On Tue, Jan 18, 2000 at 05:38:50PM +1030, Conn Copas wrote:
Tom
Some things to check with r.los:
1. Use g.region to ensure that your current region extent corresponds with that
of your elevation map. You can also coarsen up the resolution initially to make
things run faster whilst you are tweaking parameters.2. The "max_dist" parameter is fairly crucial, although it doesn't sound like
that is your problem.3. As a last resort, make sure that the version of Grass which created the
elevation map is compatible with that running r.los (thinking here of, eg,
floating point changes to the data structures).>From memory, you are correct that Grass assumes a flat earth and hence is of
limited value at the larger scales in which you are interested. Arc doesn't make
that assumption. Also, neither system will warn one about doing dubious things
like computing a LOS within a lat-long co-ordinate system.
Thanks for your comments. I have found another LOS program that seems to
produce reasonable results, and is geared for vhf/uhf progation by including
a 'k-factor' to flatten the earth by 1.3. Reads USGS DMA-formatted DEMs,
or it's own format:
http://www.mindspring.com/~otterson/cs485/
If I can find the time, I'll try a port to use GRASS rasters.
--
Tom Poindexter
tpoindex@nyx.net
http://www.nyx.net/~tpoindex/