The computation of radiowave attenuation in the atmosphere is a series of radio propagation models and methods to estimate the path loss due to attenuation of the signal passing through the atmosphere by the absorption of its different components. There are many well-known facts on the phenomenon and qualitative treatments in textbooks. A document published by the International Telecommunication Union (ITU)
provides some basis for a quantitative assessment of the attenuation. That document describes a simplified model along with semi-empirical formulas based on data fitting. It also recommended an algorithm to compute the attenuation of radiowave propagation in the atmosphere. NASA also published a study on a related subject. Free software from CNES based on ITU-R recommendations is available for download and is available to the public.
The model and the ITU recommendation
The document ITU-R pp. 676–78 of the ITU-R section considers the atmosphere as being divided into spherical homogeneous layers; each layer has a constant refraction index. By the use of trigonometry, a couple of formulas and an algorithm were derived.
Through the use of an invariant, the same results can be directly derived:
An incident ray at A under the angle Φ hits the layer B at the angle θ. From basic Euclidean geometry:
By Snell's law:
so that
Notes:
One proof starts from the Fermat's principle. As a result, one gets proof of the Snell's law along with this invariance. This invariant is valid in a more general situation; the spherical radius is then replaced by the radius of curvature at points along the ray. It is also used in equation (4) of the 2005 NASA's report in an application of satellite tracking. The assumption of the refraction index varying with the latitude is not strictly compatible with the notion of layers. However the variation of the index is very small, this point is usually ignored in practice.
The ITU recommended algorithm consists of launching a ray from a radio source, then at each step, a layer is chosen and a new incidence angle is then computed. The process is iterated until the altitude of the target is reached. At each step, the covered distance dL is multiplied by a specific attenuation coefficient g expressed in dB/km. All the increments g dL are added to provide the total attenuation.
Note that the algorithm does not guaranty that the target is actually reached. For this, a much harder boundary value problem would have to be solved.
The eikonal equation
This equation is discussed in the references. The equation is highly non-linear. Given that a smooth data fitting curve n(altitude) is provided by the ITU for the refraction index n, and that the values of n differs from 1 only by something of the order 10−4, a numerical solution of the eikonal equation can be considered. Usually the equation is presented under the self-adjoint form, a more tractable equation for the ray head position vector r is given in generic parametric form:
Implementations
Three implementations to compute the attenuations exist:
Take the ray to be a straight line. Use the optical invariant and apply the ITU recommendation. Solve the eikonal equation.
The first two are only of 1st order approximation (see Orders of approximation). For the eikonal equation, many numerical schemes are available. Here only a simple second order scheme was chosen. For most standard configurations of source-target, the three methods differ little from each other. It is only in the case of rays grazing the ground that the differences are meaningful. The following was used for testing:
At the latitude of 10°, when a ray starts at 5 km altitude with an elevation angle of −1° to hit a target at the same longitude but at latitude 8.84° and altitude 30 km. At 22.5 GHz, the results are:
Note that 22.5 GHz is not a practical frequency but it is the most suitable for algorithms comparison. In the table, the first column gives the results in dB, the third gives the distance covered and the last gives the final altitude. Distances are in km. From the altitude 30 km up, the attenuation is negligible. The paths of the three are plotted:
Note: A MATLAB version for the uplink (Telecommunications link) is available from the ITU