On Radio-Wave Propagation In: Forest Environments
On Radio-Wave Propagation In: Forest Environments
On Radio-Wave Propagation In: Forest Environments
VI. CONCLUSIONS
A practical technique is developed forthe analysis of
scattering by a thin wirewithfinite length and imperfect
conductivity. The method is based on an integral equation
for the current induced in the wire by an incident wave. This
integral equation is enforced at discrete points on the axis
of the wire to generate a system of linear algebraic equations
for thecoefficients in aseries expansion for the currentfunction.
Numerical results are presented for the backscatter echo
area of copper, platinum, and bismuth wires at the broadside aspect. These results show good agreement with experimental data and variational calculations which have been
published for wire lengths up to 0.6X. In addition, graphs
are included to show the current distributions onthese wires
at the second resonance, the phase-angle and echo-area
patterns for oblique incidence, and the broadside echo-area
curves for perfectly conducting and copperwires withlengths
up to 3.54X. Bistatic scattering data can readily be calculated with this technique. The calculations are rapid and
REFERENCES
P I R. G. Kouyoumjian, The calculation of the echo areas of perfectly conducting objects by the variational method, Ph.D. disserta-
I. INTRODUCTION
T HAS LONG been recognized that communication by
means of radio waves in forest environments is hampered by transmission losseswhich aresubstantially
higher than those occurring in the absence of vegetation.
Extensive experimental investigations111-[31of propagation
conditions in forests haverevealed that the transmission
losses may be characterized as follows.
>
.
I
a07
n2 =
u1
- j -= e
- j6OulXo
(1)
WE0
where
denotes the average relative permittivity and u1
indicates the average conductivity of the forest medium; eo
and X. are the absolute permittivity and the wavelength in
air (vacuum) of a wave with frequencyf. A time dependence
ejwtis assumed and MKS units are implied unless otherwise
specified.
1 Xore added i n proo) A similar treatment was subsequently carried
out by J. R. Wait in Asymptotic Theory for Dipole Radiation in the
Presence of a Lossy Slab Lying on a Conducting Half-space, IEEE
Trans. Anteruzas urd Propagation, vol. AP-15, pp. 645-648, September
1967.
808
NOVEMBER 1967
< H.
(3)
GROllND
The wavelength X. must be sufficiently large if the representation of the forest in terms of a uniform, continuous
medium is valid. Noting that the average separation between
trees is1-5 meters and that the intervening space is usually
Wed with foliage and other vegetation, an upper frequency
off = 100 MHz (i.e., a minimum wavelength ofho= 3 meters)
seems reasonable. The lower frequency is restricted to 1
MHz due to theoretical considerations mentioned in Section I11 and discussed in the Appendix. These considerations also prescribe a minimum range of observation p such
that, in general, p> 1 km. However, it is shown in the Appendix that p may be considerably smaller than 1 km at the
higher frequencies. For practicalreasons,the
discussion
will also be restricted to p< 100 km since forests are not
expected to preserve sufficient uniformity at longer ranges.
Theforestparameters el and ul turnout to be rather
critical and, unfortunately, only a very limited amount of
data concerning their actual values are available. Certain
theoretical considerations predictI7l -[li1 that el = 1.1 - 1.2
while u1 is of the orderof 1 P mho/m. To ensure an exhaustive converage for all possible circumstances, theranges
considered here are
1.01 5
5 1.5, a.nd
u1 2 10-j mho/m.
FIELD
E(F' = 3011
.fd
___ + f r
e-1::
p
)
rr
(4)
SO9
I \
*IR
Ionospnere
If the forest medium was lossless, any additional diffraction fields would be negligible in comparison to the above
geometric-optical contributions. In the present dissipative
case, however, it turns out that
a diffraction component
provides a dominant portion of the field over a large range
of the parameters involved.
The component in question is the lateralwave which may
be described by the quasi-optical ray trajectory TABR in
Fig. 2. This wave corresponds to the radiation emitted a t
the critical angle of total reflection 0, given by
sin 0,
( 10)
=11.
nl
EL sin 4 cos y,
(13)
where3
where
s=2h-zo-z
(9)
__
60 I1 e- jJ:o (p+n-l
EL = ___.
n3- 1
P?
a)
(14)
2 Note that n2#1 since PZ is restricted to values spelled out in Section 11.
810
One may easily verifythat the first and second terms in (11)
through (13) correspond to the horizontal andvertical components of the current element, respectively. The 'v sign
indicates that the aforementioned results were obtained via
an asymptotic evaluation which is, however, very accurate
at large distances p. The dependence of the lateral wave on
P - ~ , as compared to a geometric optical variation of p-l,
reflects the fact that the lateral wave is only a quasi-optical
contribution. Its more pronounced geometric attenuation is
due to the continuous energy leakage across the interface
along the lateral portion AB of its path. Although the general behavior of this wave was already found by Moore and
Blair['61 in their studies of radiation from sources immersed
in highly-dissipative media, they restricted themselves to the
case In 1>>1 which is not applicable here.
This completes the listof major contributions to the field
and any other diffraction componentsare ignored since
they are of lower order.[12;
one may replace thesec Bi terms in (8) with unity and obtain
ears,
with
a)?
27r
Im ( d n 2- 1 - n): (1'7)
Ao
and CY,
is therefore a relative attenuation factor which indicates the difference in exponential decay between the lateral
and the sky waves. Numerical calculationsshow that a,<0.5
dBjm for all of the pertinent ranges of cl,el, and ha. Hence,
S is close to unity, except at the largestvalues of the separation parameter s. If S=l (i.e., the case s>>l is excluded), the
lateral and sky contributions become equal at a distance
IV. COMPARISON
OF THE LATERAL
AND SKYWAVES
The attenuation produced by the forest medium affects
the various waves differently. In particular, it is noted from
(4) that the path of the forest geometric-optical contribution occurs entirely within the (lossy) forest medium and
therefore ECP) contains exponential terms which (due to n
being complex) producea decayover an extended range
r e r z p . By contrast,thelateraland
sky waves travel
mostly within the (lossless) air region. Examination of (8)
and (14) shows that the decay of these waves occurs only
over a much shorter range s<<p. It is therefore evident that
the forest geometric-optical contribution is negligibly small
compared to the other waves except possibly at very short
distances. Thus, even for small dissipation (u1= 10-5 mho/m)
the forest wave is down by at least 60 dB compared to the
lateral wave for all p> 1 km and f> 1 MHz.
One is therefore leftto compare the lateralwave with the
sky wave. The latter is available only at frequencies below
the maximum usable frequency (MUF), which is equal to
the critical frequency in the present case of nearly vertical
incidence. This establishes an upper limit of about 10 MHz,
above which the lateral wave clearly predominates. Another
important point is the choice between vertical or horizontal
polarization. Since H>_100 km, the angle Bi is smaller than
25.6" for p< 100 km. Hence, thehorizontalpolarization
is definitely preferable for exciting the skywavesince
sin &<cos ei in (5) through (8). For the lateral wave, it was
shown by Staiman and Tamir[l51that an optimum-inclination angle Y~ exists, but the improvement obtained by operatingthedipole
at
ratherthanhorizontally
(y=O), is
important only when
and/or Ep(L)are
larger
than
E+(L),i.e., if In2- 1 > 1. In the present case, I$- 1 I is of
the order of unity or less, so that Em(L)
is usually of the order
of
or larger, while Ep(L)is smaller. It is thereforeappropriate to compare E,(L) with E,(s).
Noting that
1 < see Bi
1.10'7,
(15)
where S is obtained from (17). One may now introduce approximation (2) and obtain:
A . Nighttime Comiitiorzs
Nighttime conditions prove to be the simpler case since
the variation of Ai is then relatively small and the ionospheric
reflection is produced by a single (I?)layer only. Depending
on geography, the sunspot cycle and other factors, one obt a i n ~ [ ' ~ ] - -the
[ ~ ~following
]
two extreme conditions.
811
t
(MHz1
P " q
Fig. 4. Domains of lateral- and sky-wave contributions during nighttime regime. The units for the abscissa are: p in k m , ~1 in mmho/m.
Favorable
Conditions
Absorption coefficient A;
Effective height H
Maximum usable frequency (MUF)
0.7
100
6
Unfavorable
Conditions
0.5
200 km
2 MHz
B. DaJrtime Conditions
It is rather difficult to obtain a clear-cut comparison between the sky and lateral waves during the daytime due to
exceedingly large variations of the parameters Ai, the maximum usable frequency (MUF), and the ionospheric height
H . In analogy to the favorable and unfavorable conditions
discussed above for the nighttime regime: two extreme situations are considered which cover all but rarely encountered
variations of the "quiet"ionosphere.['g]
The ionospheric losses for the twoextreme conditions are
given in Fig. 5 where the curves with the lowest and highest
losses correspond to themost favorable of unfavorable situations, respectively. The maximum usable frequencies are
Fig. 6. Domains of lateral- and sky-wave contributions during daytime regime. The units for the abscissa are: p in km, u1 in mmho/m.
7.5
104
+ 2.5 x I O - y
(31)
812
v. DISCUSSION
AND COMPARISON WITH EXPERIMENTAL
DATA
The resultsof Section IV show that, except for a restricted B. The Vegetation Factor
The presence of vegetation affects EL via a factor
range wherein the sky wave may be dominant, the field in
a forest model of the type considered here is primarily in
F , = I n2- 1 I eLs,
(24)
the form of a lateral wave, as given in (1 1) through (14).
In particular, the intensity of this field is characterized by since the refractive index n is a function of the forest parameters pl and e.Due to the separation distances appearing in
(14) which yields a magnitude
the exponential term, theheight of the vegetation aboce both
60 II
the receiving and transmitting antennas is rather important.
.-e-aLs *
IELI =
(2%
It is interesting to observe that the lateral-wave model imIn2-11
p2
813
i.25
I
I
I
1
I
I
I
1
I
wave is probably due to the effect of the ground proximity.
Experlmentalpolnts:
Since the latter effect is not accounted for in the present
model, one is justified in considering only those data which
refer to antennas that are sufficientlyelevated above the
ground and the correspondence between the theoretical and
measured variation is then very good.
As a further corroboration of the previous experimental
verification, it is noted that CYL in (23) is easily determined if
both and
u1 are known. In the case of the region where
the propagation measurements[2] were taken, it was found
at the higher frequencies (50-100 MHz) that 1.05<e1<1.15
and 0.05<u1<O. 15 mmho/m (Hagn and
AssumI
I
ing that these quantities are frequency independent, C Y L as
3
1
I
.7 1
2
4
7 10
20
40
70 100
200
L 0
obtained from (1) and (23) possesses values whichlie
f (MHz)anywhere between the two extreme curves shown in Fig. 8.
Fig. 8. Theoretical and experimental values of the lateralActual measured values of C Y Lare indicated in the figure by
wave attenuation U L versus frequency$
the letters V and H which denote vertical- and horizontalpolarization results, respectively, and the indicated points
represent smoothed-out values obtained from a large num- hand, represent theoretical predictions which are obtainedby
ber of similar measurements.[21 It is clearly seen that all of a set of completely different experimental measurements,
the measured attenuation points lie well within the range of namely those for the refractive index n. The lateral-wave
values predicted by the lateral-wave model. At frequencies model enters the picture only as the tool employed for corless than 12 MHz,the height gain was too small to be relating the two sets of data. The consistent quantitative
efficiently detected. Note also that the values of el = 1.1 and agreement obtained over the pertinent frequency range thus
ul= 10-* mho/m yield a curve, shown dashed in Fig. 8, emphasizes the plausibility of the lateral-wave model.
Another important aspect emphasized by Fig. 8 is that
which gives a very good fit up to 100 MHz. These values of
c1 and u1 are a good average of the various data obtained the height gain is critically dependent on the particular values
of both u1 and el, particularly at the higher frequencies.
by Hagn and Parker.
It is interesting to note that the comparison illustrated in Thus,the two extreme cases shown by the solid curves
Fig. 8 essentially relates two apparently independent sets yield attenuations which differ by 0.8 dB:m at 100 MHz.
of results: on the one hand, the experimental points H and Hence, an accurate determination of both u1 and el is imV were obtained by means of field-strength measurements portant for determining the total path loss in the case of
tall trees and low antennas (s large).
at various antenna heights; the solid curves, on the other
I
814
P,
I EL l2Ao2 .~ 1
Re (n)
320&
(25)
Pt = 8079
(E)2
Re (n).
(27)
f (MHz1
Fig. 9. Theoretical and experimental values of the basic path loss LbO
versus frequency f at a distance p = 1 km. The theoretically predicted values liewithin the shaded region.
E. Depolarization Effects
It is well known that the signal received by an antenna
where Lbodenotes the value of La at s=O. At the higher fre- located in a forestis considerably depolarized with reference
to the field radiated by a transmitter situated inside or outquencies, the frequency dependence of n may be neglected
so that LbOvaries asf4. Hence, the basic path loss Lbincreases side the forest. While this aspect could be discussed by a
strongly with frequency even if both antennas are close to statisticalapproach which accounts for terrainirregularthe tree tops. However, this situation may change if one of ities,[.il the depolarizing effect is actually predicted by the
the antennas (or both) are sufficiently aboce the vegetation lateral-wave model without necessitating any statistical concanopy. In that case, propagation occurs mostly by refrac- siderations.
To emphasize this feature, it is recalled that the received
tion or line of sight rather thanvia a lateralwave, but such a
field is not given by EL alone but this quantity needs to be
situation is not presently discussed.
To check theaforementioned results, relation (29) is multiplied by various factors to yield the actual field complotted in Fig. 9 for the same range of values of n that were ponents E,, E,,, and E+ of (1 1) through (13). Thus, if one
already considered in Fig. 8. For that range (1.05 <e1 < 1.I 5 uses a vertical antenna (-y=~/2),a horizontal component
and 0.05 < ~ < 0 . 1 5mmho/m), (29) predicts that LbOis given E, will be present in addition to the expected vertical comby points which lie within the region shown shaded in the ponent E,, and an analogous situation holds for a horizontal
figure. Experimental results are alsoshownand they are dipole (7 =0).
815
to the
ionosphere
ionosphere
KI 7 I"
TJK
TCDE
TJFGH
fromthe
""Y\
AIR
XYZQR
Tertiary
roys
-TCi
-Tcdi
-TAfgi
IBR
WWR
lYZBR
Fig. 10. Ray complex for the sky and lateral waves in the presence of
a reflecting ground plane. (a) Sky wave. (b) Lateral wave.
816
VII. CONCLUSION
Thestudy of radio waves in forest environments presented here has shown that propagation in the frequency
range of 1-100 MHz occurs primarily by means of a lateral
wave. The field at distances p > 1 km may then be adequately
described by a quasi-optical ray whose trajectory proceeds
along and just above the tree-top contour. By virtue of this
propagation mechanism, the lateral wave seems to be tied
to the tree-top line and is therefore capable of following its
contour even if the vegetation extends over a terrain with
certain amounts of curvature due to hills or other obstructions. A sky wave due t o reflection from the ionosophere
may also be present, but it will be important only at large
distances and low frequencies (roughly, p> 10 km and
f< 10 MHz).
The lateral waveis a diffracted field component which
varies with distances asand
therefore attenuates more
rapidly in comparison with the p-l dependence of a geometric-optical contribution.Ontheotherhand,thelateral
wave travels in the Iossless air medium, and therefore it does
not undergo the exponential decay which affects a geometricoptical ray that travels through the lossy vegetation. In addition to providing a simple explanation of the observed p-2
variation of the fieldsin forest environments, thelateral
wave variety affords a straightforward and consistent interpretation of many other propagation aspects such as the
height-gain effect, the vegetation factor, depolarization
effects, andthe path-loss variation. These features were
found to be strongly dependent on the equivalent refractive
index n describing the vegetation medium and the separation
distance s consisting of the combined depths of the (transmitting and receiving) antennas below the tree-tops. In terms
of the field associated with the lateral wave, one finds that
the height-gain effect, the vegetation factor, and the basic
path loss are all expressible by means of simple relations
which show that transmission losses increase with both s
and In Hence, the separation s of the antennaswith respect
to the tree-top contour turns out to be of major importance,
whereas the antenna height above ground plays only a minor
role in most cases. The increase of the transmission losses
with the refractive index n is in agreement with physical intuition since denser forests correspond to larger values of nl .
All of the theoretical considerations and results were compared quantitatively with available experimental data and
good overall agreement was obtained. This agreement is
found to be remarkable in view of the fact that the model
chosen to characterize the forest is a particularly simple one.
The utility of this model is enhanced by the fact that, in addition to its convenient simplicity, it also takes appropriate
account of the intrinsic properties of the forest medium in
I.
1967
such a manner as to provide a consistent physical description, as well as an excellent first-order estimate, of most of
the previously observed propagation characteristics.
APPENDIX
EVALUATION
OF NUMERICAL
DISTANCE^
IZI
which,in
yields
viewof
.4
lo<
300 meters,
GO
d p < - X lW3 X 9 X lo4 = 1720met,ers.
(33)
iT
817
grand expressing the rigorous field solution. The appropriate condition is then given (Brekhovskikh, eq. 2 2 . 1 2 9 by
I v2konp
(34)
-(1
,+ +
I
REFERENCES
- 1
\/,/122
dnz - 1
v-
/1
I I
? 1.