Wind energy is the world´s fastest-growing renewable energy source. Thus, the amount of people exposed to wind farm noise is increasing. Due to its broadband amplitude modulated characteristic, wind turbine noise (WTN) is more annoying than noise produced by other common community/industrial sources. Aerodynamic noise along the blade span is the dominant noise source of modern large wind turbines. This type of noise propagates through the atmosphere in the proximity of wind farms. However, modelling and simulating WTN propagation over large distances is challenging due to the complexity of atmospheric conditions. Real temperature, wind velocity and relative humidity measurements typically show a characteristic nonlinear behavior. A comprehensive propagation model that addresses this problem while maintaining high accuracy and computational efficiency is necessary. A Hamiltonian Ray tracing (HRT) technique coupled to aerodynamically induced WTN is presented in this work. It incorporates acoustic wave refraction due to spatial speed of sound gradients, a full Doppler Effect formulation resulting from wind velocities in any arbitrary direction, proper acoustic energy dissipation during propagation, and ground reflection. The HRT method averts many of the setbacks presented by other common numerical approaches such as fast field program (FFP), parabolic equation methods (PE), and the standard Eikonal ray tracing (ERT) technique. In addition, it is not bounded to the linearity assumptions made for analytic propagation solutions. A wave phase tracking analysis through inhomogeneous and moving media is performed. Curved ray-paths are numerically computed by solving a non-linear system of coupled first order differential equations. Sound pressure levels through the propagation media are then calculated by using standard ray tubes and performing energy analysis along them. The ray model is validated by comparing a monopole’s ray path results against analytically obtained ones. Sound pressure level predictions are also validated against both FFP and ERT methods. Finally, results for a 5MW modern wind turbine over a flat acoustically soft terrain are provided.