Abstract: Capon’s estimator was originally developed for frequency-wavenumber power spectral density analysis. Theoretically, it can be treated as maximum likelihood estimator of unknown vector parameter of multidimensional linear system (MLS) model with one input and m outputs.  Practically important and non-trivial case is considered, when the number of nuisance parameters, which are values of input deterministic signal, tends to infinity. Under assumptions that MLS output signals distorted by random Gaussian stationary  EMBED Equation.DSMT4 -dimensional noise with known matrix power spectrum and the likelihood function of observations satisfies the conditions of local asymptotic normality, the analytical form of limiting covariance matrix of the Capon’s estimate is given.Steered response power phase alignment algorithm (SRP-PHAT) is devoted to subclass of so called phase-based robust estimators. Mathematical basis of SRP-PHAT estimator shows its applicability to small-aperture seismic array data to solve two different tasks: frequency-wavenumber analyses and seismic source localization problems. Several experiments with real data demonstrate its advantage compared to conventional methods that are involved in most of free and commercial seismological software.