Long range sound propagation in the atmosphere is influenced by several effects: atmospheric turbulence, sound-speed gradients, ground properties (impedance, rugosity) etc. In the context of supersonic aircraft, nonlinear propagation of the sonic boom has to be taken into account. To evaluate the influence of these different effects, a statistical analysis is needed. However, field measurements suffer from a lack of control on atmosphere characteristics, and the statistical analysis remains circumscribed by the limited number of aircraft flight tests. An alternative to outdoor measurements is to perform experiments under well-controlled laboratory conditions. These experiments allow to study the effects purely related to the turbulent layer and to the sound-speed gradient. The propagation of high amplitude and short duration N-waves through thermal turbulence is studied. In particular, the influence of a rigid boundary and a negative sound-speed gradient resulting in a shadow-zone near the boundary, are pointed out. An experimental setup has been designed : N-waves are generated using a spark source. Thermal turbulence is obtained with a grid of electrical resistors, and the shadow-zone is obtained used a curved boundary. Measurements are performed using 1/8" microphones, and a schlieren shadowgraphy technique. The real pressure waveform (different from the ideal N-wave) delivered by the spark source has been obtained using the strioscopy technique, up to a multiplicative constant. Weak-shock theory enables to estimate the pressure level under the assumption that the pressure waveform is an ideal N-wave. This method has been modified to take into account the real pressure waveform. Because of the geometrical scaling, laboratory-scale experiments lead to a shift to a higher frequency domain : typically, 10 kHz - 1 MHz. In this frequency range, no calibration method is available up to now. A new method has been proposed and successfully applied. Free-field propagation of high-amplitude N-waves through the turbulent layer leads to a mean attenuation of the pressure level. However, random focus of the pressure waves can be observed, up to factor of 3. Probability densities of the shock overpressure have been described with an excellent agreement by a generalized Gamma probability distribution. With a rigid boundary, irregular reflections of shockwaves can be observed because of the high pressure level. This type of irregular reflections has been outlined either with a plane or a cylindrical boundary. Propagation in an acoustical "shadow zone" lead to an amplification of the pressure, contrary to the free-field observations. An important result of the experiment is that the probability to observe an attenuation of the pressure level is null, for every single wave propagating in the "shadow zone". This result suggests that, with turbulence, the dominant propagation mechanism into the shadow-zone is scattering by sound-speed in homogeneities. In the context of sonic boom exposure on the ground, this would lead to an extension of the primary carpet of the aircraft.
