The present invention relates to a machine implemented method for spectral analysis that determines a measure of cross coherence between application of two spectral estimation filters to data; and identifies a spectral feature of the measure of cross coherence. One example embodiment of the present invention provides a complete statistical summary of the joint dependence of the Bartlett and Capon power spectral statistics, showing that the coupling is expressible via a 2×2 complex Wishart matrix, where the degree coupling is determined by a single measure of cross coherence defined herein. This measure of coherence leads to a new two-dimensional algorithm capable of yielding significantly better resolution than the Capon algorithm, often commensurate with but at times exceeding finite sample based MUSIC.