On Tue, Aug 19, 2008 at 10:41 AM, Kelly Martin-AB9RF <[EMAIL PROTECTED]> wrote:
>> Hmm... it's an interesting idea. Currently the splat analysis is raster
>> data, but I think there is some way to have it calculate vector data. Might
>> be easier to run the raster analysis through a line tracer. I'd love to see
>> this done with radar data too.
>
> Getting vector data out of splat would be best done by generating a
> ".plo" file, which is just a huge litany of points (they're actually
> generated along each azimuth line going outward) with the
> corresponding path loss. Building contour lines from that data ought
> not to be terribly hard. .plo files are freaking large, though (4
> megs for a 100-mile radius calculation, which is actually inadequate
> for some repeaters sited up on top of a mountain).
Herein lies a problem that needs to be addressed rather early in this
type of discussion. There's no way that we can entertain sending 4
megs of data out over the air to describe a repeater coverage area at
1200 baud. Obviously that's not needed to draw a coverage limit
contour. We would only need the location along the azimuth line that
corresponds to the desired signal level. So, how many azimuth lines
are required to describe the coverage contour? The more lines,
obviously the greater detail can be conveyed, but that is done at the
expense of airtime used to send that data.
The APRS spec has an area object descriptor, which allows us to send
points to describe an area. How many points surrounding a repeater
would be needed to adequately give a reasonable facsimile of the
repeater coverage?
Jon NG0E does a pretty good job of showing propagation openings by
using a 12 radial descriptor, and then using bezier curves to create a
coverage plot that isn't simply a 12 sided polygon.
Here's how he describes calculating and drawing the propagation plots.
**
The propagation maps on this site are derived as follows:
* Only stations which have received 5 or more distinct
station-locations in the past hour are analyzed.
* For each of these stations, the average and standard deviation
of reception distances is calculated for 12 equally spaced 30 degree
slices; the maximum propagation distance in each direction is computed
as three standard deviations over the average distance.
* The computed maximum propagation distances for each of the12
slices are converted back to latitude and longitude for mapping.
* The area formed by connecting these twelve spokes is smoothed
using bezier curves.
**
You can see this at work here:
http://www.mountainlake.k12.mn.us/ham/aprs/path.cgi?map=na
Obviously these maps show approximate propagation openings, and not
areas where you would be guaranteed to find openings. Similarly one
should not expect the repeater coverage described by an APRS object to
be 100% accurate either.
What I would suggest is to do something similar for repeater coverage,
take that Splat! output, and reduce that 4 meg .plo file down to 12
lat/long points spaced out every 30 degrees, and output that as an
area descriptor. That would give immediate capabilities to any client
that can handle area objects. I would also suggest adding an addendum
to the descriptor that would indicate that bezier smoothing should be
used, to make the dodecagon into a "blob". Bob might object, but
Xastir could lead the way, and being an addendum to the descriptor,
other area object renderers would not be hindered by it.
Another 2 bits worth of conceptualization free of charge!
James
VE6SRV
___
Xastir mailing list
Xastir@xastir.org
http://lists.xastir.org/cgi-bin/mailman/listinfo/xastir
