Shear-wave structure of southern Sweden from precise phase-velocity measurements of ambient-noise data

Summary Rayleigh-wave phase-velocity tomography of southern Sweden is presented using ambient seismic noise at 36 stations (630 station pairs) of the Swedish National Seismic Network (SNSN). We analyze one year (2012) of continuous recordings to get the first crustal image based on the ambient-noise method in the area. Time-domain cross-correlations of the vertical component between the stations are computed. Phase-velocity dispersion curves are measured in the frequency domain by matching zero crossings of the real spectra of cross-correlations to the zero crossings of the zeroth-order Bessel function of the first kind. We analyze the effect of uneven source distributions on the phase-velocity dispersion curves and correct for the estimated velocity bias before tomography. To estimate the azimuthal source distribution to determine the bias, we perform inversions of amplitudes of cross-correlation envelopes in a number of period ranges. Then, we invert the measured and bias-corrected dispersion curves for phase-velocity maps at periods between 3 and 30 s. In addition, we investigate the effects of phase-velocity bias corrections on the inverted tomographic maps. The difference between bias corrected and uncorrected phase-velocity maps is small ($< 1.2 \%$), but the correction significantly reduces the residual data variance at long periods where the bias is biggest. To obtain a shear velocity model, we invert for a one-dimensional velocity profile at each geographical node. The results show some correlation with surface geology, regional seismicity and gravity anomalies in the upper crust. Below the upper crust, the results agree well with results from other seismological methods.