Accuracy requirements for unmanned aerial vehicle-based acoustic atmospheric tomography

If the acoustic signature is measured onboard a propeller-driven unmanned aerial vehicle and by an array of microphones on the ground, comparison between the projected and observed signals allows the propagation delays—and hence effective sound speed values of the intervening medium—to be determined. As the ray paths along which energy travels intersect within the atmosphere at multiple locations and angles, temperature and wind velocity fields may be estimated using tomography. This paper describes signal processing strategies for determining propagation delays at sub-millisecond levels of precision; the inverse problem and a model that allows meteorological observations to be incorporated into the tomographic solution; and the algebraic iterative reconstruction technique used to overcome the ill-posed nature of the inversion. A weakly sheared daytime convective atmospheric boundary layer—synthesised through use of large eddy simulation code that utilises pseudo-spectral differencing and solves an elliptic pressure equation—is used to show sub-millisecond levels of observation noise permits faithful reconstruction of target atmospheres. Particular attention is paid to the atmospheric layer below 100 m as this region typically experiences the greatest spatio-temporal variation in temperature and wind speed.