DAMAS2 Using a Point-Spread Function Weakly Varying in Space

T HE resolution of noise-source maps processed using phasedarray data has been substantially improved by introducing a deconvolution approach for the mapping of acoustic sources (DAMAS [1]), but its practicality is limited due to high computational cost. Subsequently, an improved algorithm, known as DAMAS2 [2], was developed to reduce the computational cost dramatically. Assuming that the monopole response (i.e., the pointspread function) on the source map is invariant over the entire map, referred to as shift invariant, a discrete Fourier transform and its inverse are introduced for deconvolution in DAMAS2. The drawback is that noise-source maps tend to deteriorate away from the point where we defined the point-spread function. Distortion of the monopole response is often severe in wind-tunnel tests, where there are constraints on the location of the microphone array. This note proposes amethod that suppresses such deterioration associatedwith the shift-invariant assumption for DAMAS2 with minimal extra computational cost.