8:30 AM - 8:45 AM
[S03-3-01] Bias in velocity measurements from ambient noise due to anisotropic source distributions
Secondary microseismic sources are strong off the Atlantic coast of Norway and dominate ambient noise recorded in Scandinavia. At the same time, the source distribution is quite heterogeneous and
sources from the south and east (Baltic Sea) are weak or absent. This source heterogeneity causes significant bias in measurements of velocity of surface waves extracted from ambient noise by correlation. We study this bias effect synthetically and characterize it in terms of Weaver's asymptotic formula at large inter-station distance, r, or short wavelength, L, and in terms of Fresnel angles at short inter-station distance or large wavelength. We find that the bias of group-velocity measurements is significantly higher than that of phase velocity. We find that Weaver's formula applies to within several percent when the ratio r/L > 10 and when the local anomaly angular width exceeds the width of the Fresnel angle. When the source distribution varies on a small angular scale and inter-station distances are relatively short the phase-velocity bias can be as large as several percent. We quantify the bias effects of the mapped source distribution around Scandinavia and find that in this region of relatively weak crustal heterogeneity (3%) the bias often exceeds the heterogeneity level, particularly for group velocity. This highlights the need for careful study of the effective source distribution of microseisms and its potential bias effects in regions of weak heterogeneity before inverting dispersion measurements for 3D velocity structure. Phase velocity is to be preferred over group velocity and careful data selection based on the ratio r/L and the azimuthal distribution of the source is vital.
sources from the south and east (Baltic Sea) are weak or absent. This source heterogeneity causes significant bias in measurements of velocity of surface waves extracted from ambient noise by correlation. We study this bias effect synthetically and characterize it in terms of Weaver's asymptotic formula at large inter-station distance, r, or short wavelength, L, and in terms of Fresnel angles at short inter-station distance or large wavelength. We find that the bias of group-velocity measurements is significantly higher than that of phase velocity. We find that Weaver's formula applies to within several percent when the ratio r/L > 10 and when the local anomaly angular width exceeds the width of the Fresnel angle. When the source distribution varies on a small angular scale and inter-station distances are relatively short the phase-velocity bias can be as large as several percent. We quantify the bias effects of the mapped source distribution around Scandinavia and find that in this region of relatively weak crustal heterogeneity (3%) the bias often exceeds the heterogeneity level, particularly for group velocity. This highlights the need for careful study of the effective source distribution of microseisms and its potential bias effects in regions of weak heterogeneity before inverting dispersion measurements for 3D velocity structure. Phase velocity is to be preferred over group velocity and careful data selection based on the ratio r/L and the azimuthal distribution of the source is vital.