The propagation of sound waves depends on the structure of the atmosphere. The effective sound velocity is a function of wind and temperature (T) of the atmosphere, following:

Rays are traced on the basis of Snell's law for a moving atmosphere:

Thus, detailed information on each atmospheric layer is necessary in this analysis. Since the sound velocity structure of atmosphere serves as input to the raytracing algorithm, wind and temperature structure of the atmosphere should be approximated as good as possible. The lower atmosphere, up to 15 km heights, is well known, because the weather takes place in this so-called troposphere. Models of the troposphere can be validated with actual measurements (p.e. balloon data). The wind and temperature structure of the stratosphere (15 up to 55 km), mesosphere (55 up to 85 km) and thermosphere (85 up to 200 km) are not known in much detail. They are assumed not to vary too much, mostly on a seasonal scale. Rocket and satellite data give some insight into the wind and temperature layering at these heights. As input model for raytracing, one often depends on empirical models.

Raytracing through atmospheric models can be used to explain observed data or to predict possible ray trajectories for an event. Garcés et al., 1998 rewrote the tau-p method as known from seismology (Buland and Chapman, 1983) to make it applicable to infrasonic wave propagation. To, amongst others, interpret the results of raytracing, the delay time function is used and defined as:

The delay time function represents, the amount of time a rays spends traveling vertically through a layer. If the delay time function is close to zero, the ray will travel more horizontally than for larger values of the delay time function. Negative, or imaginary, values of the delay time function are physically impossible and are defined by an area where the ray can not travel and should therefore have bended before reaching these layers.