The single circular array (SCA) method is a spatial autocorrelation (SPAC)–like technique for ambient noise exploration. Its main feature is the possibility of calculation of Love-wave dispersion curves by using centerless circular arrays of 3-component seismometers, allowing independent processing of each circle. Situations in which Rayleigh-wave and Love-wave arrivals or waves coming from different azimuths are mutually correlated are also correctly dealt with in this method. An algorithm for practical calculation of the SCA coefficient B is described. The algorithm includes averaging over a set of time windows and minimizes the number of spectral ratios to be computed for the purposes of stability. Numerical tests show that SCA coefficients estimated in this way have quite a robust behavior. Bias due to use of a finite number of sensors, as well as to effects of nonpropagating incoherent noise, has been theoretically studied in both the deterministic and the stationary random-field formulations. Using a finite number of stations is a cause of bias even under isotropic illumination conditions. Nevertheless, its effect can be neglected for wavelengths-to-radius ratios above a threshold that depends on the number of evenly distributed sensors. By contrast, uncorrelated noise may affect the whole frequency band and is behind the limitations of the method at low frequencies.
Finally, we present the first real data test of this method, consisting of a comparison between theoretical and experimental Love-wave dispersion curves for a site where the structure is known. In practice, the minimum wavelength for direct velocity retrieval for a pentagonal array with radius r was approximately , although this value depends on the signal-to-noise ratio. Experiments demonstrate that the usable range can be extended, mainly toward shorter wavelengths, if the effects of noise and of the finite number of sensors are included in the analysis.