you mention, multipath, I don’t model but could, although I do include ground and sea clutter (K-distributed). I haven’t considered ducting, but we have SW here (at work!) which allows us to define whatever refractive profile (with height) we want and give us path losses through ray-tracing, but I’m generally interested in the “standard” atmosphere.

   Thanks again

   Comment by Carlos — December 2, 2010 @ 17:26

10. 

    Thanks, Austin, I’m partway there with gnuplot, anyway.
    “pgnuplot” at any command prompt starts the program.
    splat.gp disappears. But no graph is written anywhwere.
    But “pgnuplot splat.gp” from a command prompt, if I check the preserve gnuplot box, gets me the graph.
    Progress is good.
    Bill

    Comment by Bill Weeks — December 3, 2010 @ 21:04

11. 

    I have conducted some research into the ASTER format, and downloaded the files for my area. On the surface, it should be trivial to convert the GeoTiff format to SPLAT sdf files. It’s just a matter of jumping past the header offset and converting the 16-bit elevation data to the long integer format used in SPLAT, starting at the bottom left coordinates, eliminating the top and right rows of pixels.

    The problem is that ASTER data tends to have numerous voids and anomalies. We are surrounded by Great Lakes here, and the surface is very irregular, with deep banding. This would translate into obstructions that don’t exist. The bands make strips of higher or lower elevation that skew the propagation analysis. While mostly visible over water, these do extend over the land in some areas.

    That said, the remaining data seems to be very detailed and good quality. I think it could be useful in some situations, especially where the 1 arc-second STRM data isn’t available. The solution may lie in digital imaging. This is my theory:

    First, we’ll need the 3 arc-second STRM data from the .hgt files converted to GeoTiff. This won’t be terribly complicated, as the files cover 1 degree squares and we’re not using the tags. In PhotoShop, these can be resampled up from 1201×1201 to 3601×3601 pixels. This will form the background layer. Then we paste the ASTER GeoTiff file on top in a new layer. (GIMP can’t handle 16-bit TIFF files.)

    Now, it’s simply a matter of using the erase tool to remove the voids, which show as black spots. Use a brush that’s an appropriate size, with soft edges. Erase the garbled data over water. Any defect that’s erased will reveal the (3 arc-second) STRM data from the background layer. Once everything is fixed up, flatten the image and save it as a new TIFF file.

    This should produce a 16-bit TIFF with all the anomalies fixed up. It’s a bit tedious, but far more accurate than any automatic algorithm. The master files will only need to be done once. Then we can convert it to sdf files for use in SPLAT.

    One added advantage is that data can be corrected or added visually. For example, we have a garbage dump that’s grown 100 feet in elevation since the data was collected, and it happens to block the path of a proposed link. This can be drawn on the image at the appropriate place. The foreground shade of grey used on the paintbrush will correspond to a particular height.

    With this theory in mind, I’ll try writing some code to make the necessary conversions, and see how it works out in practice. Then I’ll post it on a new page with the programs, and add instructions on getting the ASTER files.

    Comment by VE3NCQ — December 13, 2010 @ 20:34

12. 

    Hi

    I would like to know if there is any technical documentation for Splat!. In other terms, What models were used, why, how, etc…

    Thank you in advance

    sara

    Comment by sara — January 14, 2011 @ 05:52

13. 

    SPLAT uses the Longley-Rice Irregular Terrain propagation model. There is some good documentation in the docs folder of all the SPLAT distributions, and the manual is also on the KD2BD site.

    Several authors have posted articles on SPLAT. These are some that I know about:

    SPLAT Home Page
    Longley-Rice Irregular Terrain Model
    John McMellen, KC0FLR
    W5GFE Splat Web Interface
    ARRL-Surfin’: Splat That Path Online
    Analyzing DTS Coverage With Longley-Rice, SPLAT
    SPLAT! Resource Page – Describes Sid Shumate’s fix to ITM.cpp
    Improvements in Propagation Models by Doug Lung

    Comment by VE3NCQ — January 18, 2011 @ 10:16

14. 

    I tried to extract the archive splat 1.3.1 but it gives me corrupt archive.

    Comment by Andrea — February 9, 2011 @ 16:13

15. 

    Try saving it to a different location. The MD5 checksum of the file is 4efeeb3e099fbe625bfb55f4398813d6. If that doesn’t match, the file has been corrupted along the way.

    MD5(SPLAT-1.3.1.zip)= 4efeeb3e099fbe625bfb55f4398813d6

    Comment by VE3NCQ — February 9, 2011 @ 17:31

16. 

    The MD5 checksum splat 1.3.1 ebfb6ad5070bbeed542652ca6602f471

    Comment by Andrea — February 10, 2011 @ 16:53

17. 

    Hi. Very useful tool. I had a bit of an issue getting it to work to start with, when i entered the details hit splat, nothing happened. Found the text file it had generated and it was looking for splat-130-SD16.exe. I renamed a copy of the mx16 exe to this and it worked.
    Thanks

    Comment by Rob — March 15, 2011 @ 07:32

18. 

    Could someone help me with error warnings please?

    When doing LR Path Loss Analysis, I have got an error warning “Longley-Rice model error number: 3, Warning: A combination of parameters is out of range. Results are probably invalid.”

    What does this error mean? Below are my settings, I cannot find anything wrong with these settings:
    Longley-Rice path calculation parameters used in this analysis:

    Earth’s Dielectric Constant: 15.000
    Earth’s Conductivity: 0.005 Siemens/meter
    Atmospheric Bending Constant (N-units): 301.000 ppm
    Frequency: 884.800 MHz
    Radio Climate: 5 (Continental Temperate)
    Polarization: 1 (Vertical)
    Fraction of Situations: 50.0%
    Fraction of Time: 90.0%
    Transmitter ERP: 3.7 Watts (+35.66 dBm)
    Transmitter EIRP: 6.0 Watts (+37.80 dBm)

    Please any help would be appreciated.

    Goanna
    PS. any pointers to slat! discussion groups would also be appreciated.

    Comment by goannae — April 5, 2011 @ 19:48

19. 

    I would like to try splat for NB.

    Please send me your program so I can convert the Maritime e00 files

    Thanks!

    Comment by Graeme — September 25, 2011 @ 14:11

20. 

    I am going through a big move right now to W4, and my network is being disassembled while I pack furiously under deadline. Unfortunately I’ll be offline for a bit without access to my programs. Can you wait until I get back up? I believe you are looking for the utility I wrote to convert e00 to SPLAT dat files. I think I did that quick and dirty in Visual Basic but I might port it to PHP. In the meantime, try SPLAT with the US data, which ought to get you some lines. Do you know VB? I could maybe post the source code, which requires a couple of easy modifications to grab the correct input files. Apparently I never did a proper GUI for it, and I can’t compile it the way it is since I only ran it in the VB development environment.

    Comment by VE3NCQ — September 25, 2011 @ 14:43

21. 

    Thanks for providing the Windows version of Splat! Do you have plans to make the 1.40 version available for Windows?

    73,

    Brandon Pierce
    KC0BTP

    Comment by Brandon Pierce — December 2, 2011 @ 15:06

22. 

    I will look at it once I get settled in Virginia. All of my development computers are in storage right now in Canada so I can’t access anything. The GUI shouldn’t need to be updated unless additional command line options were added to the latest version. This would probably be a project for the new year.
    73,
    Austin
    VE3NCQ/AK4OG

    Comment by VE3NCQ — December 2, 2011 @ 17:01

RSS feed for comments on this post. | TrackBack URI

Leave a comment