RADIO SCIENCE, Vol. 3 (New Series), No.

l, January 1968

Exact Ray Calculations in a Quasi .. Parabolic

Ionosphere With No Magnetic Field

Thomas A. Croft

Radioscience Laboratory, Stanford University, Stanford, Calif. 94305, U.S.A.

and

Harry Hoogasian
Raytheon Company, Spencer Laboratories, Burlington, Mass. 01803, U.S.A.
(Received August 9, 1967)

The equation of the parabolic ionosphere model is modified in a manner which introduces very
slight changes in the electron-density distribution. Using this new "quasi-parabolic" (or QP) model,
it is possible to derive exact expressions describing radio-ray trajectories without introducing the
approximations which are needed when dealing with the parabolic ionosphere, although magnetoionic
effects are not included. The consequent accuracy of computed rays is useful as an aid in evaluating
the accuracy of more versatile ray-calculation methods. Also, the exact ray expressions can be of
direct !lse in calculating slight differences between rays which are nearly identical, for in this process,
small inaccuracies can become magnified.

1. Introduction

The parabolic model of the ionosphere was introduced by Forsterling and Lassen (1931), who
showed that radio-ray trajectories in this model could be readily calculated if a few reasonable
approximations were made. Subsequent investigators made refinements (e.g., Appleton and
Beynon, 1940; Rawer, 1948) but the method has remained approximate to this day. The tractability
of the ray equations has made the parabolic ionosphere very popular as a means for analysis of
ionospheric radio propagation.
As implied by the name, the model is defined by the equation of a parabola in electron density
versus height. In the notation which is most convenient for our purposes, the parabolic layer is

ro < r< rm+Ym (1)

elsewhere

where Ne =electron density, having maximum value N m

r= radial distance from earth center (height+ earth radius)
rm=value of r where Ne=Nm (hm+earth radius)
r0 = value of r at the layer base= r m-y m
Ym =layer semithickness.

We wish to point out that a very slight modification to the parabolic model permits the deriva-
tion of exact equations for ray-path parameters. This modified parabolic ionosphere will be termed
the "quasi-parabolic" or, more simply, the "QP" model. The exactness of the resulting ray equa-
tions probably provides a negligible benefit in many of the more approximate applications of the·
69
70 Croft and Hoogasian

parabolic layer. However, if one wishes to investigate small differences between rays which are
nearly identical, the use of the QP model will eliminate any suspicion that differences are intro-
duced by the parabolic layer approximations. Also, a very important application for the QP model
is die service it can provide as a primary standard of accuracy for testing more versatile ray-tracing
methods.
The QP model was described in the literature by de Voogt (1953) in a form which bears little
superficial similarity to the presentation here but which nevertheless was fundamentally the same.
This original work was not widely used, apparently because of the difficulty in evaluating the ex·
pressions for the ray parameters using manual techniques. Probably for this reason, de Voogt
did not publish the expressions. We believe that the QP ionospheric model will become increas-
ingly useful because of the widespread availability of digital computers; the ray-parameter expres-
sions described here may readily be evaluated by any computer and even by many of the small
desk calculators now coming into use. Thus, in a sense, the justification for using the approxima-
tions of a parabolic ionosphere is disappearing with the advent of efficient computing machinery.
The QP layer is defined by

-{Nm[l-(T-Tm)
Ne- Ym
2
(!:!!.)
T
2
]; To< T< Tm(-!!--)
To Ym .
(2)
- 0 elsewhere

The significant change from (l) is the addition of a multiplier, (T0/T) 2. Since To= Tin the region of
interest, the electron density versus height is nearly the same as that of a parabola having the
same Tm and Ym· The semithickness is still Ym, the geocentric height of the maximum is still Tm, and
the geocentric base height is still To. The difference between this QP and a parabolic layer is the
precise variation of electron density between the peak and the bottom of the layers. Figure l shows
overlapping plots of parabolic and QP models computed ~nd plotted by machine in order to preserve
unbiased accuracy. A fine plotting pen has been used to permit the difference between layers to
be seen, and even with this precaution the distinction is readily visible only when the semi thickness
is 200 km. This shows that the QP model is indeed very close toits parabolic predecessor. The
separation between the models is more pronounced ah!?ve the layer maximum, hut this is not sh9wn
on figure I because it cannot affect the rays which will be considered here.

2. Calculation of Ray Parameters

It is convenient to define F= f/fc, the ratio of operating frequency to critical frequency. If
n
we neglect the geomagnetic field and collisions, then = 80.6 Nm in MKS units and the refractive
index, f.L, is defined by

(3)

When Snell's law is applied to such a spherically symmetric system, the law of refraction
assumes a special form (Bouger's rule) best ~xpressed by noting that the product Tf.L cos f3 does not
vary along a single ray. Using the notation of figure 2, it is seen that Tf.L cos f3 =To cos f3o. By con-
sidering the geometry of a differential ray segment and noting the symmetry about the ray apogee,
the application of this law leads to integral representations of the desired ray parameters:

D, the distance traversed, measured along the earth's surface

P', the group path (the signal transit time multiplied by c)
P, the phase path (the wavefront transit time multiplied by c, or the total number of wavelengths
along the ray multiplied by the free~space wavelength).
Exact Roy Calculations in o Quasi-Parabolic Ionosphere
71

Center of Earth

FIGURE L Comparison of parabolic and quasi-parabolic FIGURE 2. Ray path geometry.

models of the ionosphere.

J J r 2 cos {3o
D = 2ro
fo
llt
dfJ= 2ro
rt

r0
dr
---= 2
r tan {3
rt
°
ro rv' r 2 p, 2 - r~ cos 2 f3o
dr (4a)

(4b)

(4c)

These three equations apply to any ionosphere concentric with the earth. However, the
difficulties in evaluating the integrals in closed form limit their application to a few specific profiles
such as the QP layer. In this instance, the profile has been defined in (2) as a piecewise function
of altitude, so that (4a, b, c) must be separated into their free space and ionospheric parts. For
the ionospheric component, the expression under the radical becomes

with
72 Croft end Hoogosion

A = l - _l + (__!.!_) 2
F2 Fym

2
rbr ') - r~ cos 2 f3o-
C = ( ~.
Fym

In free space the refractive index is unity. With the foregoing substitutions, (4a, b, c) become

D=2r~cosf3o{frb dr +Jrt-:::.-:-r:;:::::::=d=r=;::===:::} (Sa)

ro rVr 2 - r~ cos 2 {3_ 0 rb rVAr 2 + Br+ C

(Sb)

P=2 {I rb
_y 2_
rdr
2 2 {3.+
rtIVA 2
2
TJL dr
+B
1
+C1,withJL 2 from(3). (Sc)
ro r r 0 cos o rb· r r

These integrals are standard forms whose solutions are given in many tables. In carrying out the
necessary algebra, it is convenient to introduce -y, the angle of the ray at the bottom of the iono-
sphere as shown in figure 2. The final results are:

D=2ro\ (-y-{3o) rocosf3oln B~-4AC l 2\ (6a)

2
\J'c 4C (sin -y + rb VC + 2VGB)
P =
I 2{ rb Sin. -y - ro Sin. {3o+ -Al[ - Tb Sin. -y - 4vA
B 1n~ r7 B -4AC
(2Arb+B+2rb vA sin-y)2
]}
2
~ r7 (6b)

P=2(-rosin{3o+!!_[ l ln
4 VA (
BZ-4AC
B ~
2
+.r;;ln
vC ( C
sz;;;;c J)·
B )
21
(6c)
4 Arb+ 2+vArb sin 'Yj 4C sin-y+----;:;;-+ \J'c
2
Some useful ancillary results for the QP layer are the following:

Ar'f + Br1 + C = 0 Equation of ray apogee height

ro
cos -y=- cos {30 Angle of ray at bottom of ionosphere
Tb
B
Tt, max=- 2A Maximum apogee height (steeper rays will penetrate the ionosphere).

3. Sample Calculations

A number of calculated rays are presented here in a compact graphical format; also, a few
rays will be described numerically to preserve accuracy. These rays are given to provide sample
answers for readers who do not wish to evaluate (6a, b, c) and also to aid in checking calculation
routines designed to evaluate the equations. The present authors have independent computer
Exact Ray CalCulations in a Quasi-Parabolic Ionosphere 73

programs which evaluateD, P', and P; and the answers agree within 1 part in 107 for distance and
group path and within 1 part in 106 for phase path. The lesser accuracy on the phase path calcula-
tion arises because there is a subtraction of two numbers which are nearly equal, thereby degrading
the accuracy of the result by about an order of magnitude.
Figure 3 shows a "reflectrix" family at intervals of 1 MHz, computed for a representative QP
ionosphere. This format has been chosen because it simultaneously shows takeoff angle, ground
distance, and virtual height. For any reflectrix point, the distance is determined by dropping a
radial onto the curved base line and noting the scale provided. The virtual height is the length of
this projection, a scale for which is given on the left axis. The takeoff angle is measured as an
angle above the optical horizon, which is shown as a separate line. In addition; the group delay
may be obtained by measuring the distance from the range-height origin to the point in question,
using the group delay scale shown at the bottom of the plot. This measure is actually the length
of the equivalent triangular ray and is based on a theorem of Breit and Tuve (1926) which is exact
only for a flat earth-ionosphere geometry. Thus, the group delay scale is approximate while the
distance, takeoff angle, and virtual height scales are accurate. This reflectrix construction was
introduced by Lejay and Lepechinsky (1950), and a more recent description of it has been given
by Croft (1967). Figure 3 was computed and plotted by accurate automatic techniques, but there
may be some slight distortion in the reproduction.
As an aid in the checkout of computation methods, table 1 gives a few ray descriptions in
numerical form. The agreement between our independent computations leads us to surmise that
these numbers are corr.ect to the last digit given but it is possible that the agreement was fortuitous.

4. Applications and Limitations of the Exact Equations

The accuracy of the technique given in this paper serves only as a mathematical convenience
and does not represent a significant increase in the realism of the calculated rays when compared
to those which may be obtained from a parabolic model. Both the parabolic and the QP ionospheres
are grossly simplified representations of the real ionosphere. In particular, the rays which make
only a shallow penetration into the parabolic or QP models are not much different from the rays
which could be obtained by simply reflecting them off an equivalent ionospheric mirror. The
geomagnetic field is ignored in both the parabolic and the QP calculations, but this is unavoidable
because there is no exact ray solution known to the authors which includes the field. The fact
that table 1 is good to six digits rather than four is an advantage for many purposes, but it should

QP Ion~s~here
Nm:ld 7m ot hm:300km
Ym:IOOkm
3000

0 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16
Approximate Group Delay, msec

FIGURE 3. Rejlectrixes showing radio propagation in a representative QP ionosphere.

From any point on a reftectrix for a selected frequency. drop a radial to the curved baseline to find ground distance of ray. The length of the radial
ia the virtual height. using the scale given on the left. From the same point draw· a line to the distance-height origin; the angle between this line and the
optical horixnn io the ray takeoff angle. F"mally, the length of the line ia (approximately) the ray group delay, using the lower scale.
74 Croft and Hoogasian

TABLE l. Computed Rays in a QP layer; Nm= 10 12 /m", hm=300 km

(=rm-ru), Ym=lOO km

Radio frequency. Takeoff angle. Ground distance, Group path. Phase path,
MHz degrees km km km

20 0 3428.874 3513.514 3500.35

4 2673.309 2760.273 2745.35
8 2161.267 2256.988 2236.08
12 1842.605 1955.747 1922.15
16 1685.174 1829.296 1769.19
20 1841.034 2065.866 1916.15

29 0 4128.680 4268.531 4210.13

I 3923.094 4063.916 4004.56
2 3752.455 3896.388 3833.98
3 3622.093 3771.912 3703.74
4 3546.615 3706.676 3628.40
5 3577.037 3756.917 3658.70

be realized that the second (or even the first) digit may be wrong if the rays are used as a substitute
for those which occur in the real ionosphere.
When these results are used to test versatile ray-tracing programs, one must still be cautious
about interpreting the significance of high accuracy. For example, almost all ray-tracing programs
ignore the fact the the earth's radius varies by 20 km depending on observer location. (What, then,
is a "non tilted ionosphere'?") Similarly, it is known that the index of refraction exceeds unity by a
significant amount in the lowest 20 km of altitude, and that the consequent bending (usually
neglected) will increase the ground range oflow-angle rays from l to 3 percent. Perhaps the greatest
source of error in all ray calculations is the uncertainty in the specific ionospheric models which
must be used.
The extreme accuracy which can be obtained by numerical ray tracing does serve a useful
purpose when comparisons must be made between two rays which are nearly identical. Such
comparison often involves the calculation of a small difference between two large numbers, and
in this circumstance the fourth, fifth, or sixth digit may be crucial.

The part of this work performed at Raytheon was supported by the Propagation Sciences
Laboratory of the Air Force Cambridge Research Laboratories under contract AF19(604)-5230.
The part conducted at Stanford was supported by the Advanced Research Projects Agency under
contract Nonr-225(64).

5. References

Appleton, E. V., and W. J. G. Beynon (1940), The application of ionospheric data to radio-communication problems: Pt. 1.
Proc. Phys. Soc. London 52, No. 202, 518,...533.
Breit, G., and M. A. Tuve (1926), A test of the existence of the conducting layer, Phys. Rev. 28, No. 3, 554-573.
Croft, T. A. (1967), HF radio focusing caused by the electron distribution between ionospheric layers, J. Geophys. Res.
72, No.9, 2343-2355.
de Voogt, A. H. (1953), The calculation of the path of a radio-ray in a given ionosphere, Proc. IRE 41, No.9, 1183-1186.
Forsterling, V. K., and H. Lassen (1931), Die Ionisation der Atmosphiire und die Ausbreitung der kurzen elektrischen
Wellen (IQ-100 m) iiber die Erde. III, Zeitschrift fur technische Physik, 12, 502-527.
Lejay, P., and D. Lepechinsky (1950), Field intensity at the receiver as a function of distance, Nature 165, No. 4191, 306-
307.
Rawer, K. (1948), Optique Geometrique De L'Ionosphere, Rev. Sci. Paris, No. 3298, 585--600 (4 Rue Pomereu, Paris).

(Paper 3-l-330)