A 70-microphone phased array, protected to withstand the harsh environment of a rocket test facility, was used to identify noise sources in the Ares I Scale Model Acoustics Test. In the unobstructed burn of a single solid rocket motor, the free-flowing plume itself was found to make a long noise source. The scenario changed completely in launch configurations where a 5%-scale model of the Ares I vehicle was tested in static firings. It was found that the impingement by the plume on various regions of the launch pad constituted the primary noise sources. The scenario is very different from current models, which assume that the plume itself is the noise source and do not account for impingement sources. As expected, the addition of water in the trench and the hole for the plume passage attenuated the associated noise sources. Water injection on the top of the pad ("rainbird") was found to attenuate only the peripheral sources around the primary plume impingement zone. The noise maps suggest that the minimization of impingement by reducing vehicle drift, reducing plume spillage via increasing the size of the hole, and covering-up leakage paths for the sound waves from the trench will attenuate the liftoff acoustics level. Nomenclature b = beamformed output d = aperture of the array G = cross-spectral matrix R = radial position of a microphone from array center S = weight applied to individual microphones St = Strouhal frequency U = plume velocity at nozzle exit w = steering vector θ = angular position from the image center λ = wavelength Subscripts D = nozzle exit diameter f = frequency i, j = index for interrogation grid m = microphone index R = Rayleigh resolution