[TowerTalk] antennas in the trees

[Jim Lux]

Interestingly, in the most recent of IEEE Transactions on Antennas and
Propagation which arrived today, there's a paper:

"Wave Mode and Path Characteristics in a Four-Layered Anisotropic Forest
Environment" by Le-Wei Li, Chee-Keong Lee, Tat-Soon Yeo, Mook-Seng Leong, pp
2445-2455, V52, No. 9, Sept 2004

They talk about modeling a forest as multiple slabs of anisotropic (i.e. not
the same for vertical and horizontally polarized waves) media. One slab for
trunks, one slab for Canopy. (the other layers are the earth and the air
above the forest).

Of particular interest is that they cite some references on HF propagation,
which some folks might be able to get a hold of (I'll try and get copies at
work.. the IEEE ones should be easy):

1) T. Tamir, "On radio-wave propagation in forest environments", IEEE Trans
Ant Prop, Vol AP15, pp806-817, Nov 67
2) D.L Sachs, P.J. Wyatt, "A conducting-slab model for electromagnetic
propagation within a jungle medium", Radio Science, v3, #2, pp125-135, Feb
68
3) D. Dence, T.Tamir, "Radio loss of lateral waves in forest environment",
Radio Science, v4, #4, pp307-318, Apr 69
4) T. Tamir "Radio wave propagation along mixed paths in forest
environments", IEEE Trans Ant Prop, AP25, pp471-477, Jul 77

#3 includes some empirical measurements at HF frequencies, judging from the
graphs in the new paper.

Naturally, they're focussing more on higher frequency, since that's where
the hot interest is these days, but they do talk about lower frequencies.
They claim that below 200 MHz, one can just model the forest as a basically
isotropic medium (perhaps in layers).

They also present some interesting modeled data, compared against
experimental measurements (which come from Dence&Tamir)

vertical dipoles and measurement points at 3m and 10m off the ground.
1km separation
Forest is 1m of trunk
9m of canopy
epsilon canopy = 1.12, sigma = 0.12 mS/m
epsilon trunk = 1.03, sigma=0.03 mS/m
epsilon earth = 20, sigma = 10 mS/m

The excess loss over the 1km path above the free space loss of
32.4+20log10(Dkm)+20log10(FMHz) is
1 MHz    32.4 + 14 dB
10 MHz    52.4 + 33 dB
30 MHz    62 + 38 dB

Looks like for lower frequencies that the loss (E field) is proportional to
frequency (i.e.power goes as 20dB/decade)

I'm still working through the math to see whether the loss follows a linear
function of distance (the usual absorption thing, on top of the inverse
square spreading)

I believe the forest model is for fairly dense coniferous forest. (they cite
Cavalcante, et al., Radio Sci, v19. #5, pp691-695, Sep 83)

The propagation model includes bouncing up and down between layers, etc.