A new method devised for solving the three-dimensional (3D) atmospheric refraction
problem is presented. It considers the situation when the origin and the end of the
refracted beam are given (boundary value problem) and is based on Fermat's variational
principle. Original points of the method are the 3D geometry with ellipsoidal
Earth, no sampling of exo-atmospheric beam and treatment of mirage effects.
The method is first described and great care is given to the particular case of grazing
incidence yielding mirages. In contrast, the case of exo-atmospheric paths (satellite
links) is optimized by using a variable integration step, large outside and small inside the
atmosphere. Validation is then performed by comparison with numerical and analytical
models in the case of astronomical refraction and with MODTRAN 3.5 in the case of
terrestrial refraction. The influence of the Earth's oblateness is shown to modify the
refraction angle by 1.3% in absolute magnitude. Consistency is also checked with the direct
method where the origin, direction of sight and distance are specified (initial value
problem).
In a companion paper (Berton, Part II) the influence of significant parameters such as
observer and target distance and altitude, vertical profiles of temperature, pressure and
humidity, wavelength and sampling step will be analysed in detail. The accuracy of the
refractive index formula will also be tested.